Luận văn thạc sĩ Quản lý xây dựng: Public opinion analysis for management of urban infrastructure systems: Social media data mining approach

INTRODUCTION

B ACKGROUND OF THE S TUDY

Adequate, reliable, and efficient urban infrastructure systems (UIS) are fundamental to sustainable development, social mobility, and economic vitality (Sánchez-Silva et al., 2016; P Wang & Shi, 2018) All members of the society, from individuals to global conglomerates, rely on basic infrastructure services to support their daily communal functions (Harris et al., 2017) For this reason, major components of UIS such as roads, light rails, and water supply networks are subject to heavy use throughout their operable life As a result, the system deteriorates rapidly over time (Flintsch & Chen, 2004; Frangopol & Kim, 2011; Sánchez-Silva et al., 2016; Sánchez-Silva & Klutke, 2016) Under these anticipated situations, it is crucial that efficient infrastructure management strategies are in place to ensure that UIS components are maintained in excellent condition

Management of urban infrastructure systems involves multi-faceted and complex decision-making processes throughout the life-cycle of the project (Guillaumot et al., 2003) Implementation of maintenance and monitoring activities is one of the core decisions involved in this aspect (L Chen et al., 2015; Mild & Salo, 2009) Normally, at the project programming level, monitoring, maintenance, and repairs are established as predefined activities with earmarked budgets that are spread across the planned operations and maintenance horizon (Giustozzi et al., 2012; Kuhn, 2010; Mild & Salo, 2009) These undertakings include routine infrastructure health monitoring activities that are usually carried out by technical personnel of responsible public agencies who are tasked to perform measurements and draw reports based on field inspections (Harris et al., 2017; Mishalani & Gong, 2009) Accordingly, maintenance treatments that can either be routine, rehabilitation, or preservation are implemented depending on the system adopted by the public authorities (Torres- Machi et al., 2018)

There are several identifiable flaws in the existing practices of monitoring and maintaining urban infrastructure systems First, maintenance and monitoring programs as preplanned activities that are funded through reserved budgets discount the evolving circumstances of infrastructure use, especially when it comes to demand

A public road that was constructed ten years ago may be seeing heavier use at present, especially in the course of rapid urban development Variations in demand, coupled with other external and internal conditions, can dramatically accelerate the deterioration of urban infrastructure (Sánchez-Silva et al., 2016) If public agencies rely on routine monitoring alone, these situations may go unnoticed In such cases, untimely degradation of structures may be imminent As premature deterioration of public infrastructure necessitates urgent corrective measures, this may result to expenditures in excess of the budget allocated for routine maintenance programs This inadequacy brings attention to a need for a more proactive and timely approach in infrastructure monitoring

Secondly, the current approach in infrastructure monitoring is confronted by two limitations: subjectivity and cost-effectiveness (Chang et al., 2003; Harris et al., 2017) As infrastructure health monitoring activities are centralized within the functions of public authorities, these activities rely heavily on subjective assessments of technical personnel of the responsible public authorities (Harris et al., 2017) As this system fails to provide opportunities for public engagement, especially within the immediate locality of the UIS component, this may lead to inaccurate judgment on the actual condition of the infrastructure In an urban infrastructure system, it is important to consider that end users, as direct stakeholders, have incontestable merit to assess the reliability and performance of public infrastructure based on how it serves their best interest (P Wang & Shi, 2018; Y Zhang et al., 2019)

Finally, in terms of cost, deploying large-scale infrastructure monitoring activities exhausts a huge amount of resources (Harris et al., 2017) For vast networks of urban infrastructure systems with interdependent components requiring equally rigorous attention, the current approach is essentially impracticable

This research focuses solely on the local context of Urban Infrastructure Management in the Philippines With that, it is imperative that its setting as the

3 research domain is analyzed in order to gauge the study’s essence, practicability, and relevance

According to a report by KPMG (2015), a global audit and financial services firm, there has been a consistent evidence of underinvestment and underperformance of roads, mass transit, and water infrastructures in the country Coupled with the findings of the survey from World Economic Forum Global Competitiveness Report from 2013 to 2014, the Philippines was found to have very poor or inadequate overall quality of infrastructure, lagging behind its neighbors Singapore, Malaysia, Thailand, and Indonesia and only slightly above Vietnam Alarmingly, the same relative ranking reflecting the performance of infrastructure in the Philippines was still apparent based on the findings from 2017 to 2018 Evidently, these deficiencies in infrastructure translate to further consequences in the economic performance of the country Ito (2019) claimed that this inadequacy is mainly driven by low-effort investment in development and management of urban infrastructure, alongside complex political structures that may not be ideally complementary with the goals of robustly progressing the quality of infrastructure in the country

The deficiencies in the quality of urban infrastructure in the Philippines, along with the persistent issue of conservative budgeting in development and management of UIS brings attention to a solution that can be effectively utilized considering the given constraints In this light, the Philippines, as a research setting, was identified to be fitting.

S TATEMENT OF THE P ROBLEM

In light of the aforementioned constraints on UIS management, this study finds motivation in revitalizing and supporting current urban infrastructure monitoring strategies by presenting a novel approach that is proactive, cost-effective, and provides an improved avenue for public engagement and feedback Primarily, this study will exploit the possibility of extracting meaningful information from publicly available user-generated data in social media (SM) using various data mining and text mining techniques.

R ESEARCH O BJECTIVES

This study aims to analyze public opinion in the form of user-generated social media data as a means to extract meaningful information and derive actionable knowledge in relation to the performance of urban infrastructure systems Specifically, this study seeks to:

1 Investigate common and recurring issues in public infrastructure based on public opinion

2 Investigate the occurrences and emerging trends on public opinion regarding urban infrastructure and its attached services

3 Identify the relevant demands of the public with respect to public infrastructure.

S COPE AND L IMITATIONS OF THE S TUDY

In order to realize the above stated objectives, this research shall be confined within the following bounds:

1 This study considered the urban infrastructure system of The Philippines Specifically, this study was limited to existing public infrastructure in The Philippines such as roads, bridges, light rails, and water supply networks along with the attached public agencies responsible for each infrastructure component

2 For ease of data collection and consistency of approach, the social media platform considered for the study was limited to Twitter.

S IGNIFICANCE OF THE S TUDY

Efficient urban infrastructure monitoring and maintenance strategies are key to ensuring that the system will perform well and remain serviceable throughout its life span Current strategies, however, remain confronted with various limitations including adaptability to changing conditions, lack of public engagement, and cost- effectiveness In this view, this study seeks to explore a proactive, user-involved, and cost-effective approach to revitalize conventional infrastructure monitoring processes Capitalizing on the availability and abundance of user-generated data in social media platforms as a potential source of meaningful and actionable information, the findings of this study can help support planning and response strategies in routine maintenance, preservation, and improvement of urban infrastructure systems Moreover, understanding public opinion concerning the reliability, performance, and demands for improvement of public infrastructure can play a major role in decision making and infrastructure planning

Ultimately, this study aims to contribute foundational research on the use of social media data for infrastructure research, especially in the context of The Philippines

This Chapter sets and summarizes the problem at hand, the motivations behind the pursuit of research, and the objectives committed All frameworks, methods, and approaches employed toward the completion of this research are discussed in detail in the subsequent Chapters Chapter 2 hovers on previous related literature that were used as foundations for this undertaking; Chapter 3 discusses the research design implemented; Chapter 4 discussess, in-depth, the results achieved and their implications in UIS management Ultimately, this study closes with the final Chapter with conclusions and recommendations – both for stakeholders and future similar endeavors.

LITERATURE REVIEW

U RBAN I NFRASTRUCTURE M ANAGEMENT

Urban infrastructure is an umbrella term for services and assets that provide the fundamental physical and organizational systems that support economic and societal functions (World Bank, 2020) More specifically, “economic infrastructure” is coined to collectively represent public utilities, public works, and other transport sectors (World Bank, 1994) The United Nations Human Settlement Programme (UN- Habitat) defines urban infrastructure, and its attached basic services to encompass essential components including delivery of clean water, sanitation, waste management, transportation, and management of public spaces (UN-Habitat, 2015)

Numerous benefits can be derived from reliable and efficient UIS An urban infrastructure system that performs efficiently and responds effectively to the demand of service operates as a catalyst for progressive economies and sustainable communities (World Bank, 1994); its optimum operation is vital to sustain socioeconomic activities and promote sustainable urban growth (Sánchez-Silva et al., 2016; P Wang & Shi, 2018; World Bank, 2020) For these reasons, government agencies and private entities responsible for infrastructure development seek effective infrastructure management strategies to ensure all infrastructure assets remain serviceable and in excellent condition (Flintsch & Chen, 2004)

Guillaumot et al (2003) and Mishalani & Gong (2009) define infrastructure management as a process in which public agencies collect and analyze data regarding urban infrastructure systems, assess and forecast the condition of infrastructure components, and make necessary decisions on maintenance, repair, and rehabilitation (MR&R) based on the assessments Fig 1 illustrates this process as an event chain that builds the general working of infrastructure management

Figure 1 Infrastructure management event chain Adapted from Guillaumot et al (2003)

Flintsch & Chen (2004) presented a general functional framework for infrastructure asset management that deconstructs the system into different levels of decision-making and highlights the importance of different aspects of infrastructure management (Fig 2) Notably, “feedback”, as a consequential output of performance monitoring of infrastructure components, is shown to deliver supplemental information needed for asset inventory, condition assessment, maintenance strategies, performance prediction, and needs analysis

Figure 2 Infrastructure asset management framework Reprinted from Flintsch & Chen (2004)

From a general perspective, infrastructure management systems are ultimately concerned about the condition and performance of the infrastructure components In this regard, monitoring and maintenance of UIS becomes central to infrastructure asset management (L Chen et al., 2015; Tafazzoli, 2017) This is further supported by the growing number of research studies focusing on this aspect

Improving urban infrastructure management has been extensively studied by many researchers in the past several years These studies crossed multiple and varying objectives within the realm of infrastructure management research Minimizing life- cycle cost (Amador & Magnuson, 2011; Ghodoosi et al., 2018; Kim & Frangopol,

2017; Rashedi & Hegazy, 2016; Ward & Savić, 2012), maximizing asset performance (Abu Samra et al., 2018; Amador & Magnuson, 2011; Elbehairy et al., 2009; Kim & Frangopol, 2017; Rashedi & Hegazy, 2016; Salman et al., 2013), assessing environmental impact of maintenance strategies (Giustozzi et al., 2012; Torres-Machi et al., 2018), and overall improvement of asset management practices (Abu-Samra et al., 2020; Guillaumot et al., 2003; Kuhn, 2010; Mishalani & Gong, 2009; Tafazzoli, 2017) were among the dominant research areas that populated this field of study Although very little attention has been given to innovate data collection methods in assessing the condition of urban infrastructure systems, a growing amount of contemporary studies in this area shows evidence that it has likewise gained the interest of many researchers

The seminal work of Walker et al (2014) has laid foundation for the use of mobile computing in collecting and organizing data for infrastructure assessment In the study, a smartphone-based software named “Mobile Information Collection Application (MICA)” was developed and applied for route exploration and preliminary surveys for flood prevention infrastructure (Fig 3) Due to its advanced features, the software was evaluated to have performed exceedingly well compared to traditional route survey methods in consideration of savings in time and resources

Figure 3 Route reconnaissance trial results using MICA Reprinted from Walker et al (2014)

In the prominence of more sophisticated computing methodologies, the research on infrastructure monitoring took a leap forward The works of Koch et al (2013), Hadjidemetriou et al (2016), Radopoulou & Brilakis (2017) and Kyriakou et al (2019) harnessed the power of artificial intelligence in detecting pavement anomalies, contributing solutions to persistent challenges in pavement management systems Particularly, Kyriakou et al (2019) coupled smartphone sensors with artificial neural networks to examine the interactions between vehicles and roads as a way to detect the presence of potholes The results of the study confirmed the efficacy of the proposed method in detecting road anomalies

Figure 4 Relating smartphone and vehicle angular movements Reprinted from Kyriakou et al (2019)

These studies delivered promising solutions to outdated and inefficient data collection methods for infrastructure assessment However, a gap remains to be filled for integrating the social aspects of a UIS in the assessment of its key components In a holistic management approach, it is important to recognize that urban infrastructure systems lie within the confluence of technical and social environments (Valentin et al., 2018) One important social dimension that has been determined to have substantial influence in infrastructure development is public opinion For the most part, the wealth of information that perhaps can be obtained from analyzing public opinion remains untapped This opens a whole new domain for infrastructure management research.

P UBLIC E NGAGEMENT IN U RBAN I NFRASTRUCTURE S YSTEMS

Attributing to its magnitude, complexity, and impact, urban infrastructure systems lure more public opinion than other types of construction projects (Jiang et al., 2016; Valentin et al., 2012) The extent and tonality of public opinion is inherently unpredictable, owing to the fact there are many project aspects that may influence it; thus, it has a conceivable capability to bring about uncertainties and risk to infrastructure projects (Valentin & Bogus, 2013) On the upside, public engagement provides an opportunity for understanding and balancing the interests of multiple stakeholders in UIS, particularly those of the end users (Leung et al., 2013)

In the aspect of infrastructure management, public opinion is regarded to be of significant importance (Jiang et al., 2016; Seto & Sekimoto, 2019; Valentin et al., 2018) This importance is clearly emphasized when dealing with UIS performance indicators that are otherwise qualitative and cannot be measured numerically (Khatri et al., 2012) Additionally, public satisfaction is viewed to have a strong direct correlation with the efficiency and quality of urban infrastructure services (Van Ryzin et al., 2008; P Wang & Shi, 2018) Thus, understanding public opinion opens an explorable channel for public agencies to improve infrastructure services and achieve desired levels of efficiency for UIS Furthermore, contemporary research substantiates that the general public is increasingly becoming an expedient source of valuable information needed for infrastructure development, and engaging them often leads to novel solutions that would otherwise not surface in an exclusive and centralized decision-making environment (Lakhani & Panetta, 2016; M Nik-Bakht

& El-Diraby, 2020; Mazdak Nik-Bakht & El-diraby, 2016; von Hippel, 2005) Despite its recognized importance in UIS management, integrating public opinion in urban infrastructure management remains challenging, primarily due to shortfalls of data collection methods that are employed by public agencies

Many urban infrastructure developments have acknowledged the benefit, and are pushing for public participation particularly in operations and maintenance (O&M) phases (Sohail et al., 2005) Typically, engaging the users require personal visits, surveys, or forums organized by public authorities (Y Zhang et al., 2019)

Unfortunately, these activities are burdened with exhaustive labor and sizable use of resources (Martín et al., 2020; Tang et al., 2015, 2017; Y Zhang et al., 2019) In addition, these current practices have been shown to be inadequate in properly representing the population of interest and in eliminating biases (Martín et al., 2020;

M Nik-Bakht & El-Diraby, 2020; Tang et al., 2015, 2017) To address these shortcomings, a fair number of studies took interest in exploring innovative methods to gather public feedback as a means to assess the performance of urban infrastructure

Harris et al (2017) found leverage on the increasing prevalence of smartphones and the day-to-day interaction of general public with urban infrastructure systems Hence, a framework integrating urban infrastructure with volunteer “human sensors” called “The Citizen Engineer” was proposed (Fig 5) Through the system, the citizens are given capabilities to report infrastructure defects and issues through their smartphones, which are then aggregated to aid public authorities in thorough assessment and response

Figure 5 Illustrative framework of "The Citizen Engineer" Reprinted from Harris et al (2017)

Seto & Sekimoto (2019) utilized the information contained in “Chiba-repo” (Fig 6), a repository of citizen-generated reports for urban infrastructure issues and concerns in Chiba City, Japan The study entailed analyzing the geographic distribution and textual information contained in the reports As a result, the most frequent and major urban infrastructure issues in the locality were revealed

Figure 6 User interface of Chiba-repo Reprinted from Seto & Sekimoto (2019)

The works of Harris et al (2017) and Seto & Sekimoto (2019) demonstrated the viability of harboring public feedback in assessing the performance and condition of urban infrastructure However, the approaches presented are manifested to be dependent mainly on the availability of complex information management platforms and the public’s willingness to participate Harris et al (2017) enlisted the help of volunteers to monitor the condition of public infrastructure On a greater scale of implementation, this approach is rendered unsustainable Seto & Sekimoto (2019), on the other hand, utilized the data readily available from an existing centralized

13 repository of information Despite its efficacy, a complex information management system, such as Chiba-repo, requires high financial capital and costly maintenance, which makes it challenging to employ across all settings, especially in developing countries For these reasons, the need for a more proactive, engaged, and cost- efficient solution remains

Contemporary research points to social media platforms as an emerging social environment that has been consistently gaining attention as a gold mine of valuable data having the potential to surpass the limits of conventional means of collecting public feedback The advent of social media and its growing prevalence in the modern society offers multitude of possibilities that are yet to be fully explored in infrastructure management research.

S OCIAL M EDIA

Despite its growing familiarity, a widely accepted unifying definition for social media has yet to be established Researchers engaged in the subject arrive at rather dispersed definitions mainly due to differences in context This present study, however, subscribes to the definition provided by Kapoor et al., (2018):

Social media is made up of various user-driven platforms that facilitate diffusion of compelling content, dialogue creation, and communication to a broader audience It is essentially a digital space created by the people and for the people, and provides an environment that is conducive for interactions and networking to occur at different levels (for instance, personal, professional, business, marketing, political, and societal)

Utility of SM transcends multiple disciplines and subjects (Martín et al., 2020) Particularly, in the architecture, engineering, and construction (AEC) industry, its prominence continues to evolve (Azhar et al., 2019) Several researchers have taken a detour in exploring social media data to investigate various phenomena in AEC and

15 related fields These studies are presented in Table 1 Studies in AEC and related fields that utilized social media data

Table 1 Studies in AEC and related fields that utilized social media data

Paper Platform Phenomenon of Interest

(2016) Twitter Interdependence of human mobility and civil infrastructure

Jiang et al (2016) Online forums

Public opinion on large hydro projects

Tang et al (2017) Twitter Viability of social media data as a source of valuable knowledge for the construction industry

Tracy et al (2018) Twitter Information transfer and changes in online interaction networks after a hazard event

D Zhang et al (2018) Sina Weibo Public opinion on water conservation projects

Tang et al (2018) Sina Weibo Topics and discussion networks pertaining to construction

Y Zhang et al (2019) Sina Weibo Public opinion on transportation systems

Diraby (2020) Twitter Topics of interest and evolution of discussions pertaining to urban infrastructure projects

Martín et al (2020) Twitter Validity of social media data for analyzing evacuation mechanisms

Paper Platform Phenomenon of Interest

Diao et al (2020) Twitter Detection of litigation risk in construction projects

Yuan & Liu (2020) Twitter Post-disaster impact assessment

Y Chen et al (2020) Twitter Post-disaster impact assessment

W J Wang et al (2020) Twitter Communication and distribution of risk warning messages prior to a disaster event

One promising aspect of social media is that it allows for massive generation of free-form and dynamic data which can be utilized for a wide range of purposes (Barbier & Liu, 2011; Han et al., 2014) This abundance is perceived as an opportunity to uncover insights that have never been explored before, overcome the limits imposed on traditional data sources, and pave the way for finding solutions in a swift and cost-effective manner (Klašnja et al., 2015; Tang et al., 2017)

Because of its user density and rate of engagement, SM is considered to be a convenient solution for reaching larger sample size, including minorities and underrepresented demographics that would not have been reached through traditional sampling methods (Klašnja et al., 2015; Martín et al., 2020; M Nik-Bakht & El-Diraby, 2020) Studies have also claimed that the general population finds participating in social media appealing, which translates to increased level of engagement (Evans-Cowley & Griffin, 2012; M Nik-Bakht & El-Diraby, 2020) Moreover, using SM as a channel for public engagement removes the hindrances known to plague conventional engagement methods For instance, the availability social media data affords the opportunity to examine public sentiment and opinion without the coercing effects of artificial environments such as face-to-face meetings

17 and interviews (Klašnja et al., 2015; M Nik-Bakht & El-Diraby, 2020) SM also allows for natural emergence of trends in public opinion outside the constraints of predefined questions normally employed in survey methods (Klašnja et al., 2015)

Social media, irrefutably, presents a new breadth of possibilities unlike any other means The potential imbued in SM has been quick to clutch the interest of researchers in vast settings Despite its following however, the potential of social media as vault of valuable knowledge that can be utilized for management of urban infrastructures systems remains uncharted This present study, thus, aims to bridge this gap by uncovering insights embedded in user-generated data in social media as a response to the persistent challenges in management of urban infrastructure systems.

RESEARCH METHODOLOGY

D ATA C OLLECTION

The social media platform that was be utilized for this research is Twitter, primarily because of its unique features as a combined social and information network (W J Wang et al., 2020), availability of data, and ease of data acquisition (Klašnja et al., 2015; Martín et al., 2020)

This study entailed collecting Twitter messages or “Tweets” pertaining to urban infrastructure components in The Philippines The information on responsible authorities of interest, along with their Twitter accounts, are consolidated and presented in

Table 2 Responsible authorities and their respective Twitter accounts, supplemental keywords, and hashtags that will be used for data collection Tweets mentioning the usernames of responsible entities for each UIS component were collected by means of “web crawling”, or the process of collecting data from the web (Klašnja et al., 2015); a python script was written to automate collection of data from Twitter’s webpage After removal of duplicates and unrelated messages, a total of 70,131 Tweets were prepared for analysis Table 3 presents a data frame of sample Tweets that were collected.

D ATA P REPROCESSING

In its raw form, data collected from social media is characterized to be “noisy”, or containing unnecessary elements that are best removed prior to performing various

19 analyses (Barbier & Liu, 2011; D Zhang et al., 2018) This sequence of activities includes removal of embedded links, special characters, and addresses among many other things Other preprocessing techniques include lowercasing, tokenization, word stemming, and stop words removal (Jiang et al., 2016; Tang et al., 2017; D Zhang et al., 2018) The stopwords library utilized in this research, as long as the pre- processing functions used can be found on Appendix A.

D ATA A NALYSIS

To analyze unstructured text data, this study employed Text Mining or the process of extracting information from qualitative data (Marzouk & Enaba, 2019; F Zhang et al., 2019) Text mining utilizes natural language processing – a sub-discipline of artificial intelligence and computational linguistics that allows computers to comprehend human languages (Song et al., 2018; F Zhang et al., 2019) In line with the objective of deriving meaningful information and actionable knowledge from otherwise unstructured data, Topic Modeling and Sentiment Analysis were the principal text mining techniques performed

Topic Modeling – Latent Dirichlet Allocation

Latent Dirichlet Allocation (LDA) is a three-level hierarchical Bayesian model that has gained vast attention in research Topic models assume that a set of keywords represents a certain topic, and a set of topics, in turn, represent the entire document (Blei et al., 2003; Tang et al., 2017; Y Zhang et al., 2019) Topic models allow extraction of dominant topics from a huge set of text documents In the context of this research, a topic model was built to identify the common and recurring issues in UIS Furthermore, a preliminary event mapping and detection approach was employed to recognize trends in public opinion

Sentiment analysis is a robust tool that aids in understanding the sentiment or emotion of textual data (Tang et al., 2017) In this study, the deep learning-based

Natural Language Toolkit (NLTK) (Bird et al., 2009) was used to perform sentiment analysis that will classify Tweets as either, positive, neutral, or negative In this study, it is assumed that the underlying emotion of messages sent over by the public are representative of their individual perception on different UIS components and their attached services

In addition to qualitative analysis of social media data pertaining to different UIS components, structured interviews were conducted among professionals and experts with involvement in infrastructure development, planning, and management Following the recommendations of Harris et al (2017), the interviews were integrated in the methodology to draw stakeholder insights on benefits, limitations, and constraints of the explored approach This study made reference to prescribed minimum number of subjects of 5 to 50 participants as recommended for comprehensive qualitative studies (Dworkin, 2012) As a more context-specific reference, this present research was guided by the seminal research of Yu et al (2020) on utilization of Data Mining in construction industry considering a total of 11 interview participants which were also supported by the approaches of Shen et al (2016), Shi et al (2015), and Yu et al (2018) Considering the scope of this study, as well as the availability of interview participants in the domain, a total of 22 candidates were invited for participation with the aim of maximizing the representation of the participants for all UIS components considered in the study A total of 16 interviewees were confirmed for participation The structured interview was conducted using a pre-formed interview questions sheet which were sent to the respondents through online means as face-to-face interviews were impracticable due to logistical concerns

The whole research methodology was designed in a manner that is simplistic and appropriate for the scope and context of the research All results are deemed to answer the specific, measurable, attainable, realistic, and time-bound objectives of the study Verifiability and reproductibility of the results were also considered in the

21 study The succeeding Chapter discusses, in depth, the results achieved and their implications in urban infrastructure management

Data Collection Data Preprocessing Analysis Actionable Knowledge

Lowercasing Tokenization Word Stemming Stopwords removal

Common and recurring issues in UIS

Emerging trends in public opinion Relevant demands of the public

Overall public perception on UIS Topic Modeling

Latent Dirichlet Allocation Structural Topic Model

23 Table 2 Responsible authorities and their respective Twitter accounts, supplemental keywords, and hashtags that will be used for data collection

UIS Component Responsible Authority Classification Twitter Username

Department of Public Works and Highways Public @DPWHph

South Luzon Tollway Corporation Private @OfficialSLEX

North Luzon Expressway Corporation Private @NLEXexpressways

Department of Transportation Public @DOTrPH

Light Rail Manila Corporation Private @officialLRT1

Metro Rail Transit Corporation Private @dotrmrt3

Light Rail Transit Authority Public @officialLRTA

Philippine National Railways Public @PNR_GovPH

Water Supply, Sewerage, and Sanitation

Metropolitan Waterworks and Sewerage System Public @mwssro; @MWSSgovph

Maynilad Water Services, Inc Private @maynilad

Manila Water Company, Inc Private @ManilaWaterPH

Urban Space Metropolitan Manila Development Authority Public @MMDA

@joangarcia0214 2014-04-10T03:00:08.000Z @dpwhph I just want to know when the road construction at sta Cecilia st Sto Rosario village brgy Baritan Malabon city will resume i just want to know when the road construction at sta cecilia st sto rosario village brgy baritan malabon city will resume

@klsochan 2014-05-23T07:55:41.000Z @DPWHph will you be replacing the trees cut down from the Bulan Tree Tunnel after your road widening? will you be replacing the trees cut down from the bulan tree tunnel after your road widening re

@MarjChubby08 2014-05-29T03:24:01.000Z Road repair unfinished for almost a year now in Romualdez

St., Paco Manila @DPWHph road repair unfinished for almost a year now in romualdez st paco manila

@dinnygirlganda 2018-05-27T23:33:47.000Z @dotrmrt3 Hi Please tell your mrt opertators to make their announcements clear Hindi sila maintindihan eh Thanks hi please tell your mrt opertators to make their announcements clear hindi sila maintindihan eh thanks

@cnnphilippines 2018-05-10T01:45:37.000Z Good news for moms! These P2P companies are offering free bus rides to mothers until Friday, May 11 | via

@dotrmrt3 good news for moms these p2p companies are offering free bus rides to mothers until friday may 11 via

@peggyscdp 2020-10-18T16:28:32.000Z No water sa amin brgay Fairview QC no announcements made no water sa amin brgay fairview qc no announcements made

@gapolicarpio 2020-10-13T14:22:40.000Z @maynilad Why is there no water service in Brgy

Alabang/Cupang Muntinlupa? why is there no water service in brgy alabang cupang muntinlupa

@CharlieCanela 2018-04-24T01:30:57.000Z @MMDA sirs if i’m color coding in pasig but driving within the allowed window, can i cross c5 from megamall going to ortigas extension? sirs if i m color coding in pasig but driving within the allowed window can i cross c5 from megamall going to ortigas extension

RESULTS AND DISCUSSIONS

W ORDCLOUD V ISUALIZATION

A wordcloud visualization was generated to graphically represent the frequency of dominating subjects In a wordcloud visualization, an initial impression of the aggregated text data is obtained through illustration (Katre, 2019) The text sizes are scaled based on the relevance and occurrence of the subjects; a bigger text signifies a more relevant word Figure 8 presents wordcloud visualizations of the messages analyzed for different urban infrastructure components coming from the public, and figure 9 presents wordcloud visualizations of the messages coming from the responsible authorities

The wordcloud generated for the public’s messages underlined a general impression of frequently mentioned topics within public discussions in social media

In UIS component Roads, Bridges, and Highways, for example, it is evident that

“traffic”, pertaining to vehicular traffic congestions, surfaced as the most dominating subject in the collected messages Similarly, the prominence of the words

“interruption” and “supply”, for example, in the case of Water Supply, Sewerage, and Sanitation services points to the dominance of issues on water supply and service interruptions experienced by the public

Despite its evident lack of evident quantitative properties, wordcloud visualizations have been proven to be an effective tool to identify prominent themes in an otherwise bulky and unorganized text data which can be established as starting point for further supplemental analyses (DePaolo & Wilkinson, 2014) In the case of urban infrastructure management, these insights can be handled to initially recognize and isolate problem areas in development and management of urban infrastructure components and services Additionally, since this approach banks on online, free- form discussion, it naturally evolves with the movements of public discussion.

Figure 8 Wordcloud Visualizations for Tweets collected from the public

(a) Water Supply, Sewerage, and Sanitation (b) Roads, Highways, and Bridges

Figure 9 Wordcloud Visualizations of Tweets collected from the authorities

(a) Water Supply, Sewerage, and Sanitation (b) Roads, Highways, and Bridges

Furthermore, several insights can be derived by comparing the word cloud visualization of Tweets coming from the public against the Tweets coming from the responsible authorities Principally, since word clouds provide a general impression of the overlying subjects of interest, it hints to how the public and the authorities utilize social media platforms, such as Twitter, for urban infrastructure management

As mentioned earlier, for the Tweets coming from the public, it can be understood that the social media platform is primarily used to express concerns and overall experience of the public in interacting with different urban infrastructure components Conversely, it can be expected that the authorities responsible for management of infrastructure components and its attached services would consider the platform as an avenue to respond to these concerns and implement necessary strategies to further improve public infrastructure

For the case of Water Supply infrastructure, for example, as users consistently comment on interrupted services, the managing authorities respond by issuing

“service advisories” and reaching out (“contact”, “apologize”) to its users through direct interactions (“direct message”, “account”, “contract”), possibly hinting at their intention to resolve the users’ concerns

Notably, for the case of Transportation infrastructure, there appears to be an evident gap in this aspect Contrary to prominent mentions of “trains” by the public, the authority’s platform seems to be more dominated by advisories regarding aerial transportation (“flight”, “mnl”) with mentions of “trains” and “mrt” being only underlying, sub-dominant themes

Due to the principally qualitative nature of text analysis, and more so, word cloud visualizations, a more standardized evaluation of these kinds of observation remains to be integrated in management of UIS However, it is evident from the preliminary results that this approach warrants a consideration as a potential supplementary source of insight from the public regarding the performance of urban infrastructure components and its attached services

T OPIC M ODELING – L ATENT D IRICHLET A LLOCATION

Topic models were built to extract dominant topics for all UIS components of interest Similar to the approaches employed by Jiang et al (2016), D Zhang et al., (2018), and Y Zhang et al., (2019), the entire process relied upon optimizing the paramaters of the model and iteratively removing stop words that hold minimal to no research value, until an acceptable contextual convergence among the topics and keywords are achieved Upon optimizing the parameters of the model, the dominant topics for each UIS component, along with the ten most frequent terms per topic were obtained By principle of topic modeling, the ten representative terms for each topic were grouped by the algorithm based on similarities in word patterns, relevance, and frequency Appropriate topic labels that collectively describe the obtained keywords were then assigned by the researchers based on the context of the study and domain knowledge The topic model outputs are presented in Tables 4a to 7a and sample messages belonging to each extracted topic are presented as examples in tables 4b to 7b

Water Supply, Sewerage, and Sanitation

Table 4 (a) Topics extracted for Water Supply, Sewerage, and Sanitation

Count Assigned Topic Label Keywords

Concerns bill, supply, account, customers, metro, interruptions, service, concern, billing, payment

2 3,014 Service Interruptions interruption, city, service, advisory, brgy, supply, update, scheduled, pressure, schedule

Repair Works leak, notice, new, fix, help, pipe, need, meter, work, number

Count Assigned Topic Label Keywords

Improvement mwss (government authority), bills, duterte

(President of the Philippines), problem, project, action, ask, government, issue, response

(b) Messages belonging to the topics extracted for Water Supply, Sewerage, and Sanitation

@manilawaterph help us to compute our bill its so shocking when we recieve our monthly water bill

@maynilad what is the cause of water interruption here at proj8, congressional ave qc? what time will resume?

@maynilad still no action on the leak how long would it really take you to fix that problem? water is being wasted as we tweet

@duterte24x7 take actions senate & president duterte for no water due to pressure low & alibislas pinas

For Water Supply, Sewerage, and Sanitation infrastructure, it is apparent that messages from the public pertaining to billing and payment concerns dominated the discussions on water supply services (Table 4) The relatively high Tweet counts concerning service interruptions, repair works, and persistent calls for more reliable services also imply that these are major issues among the public Mention of key phrases and words specific to the local context such as brgy (ward), mwss

(Metropolitan Waterworks and Sewerage System – government authority regulating water supply services), and duterte (president of the Republic of The Philippines) were also evident, hinting at the relation of these words to issues such as localized

31 water supply concerns and calls to government authorities to take action for service improvement

From these results, many important insights can be drawn and utilized for improvement of UIS management practices For instance, the predominance of messages pertaining to billing and payments points to the public’s perceived inadequacy of current billing and payment systems by the water supply service providers, substantiating a need for improvement on this particular aspect Meanwhile, the prevalence of messages concerning service interruptions may be indicative of the providers’ failure to provide timely advisories and mitigating solutions for water supply interruptions, compromising the daily societal functions of the users

Table 5 (a) Topics extracted for Transportation

Count Assigned Topic Label Keywords

Operations of Light Rails LRT-1, LRT-2 and MRT-3 lrt, commute, operation, ride, mrt, service, passenger, platform, late, biyahe (journey)

2 3,755 Public Safety update, accident, running, trip, safe, line, people, pasahero (passenger), ride, new

Technical issues and breakdowns of major transportation lines operation, problem, advisory, open, technical, response, sira (broken), delayed, rides, resume

National Railway pnr, bicol (location, station), sta mesa (station), tutuban (station), clean, operations, trip, schedule, late, aircon

(b) Messages belonging to the topics extracted for Water Supply, Sewerage, and Sanitation

Operations of Light Rails LRT-1, LRT-2 and MRT-3 heads up, mrt commuters! the third dalian train set will serve passengers starting today until february 1, from 6am to 9pm | @dotrmrt3

2 Public Safety we dont want this type of accident but it totally sealed the deal for our puj modernization with safety officer includedkudos @dotrph

Technical issues and breakdowns of major transportation lines mrt-3 advisory: overhead line encountered a technical problem at the cubao station (southbound) the glitch was resolved at 8:48am | via @dotrmrt3

Operation of Philippine National Railway (Line 1) sad that there are no urgent plans to improve pnr service talks about the commuter and bicol lines are just that talk @dotrph has to move beyond press releases

For the Transportation component, the discussions regarding the operations of three operating light rails (LRT-1, LRT-2, and MRT-3) appear to take dominance while discussions pertaining to Philippine National Railway appear to be relatively dormant compared to the former Notably, the neutrality of the top keywords for both topics point to a lack of dominating sentiment among the messages which does not make it an operative indicator of an alarming concern or issue Interestingly, messages with public safety as an underlying theme emerged as a sub-dominant topic, along with messages pertaining to technical issues and breakdowns of major transportation lines

Putting these findings into context, the dominance of messages pertaining to major light rails may be presumed as an evidence of these transportation lines taking

33 the most volume of users compared to other public transportation modes available The Philippine National Railway has been established many years ahead of the light rail system, yet qualitative data from social media points to the latter having, observably, more user interaction and public engagement Acknowledging these preliminary results, certain steps can be taken to ensure that development and maintenance strategies are deployed effectively and proportional to users’ demands Additionally, in light of numerous messages pertaining to technical issues and public safety, identifying the major flaws of major transportation systems may point to a solution for an overall satisfactory experience by all stakeholders involved

For future infrastructure development strategies, such data can be used to identify the current and emerging demands of the users As mentioned, the dominance of messages regarding the Light Rail system may be taken as an indicator of volume and, if appropriate, may be considered as a preliminary indicator of the rail system’s susceptibility to becoming overwhelmed Putting these into perspective, certain measures can be implemented to ensure appropriate and proactive infrastructure planning approaches

Table 6 (a) Topics extracted for Roads, Highways, and Bridges

Count Assigned Topic Label Keywords

Major Roads and Exits southbound, exit, alabang (location), checkpoint, skyway (major highway), checkpoints, heavy, advisory, near, closed

Cashless Toll Payment System (RFID, easytrip) rfid, toll, lane, easytrip, lanes, free, exit, open, nb, pass

3 2,850 Road Accidents cause, people, accident, news, dyan (here), app, ask, budget, include, roads

Count Assigned Topic Label Keywords

4 1,412 Project Developments villar, mark (Secretary of Public

Works), build, construction, public, buildbuildbuild (government program), bridge, expressways, new, people

(b) Messages belonging to the topics extracted for Roads, Highways, and Bridges

Transport Situation in Major Roads and Exits it wasn't even 7 but the traffic's already so bad by alabang what is up, @officialslex?

Cashless Toll Payment System (RFID, easytrip) thanks how to register easytrip and rfid in one? i think there should be a info dissemination there since these confuse people and creating bottleneck especially in bocaue exit

@nlextraffic are the authorities doing something about the accident in san fernando? it's been 30 minutes the traffic is not moving

280 km of roads to be improved by adb's biggest infrastructure investment in

Similar to Transportation infrastructure, dominant topics for Roads, Highways, and Bridges also point to possible indications of volume and demand In topic 1 for example, mentions of “skyway”, a major thoroughfare, and “alabang”, a busy road network junction, along with descriptive words such as “heavy” and

“closed” indicate a relatively high-volume reporting of traffic situations in these areas The cashless toll payment system (RFID) “easytrip” (topic 2) also appears to have presence in online discussions, possibly implying a continuous increase in

35 adoption of the technology since it was introduced In parallel to the case of transportation sector, accidents and safety concerns (topic 3) also see frequent reporting in social media In prospect, because of the velocity in which social media data traverse network communications, this opportunity can be leveraged to improve emergency responses for road accidents Interestingly, in contrary to the cases of other infrastructure components, discussions on project developments surfaced as a sub-dominant topic for road infrastructure With mentions of “buildbuildbuild”, the government’s program aiming for aggressive push in infrastructure development, as well as “villar, mark”, the Secretary of Public Works and Highways, there is evidence of public interest in discussing upcoming infrastructure projects through online spaces With this information at hand, the authorities may evaluate and gauge the effectivity of social media as a channel to engage the public and collect opinion for future infrastructure projects as a way to assess public sentiment and opinion toward various development plans

Table 7 (a) Topics extracted for Urban Space

Count Assigned Topic Label Keywords

1 1,730 Number Coding policy coding, number, lifted, scheme, metro, tomorrow, people, suspended, new, happy

Traffic situation in major thoroughfares following an accident lane, edsa (major highway), accident, occupied, involving, advisory, car, vehicular, sb (southbound), bus

Traffic situation in major thoroughfares (C5 and EDSA) c5 (major highway), need, right, stop, edsa (major highway), job, drivers, best, many, situation

Traffic situation in major thoroughfares following a typhoon view, edsa (major highway), update, window, flood, passable, bike, routes, alert, magallanes (locality)

Count Assigned Topic Label Keywords

Local traffic situations (Northern Metro Manila) drive, balintawak (locality), moving, safely, edsa (major highway), fast, bound, north, street, makatitraffic (locality)

6 861 Fire alerts fire, city, turn, alarm, report, public, alert, response, could, interaction

(b) Messages belonging to the topics extracted for Urban Space

1 Number Coding policy due to the scheduled transport strike tomorrow, the number coding scheme is lifted for public utility vehicles only | @mmda

Traffic situation in major thoroughfares following an accident heavy traffic at kalayaan nb due to truck/sedan accident pls advise @mmda

Traffic situation in major thoroughfares (C5 and EDSA) traffic situation in edsa and c5 northbound?

@mmda alternates routes suggested coming from sucat thanks

Traffic situation in major thoroughfares following a typhoon flood alert: 13 inches at edsa p tramo extension nb not passable to light vehicles

Local traffic situations (Northern Metro Manila) dear @mmda, traffic along balintawak is horrendous every morning pls show us some luv and do somethin' abt it thank u

6 Fire alerts update re burning car at edsa-boni, @mmda clarifies: no fire just smoke but incident has already affected traffic flow @dziq990

The topics extracted for UIS component Urban Space are comparatively sparse relative to the other UIS components, which may be attributed mostly to its domain of responsibility The Metropolitan Manila Development Authority (MMDA) is the sole government authority handling various aspects of urban spaces management including transportation management, climate change-related policies, waste management, disaster prevention, and upkeep of public spaces Despite the lack of observable dominance among topics, traffic-related discussions remain present with mentions of major thoroughfares in the capital including “C5”, “EDSA”, and query- related keywords such as “passable”, “update”, “situation” which is indicative that the public is utilizing Twitter as an avenue to get real time traffic updates and situations especially in the event of heavy congestions, road accidents, and typhoons Prevalence of discussions pertaining to MMDA’s number coding policy – a move to regulate and limit vehicles traversing through major highways within specific window everyday – remains evident as well Similar to traffic condition queries, the presence of words “lifted”, and “suspended” hints at how the users utilize social media space to acquire information regarting the implementation of the policy on a day-to-day basis Since MMDA’s domain of responsibility is inherently more dynamic than the other governing bodies of other UIS components, social media data affords to match this required transfer of information between the using public and the government authority at very minimal cost

As noted from the outputs of topic models built to extract dominant themes of discussions pertaining to different UIS components, the utility and significance attributed to social media data, in general, vary across different domains Further, the insights that can be derived from such information also vary in importance and value Identifying issues, topics, and gauging public’s concerns on UIS components provides insight on critical areas for improvement that can be taken as more proactive in contrast to periodic monitoring of infrastructure components as it takes into account the interest of end users as direct stakeholders of UIS The velocity and volume in which social media data is received also makes it possible to understand the frequency and intensity of concerns which are both important dimensions that may not be acquired through conventional data collection techniques.

E VENT A NALYSIS

Since timestamps were also attached to raw messages collected, it was possible to analyze the development and occurrence of topics on a temporal perspective, that is, analysis of the evolution and emergence of topics in which insights such as levels of frequency of topics (high or low), and period of occurrence (duration) can be extracted Temporal analysis of Twitter data was first explored by Shah & Dunn (2019) and was found to have demonstrated the efficiency of the approach to understand and detect events across several domains including disaster response, impact of social events, and other forms of surveillance A similar analysis was also performed by Y Zhang et al (2019) with the objective of gaining comprehension on the trends of public discussions This research, specifically, relied on the theoretical baseline provided by Shah & Dunn (2019), but due to limited scope and constraints in availability of data, an approach closely parallel to Y Zhang et al (2019) was employed

As an exploratory approach, the topics taken from Water Supply, Sewerage, and Sanitation infrastructure component were taken for event analysis A truncated timeline of October 2018 to October 2020 was taken for understanding of the developments of discussions pertaining to the previously extracted topics (Figure 10)

Figure 10 Event Analysis of Topics Extracted for Water Supply, Sewerage, and Sanitation.

From the period of October 2018 to October 2019 there was no observable difference between the percentage shares of each topic in public discussions Presumably, this indicates that there were no evident dominance of a single topic against the others However, as the data showed, there was a considerable spike in discussions regarding demands for service improvement during December 2019 Putting this finding alongside the local context, it can be learned that this surge was most likely due to series of service deficiencies by the water concessionaires that led to the government’s threats of expropriating the services and seizing the operations of the private concessionaires (Corrales, 2019; Ranada, 2019) Around this time, Metropolitan Waterworks and Sewerage Systems, the chief regulating body of the two main water concessionaires Manila Water and Maynilad Water Services, also passed a decision to cancel the extension contracts of the private partners (Rivas, 2019) Public’s dissatisfaction on the water supply services, mainly, also led to the sharp plummet in the companies’ market valuation around this period (Rivas, 2019)

Another noticeable event in the extracted timeline was the uptrend of discussions pertaining to billings and payments from February 2020 to June 2020 In context, this period covered the emergence of COVID-19 pandemic present in the Philippines Since the pandemic brought several negative economic repercussions such as unemployment, closure of businesses, and overall decline in quality of living, it was understandable that the financial aspect of water insfrastructure services concerning the users, i.e., billings and payments, would get increased attention To cushion the economic impacts of the pandemic and to comply with the mass quarantine programs of the Philippine government, the concessionaires temporarily suspended payments from the beginning of March 2020 to end of May 2020, with all of the accumulated suspended payments to reflect on the June 2020 bills (Maynilad Water Services Inc., 2020; Manila Water Company Inc., 2020) Evidently, this move initiated discussions on online spaces in varying sentiments

In both scenarios, it was apparent that the trends in public opinion from social media was closely reflective of real life events, which, in turn, strengthens the argument of this research that there can be value extracted from mining social media data and several valuable insights can be derived from it Management of UIS is consisted of interconnected and complex facets, one of the most significant of which is stakeholders’ position which can be drawn from the public’s opinion Analyzing trends, movements, and developments in public discussion concerning urban infrastructure systems makes it possible for governing authorities to derive insights from historical data and make future decisions and projections based on the fact For example, in the case of Water infrastructure, the analyzed timeline gave an insight on how the sentiment on public discussion moved in response to deficiencies of their services and changes in billing processes In addition, this approach can be performed on different scales A weekly time frame, for example could possibly be beneficial for detecting events related to transportation such as volume and demands analysis

However, both in scope and approach, the methodology presented for event detection and analysis in this research is exploratory and only considers the fundamental process of analyzing aggregated qualitative data from social media In future studies, several, more advanced approach such as predictive models may be

41 derived to consider all of the dynamic elements in public sentiment and effectively apply findings for improvement of UIS management.

S ENTIMENT A NALYSIS

In line with the objective of determining the overall perception of the public on different UIS components and their attached services, the underlying sentiment of collected messages were analyzed using the built sentiment analysis model Table 8 presents in detail the distribution of sentiments across all four UIS components Tables 9 and 10 present samples exrtracted from the data pool with assigned sentiment values (positive or negative)

Table 8 Public Opinion Sentiment distribution for different UIS Components

UIS Component Positive Neutral Negative

Water Supply, Sewerage, and Sanitation 41% 22% 38%

The presence of considerable proportion of negative messages for all four UIS components were notable Particularly for water supply services and urban space, the share of negative public opinion appears to be relatively higher, which may be reflective of how satisfactorily these UIS components serve the interest of the public

As the underlying sentiment prevailing in public opinion may be suggestive of the performance of UIS, it may be possible to translate and interpret such results quantitatively For example, a specific set of public satisfaction benchmarks can be laid out by the government authorities which can be used to assess the performance of infrastructure components, in which measure of public opinion sentiment can be translated as rewards or penalties to service providers as a form of regulation Similarly, the dynamic nature of public opinion sentiment can be leveraged to analyze the changing perspective of the using public in response to the changes on the performance and reliability of different UIS components

Table 9 Sample Tweets with Negative Sentiment

UIS Component Message (Negative Sentiment Value)

Road widening? ROAD WIDENING?! Shitty work DPWH :|

C6 Expansion Project on halt!? This is causing accident and traffic

@DPWHph the corner of U.N & Paz M Guazon has been under repair for 3 months now It's still not finished Please fix the road already!!!

Dear @DPWHph, steamrollers Use 'em C5 and EDSA are terrible

@DPWHph your ill-timed, ill-planned, ill-prepared projects are very counter productive Costing everyone time, effort & money Plan better!

The glaring defects of the Dalian trains Heads must roll I remember seeing one running even rode on it? Tama ba @dotrmrt3 @DOTrPH?

We Even thought it was a testrun.Japanese, German techs detail Dalian train defects | http://Philstar.com http://po.st/3T1s3P via @po_st It's not a good day to be @dotrmrt3 today

Stopped at magallanes station All trains not operational

UIS Component Message (Negative Sentiment Value)

What's the technical problem about, @dotrmrt3?

Not being transparent with the people is worse than losing your no unloading/technical failure streak

@ManilaWaterPH 6 hours flushing? Wow that's a lot of contamination Water pressure is still low, so it's not resolved

@manilawaterph announcement was water interruption starts 9pm but we have no water since past 7 what happened?

@ManilaWaterPH late notice again? For how many more weeks the scheduled interruption will be?

@maynilad so you now have an unscheduled water interruption on our street WHAT THE HELL???

@maynilad Hi our water supply in Brgy 199 Tondo has been cut off again without prior notice It's too much of a hassle

What the hell we're stuck in traffic.no movement What's wrong with Roxas blvd @MMDA??

Can someone explain why the @mmda is wasting money on a motorcycle lane but no designated jeepney stops and no punishment for smoke belchers

Don't wait for them to kill a bunch of people with their recklessness before somthing is done Regulate! And crack down on the owners

@MMDA: Opo There was a traffic enforcer who tried to stop me fr turning right and giv me a ticket there pls investigate po.Such rudeness

UIS Component Message (Negative Sentiment Value)

Please do something about this insane traffic jam @MMDA Edsa megamall south bound

Table 10 Sample Tweets with Positive Sentiment

UIS Component Message (Positive Sentiment Value)

@DPWHph @MMDA Congrats I never thought your agencies can do it You did clear Hemady of parked vehicles Keep it like that always!

It's refreshing to see the gov't is really working for it's people Good job @DPWHph Saludo ako sa inyo!

Another legacy project I recommended to Sec Villar is the creation of a promenade on both sides of the Pasig River to allow people to walk or bike by the river I requested @DPWHph to submit an augmentation to their budget so we can locate the funds for this

The only key in achieving all the infra projects, 24x7 work without compromising the build quality Again BIG congratulations to

@DPWHph and Sec Mark Villar Action speaks louder than words!

The three road projects are worth a total of P3 billion and are under the Improving Growth Corridors in Mindanao Road Sector Project

@DPWHph https://portcalls.com/dpwh-begins-3-key-road-projects-to- improve-mindanao-growth-corridors

Another comfortable ride from the south, thanks again @DoTrPH for the @dotrmrt3 improvements #midweek (@ Yellow Line - North Avenue Station in Quezon City, Metro Manila) https://swarmapp.com/c/dKMuStbhn0Q Good news for moms! These P2P companies are offering free bus rides to mothers until Friday, May 11 | via @dotrmrt3

UIS Component Message (Positive Sentiment Value)

GOOD NEWS: DOTr MRT-3 is now running 18 trains!On May 10,

2018, at 8:27 pm, MRT-3 deployed its 18th train in the mainline.MRT-

3 last fielded 18 trains on Nov 16, 2017 | via @dotrmrt3

Not-so-crowded train coaches, doing a great maintenance job for commuters!!! @dotrmrt3 & @DoTrPH (@ Yellow Line - North Avenue Station in Quezon City, Metro Manila) https://swarmapp.com/c/9r31IY1KZEU

I have seen the efficiency of the transportation system in other countries I still believe that the Philippines can do the same

Commendations are a sign that you're doing your job right Congrats, team @ManilaWaterPH! Keep it up! :D #manilawater

@ManilaWaterPH My wish is for MWCI to continue providing world class water and used water services Cheers! #ManilaWaterat17

Yay! Water pressure here is back to normal Thanks @ManilaWaterPH

Good job, @ManilaWaterPH , for the quick reply re: the weak water pressure here at home Hopefully I get further feedback within the day

Fantastic to hear about how @maynilad has conserved water and reduced leakage so significantly, here in Manila at the WaterLinks Forum https://waterlinks.org/wl-forum-2018-1

Twitter is the portal if you want fast and reliable road info Thanks to

Thank you for the prompt responses I'm in Binondo now HOT! ,

@GhelSandra: @officialTIMYAP My beloved GMAnews &

Nice mural on the EDSA Cubao tunnel Now being stuck in traffic is no longer boring Congrats @MMDA!

UIS Component Message (Positive Sentiment Value)

Good job @MMDA putting blue bike lanes along Commonwealth Ave

& EDSA! Looking forward to having Bus Rapid Transit/ BRT too :)

Nice anti-jaywalking campaign along Libis near @eastwoodmall

From the pool of collected Tweets with assigned sentiment values, it is apparent that the nature of negatively-toned tweets mostly pertains to complaints on service or comment on dissatisfaction to underperformance of a certain infrastructure component On the contrary, positive Tweets appear to be hinting at the public’s commendation of the work being done in the upkeep of public infrastructure Furthermore, such in the case of roads, highways, and bridges, it is notable that messages referring to upcoming infrastructure projects tend to have positive sentiments

From the perspective of infrastructure management, many insights can be gained from understanding how the public perceives the quality and reliability of infrastructure components and services Specifically, since public satisfaction has been shown to have a strong correlation with the quality of urban infrastructure, understanding public sentiment plays a vital role in achieving the desired levels of performance of UIS.

B ENEFITS AND B ARRIERS TO A DOPTION

In order to further assess the benefits of utilizing social media data for UIS management, as well as the potential limiting barriers to its adoption, a series of structured interviews among experts involved in urban infrastructure development and management including infrastructure planning, implementation, operations, and monitoring and maintenance were conducted Having a minimum of 5 years

47 experience in the field, and having a decision-making role within any of the government authorities covered by the scope of the research were the other criteria considered in the selection of the interviewees A total of 16 participants were considered from the 22 that were initially invited A pre-formed interview questionnaire sheet was prepared and delivered through online means as a way to conduct the interview The profiles of the respondents, as well as the interview questionnaire can be found on Appendices B and C, respectively

Interview responses suggest that government authorities mainly involved in UIS management are generally optimistic on the potential benefits of utilizing social media data mining as an approach to extract insights from public opinion found in online spaces Principally, on matters of maintaining acceptable levels of stakeholders’ satisfaction on certain infrastructure components and its attached services, there was an agreeing consensus on the benefits that can be drawn from the approach In addition to that, most of the respondents agreed that mining social media data, instead of employing traditional methods of data collection such as surveys, would allow for a more instantaneous turnover of information at a lower cost The facility of online platforms, such as Twitter, as a bridge that closes the gap between the public and the government authorities and serves as an avenue for the users to freely express their sentiments on infrastructure services was also highlighted in the interviews Responses further suggest that several significant infrastructure management decision can directly benefit from analyzing public opinion through social media data mining Remediation of problems and decisions on maintenance programs based on the evolving needs of the public, acceleration and prioritization of projects, and improvement of contract designs were the major points cited by the experts

Despite the aforementioned benefits, social media data mining for UIS management, without a doubt, remains an uncharted territory Because of its infancy and the inherent limitations attached to social media primarily, there remains overt limitations that serve as barriers to successful adoption and integration of the approach Responses from the interviewees were indicative that the presence of irrelevant, fictitious, and malicious comments characteristic of online platforms may be a source of concern that may impede the goals of the approach Furthermore, most of the respondents were convinced that social media data is primarily constituted of negative sentiments, and may not be the best avenue for evaluating the performance of urban infrastructure systems Fully integrating social media data mining in traditional systems of assessing infrastructure performance can also problematic given the inconsistencies between the established approaches and the proposed approach For instance, in traditional record-keeping of stakeholders’ concerns regarding a UIS component, the issues are already identified and clustered beforehand, where periodic assessments are mainly focused on reusing the same evaluation criteria On the other hand, in social media data mining, where topic models can be extracted, the issues or concerns are iteratively identified, adapting to dynamic changes in public opinion and social sentiment In other words, the variability of results expected of social media data mining remains incompatible against the otherwise consistent and static conventional data management practices in management of urban infrastructure systems

CONCLUSIONS AND RECOMMENDATIONS

C ONCLUSIONS

In view of inherent limitations of current management strategies in UIS, particularly in performance monitoring and evaluation, this study sought a more proactive, user-involved, and cost-effective approach that can help support planning and response strategies in routine maintenance, preservation, and improvement of urban infrastructure components This study also highlights public opinion as both an important social aspect and a potential key driver for a more user-involved decision making in infrastructure management

This research explored the possibility of extracting valuable insights pertaining to urban infrastructure systems from user-generated social media data through various text mining techniques Wordcloud visualization provided a general overview of salient topics in public discussions that reflected the common issues of UIS components from the perspective of the end users Similarly, the topic model was able to extract dominant topics from unstructured collection of qualitative data which proved beneficial for understanding the present and emerging concerns of the public regarding UIS Additionally, event analysis was explored to determine the applicability and significance of the topics derived when analyzed on a temporal perspective, giving light to further understanding the occurrence, emergence, and development of public opinion and how insights derived from social media data mining can be reflective of real and practical events Furthermore, with the goal of comprehending how the using public perceives the quality and reliability of UIS components and their attached services, sentiment analysis was performed as an evaluation measure of the overall public perception which can be used as a determining factor of the level of satisfaction of the users on the level of performance of UIS components Ultimately, with the goal of assessing the potential benefits of the approach, and the limiting factors that serve as barriers for its adoption, opinions of experts in urban infrastructure management were sought The interview results were indicative that social media data mining as an approach to extract valuable insights from public opinion for the benefit of UIS carries strong potential as a tool to evaluate stakeholders’ position on infrastructure components and its attached services, owing to its capability to draw insights in an expeditious manner However, there remains overt limitations that serve as barriers to its adoption including the presence if irrelevant, fictitious, and malicious information on social media and the incompatibility of the approach to current data management systems.

R ECOMMENDATIONS FOR S TAKEHOLDERS

The use of social media data mining for UIS management affords many possibilities that were otherwise not achievable in traditional data collection method and techinques such as surveys The advent of social media itself, as a new paradigm, has made it possible for most individuals to access and create voluminous data incomparable to means available before This research explored this potential and brought to light, an approach that could potentially improve stakeholder management and connection crucial to any urban infrastructure management system It is recommended that government authorities, in case of eventual adoption of the approach, evaluate and assess properly the applicability, accuracy, and legitimacy of the process Improving a system of approach that is in its infancy is a daunting task, hence expert judgment and discretion will always be critical for further development and use of this approach.

R ECOMMENDATIONS FOR F UTURE R ESEARCH

Social media data mining offers promising potential to improve and revitalize current urban infrastructure management practices However, for it to realize its full efficiency, its existing limitations must be considered and its vulnerabilities must be minimized In future studies of the same interest, the author of this research recommends looking into ways to improve the quality of collected social media data Furthermore, developing predictive models to derive topics and detect events relevant to UIS management could also improve robustness of the approach

Abu-Samra, S., Ahmed, M., & Amador, L (2020) Asset Management Framework for Integrated Municipal Infrastructure Journal of Infrastructure Systems,

26(4), 04020039 https://doi.org/10.1061/(asce)is.1943-555x.0000580

Abu Samra, S., Ahmed, M., Hammad, A., & Zayed, T (2018) Multiobjective Framework for Managing Municipal Integrated Infrastructure Journal of

Construction Engineering and Management, 144(1), 1–13 https://doi.org/10.1061/(ASCE)CO.1943-7862.0001402

Amador, L E., & Magnuson, S (2011) Adjacency modeling for coordination of investments in infrastructure asset management: Case study of Kindersley, Saskatchewan, Canada Transportation Research Record, 2246, 8–15 https://doi.org/10.3141/2246-02

Azhar, S., Riaz, Z., & Robinson, D (2019) Integration of Social Media in Day-to- Day Operations of Construction Firms Journal of Management in

Engineering, 35(1), 1–7 https://doi.org/10.1061/(ASCE)ME.1943-

Barbier, G., & Liu, H (2011) Social Network Data Analytics Social Network Data

Blei, D., Ng, A., & Jordan, M (2003) Latent Dirichlet Allocation In J Lafferty (Ed.), Journal of Machine Learning Research (Vol 3, pp 993–1022) Elsevier https://linkinghub.elsevier.com/retrieve/pii/B9780124115194000069

Chang, P C., Flatau, A., & Liu, S C (2003) Review paper: Health monitoring of civil infrastructure Structural Health Monitoring, 2(3), 257–267 https://doi.org/10.1177/1475921703036169

Chen, L., Henning, T F P., Raith, A., & Shamseldin, A Y (2015) Multiobjective optimization for maintenance decision making in infrastructure asset management Journal of Management in Engineering, 31(6), 1–9 https://doi.org/10.1061/(ASCE)ME.1943-5479.0000371

Chen, Y., Wang, Q., & Ji, W (2020) Rapid Assessment of Disaster Impacts on Highways Using Social Media Journal of Management in Engineering, 36(5), 1–11 https://doi.org/10.1061/(ASCE)ME.1943-5479.0000836

Diao, C., Liang, R., Sharma, D., & Cui, Q (2020) Litigation Risk Detection Using Twitter Data Journal of Legal Affairs and Dispute Resolution in Engineering and Construction, 12(1), 1–9 https://doi.org/10.1061/(ASCE)LA.1943-

Elbehairy, H., Hegazy, T., & Soudki, K (2009) Integrated Multiple-Element

Bridge Management System Journal of Bridge Engineering, 14(3), 179–187

Evans-Cowley, J S., & Griffin, G (2012) Microparticipation with Social Media for Community Engagement in Transportation Planning Transportation Research

Flintsch, G W., & Chen, C (2004) Soft computing applications in infrastructure management Journal of Infrastructure Systems, 10(4), 157–166 https://doi.org/10.1061/(ASCE)1076-0342(2004)10:4(157)

Frangopol, D M., & Kim, S (2011) Service life, reliability and maintenance of civil structures Service Life Estimation and Extension of Civil Engineering

Ghodoosi, F., Abu-Samra, S., Zeynalian, M., & Zayed, T (2018) Maintenance cost optimization for bridge structures using system reliability analysis and genetic algorithms Journal of Construction Engineering and Management, 144(2), 1–

10 https://doi.org/10.1061/(ASCE)CO.1943-7862.0001435

Giustozzi, F., Flintsch, G W., & Crispino, M (2012) Environmental Analysis of Preventive Maintenance Treatments on Road Pavements 8th International

Conference on Managing Pavement Assets, 1

Guillaumot, V M., Durango-Cohen, P L., & Madanat, S M (2003) Adaptive optimization of infrastructure maintenance and inspection decisions under performance model uncertainty Journal of Infrastructure Systems, 9(4), 133–

Hadjidemetriou, G M., Christodoulou, S E., & Vela, P A (2016) Automated detection of pavement patches utilizing support vector machine classification

Proceedings of the 18th Mediterranean Electrotechnical Conference:

Intelligent and Efficient Technologies and Services for the Citizen, MELECON

2016, April, 18–20 https://doi.org/10.1109/MELCON.2016.7495460

Han, B., Cook, P., & Baldwin, T (2014) Text-based twitter user geolocation prediction Journal of Artificial Intelligence Research, 49, 451–500 https://doi.org/10.1613/jair.4200

Harris, D K., Alipour, M., Acton, S T., Messeri, L R., Vaccari, A., & Barnes, L

E (2017) The Citizen Engineer: Urban Infrastructure Monitoring via Crowd- Sourced Data Analytics In J G Soules (Ed.), Structures Congress 2017 (pp 495–510) American Society of Civil Engineers

Jiang, H., Lin, P., & Qiang, M (2016) Public-opinion sentiment analysis for large hydro projects Journal of Construction Engineering and Management, 142(2), 1–12 https://doi.org/10.1061/(ASCE)CO.1943-7862.0001039

Kapoor, K K., Tamilmani, K., Rana, N P., Patil, P., Dwivedi, Y K., & Nerur, S (2018) Advances in Social Media Research: Past, Present and Future

Information Systems Frontiers, 20(3), 531–558 https://doi.org/10.1007/s10796-017-9810-y

Khatri, K B., Vairavamoorthy, K., & Akinyemi, E (2012) Framework for

Computing a Performance Index for Urban Infrastructure Systems Using a Fuzzy Set Approach Journal of Infrastructure Systems, 17(4), 163–175 https://doi.org/10.1061/(ASCE)IS.1943-555X.0000062

Kim, S., & Frangopol, D M (2017) Efficient multi-objective optimisation of probabilistic service life management Structure and Infrastructure

Klašnja, M., Barberá, P., Beauchamp, N., Nagler, J., & Tucker, J A (2015)

Measuring public opinion with social media data In The Oxford Handbook of

Polling and Polling Methods (Issue April 2018) https://doi.org/10.1093/oxfordhb/9780190213299.013.3

Koch, C., Jog, G M., & Brilakis, I (2013) Automated pothole distress assessment using asphalt pavement video data Journal of Computing in Civil Engineering,

27(4), 370–378 https://doi.org/10.1061/(ASCE)CP.1943-5487.0000232

Kuhn, K D (2010) Network-level infrastructure management using approximate dynamic programming Journal of Infrastructure Systems, 16(2), 103–111 https://doi.org/10.1061/(ASCE)IS.1943-555X.0000019

Kyriakou, C., Christodoulou, S E., & Dimitriou, L (2019) Smartphone-Based Pothole Detection Utilizing Artificial Neural Networks Journal of

Infrastructure Systems, 25(3), 1–8 https://doi.org/10.1061/(ASCE)IS.1943-

Lakhani, K., & Panetta, J (2016) The principles of distributed innovation

Successful OSS Project Design and Implementation: Requirements, Tools, Social Designs and Reward Structures, April, 7–26 https://doi.org/10.1162/itgg.2007.2.3.97

Leung, M., Yu, J., & Liang, Q (2013) Improving Public Engagement in

Construction Development Projects from a Stakeholder’s Perspective Journal of Construction Engineering and Management, 1–13 https://doi.org/10.1061/(ASCE)CO.1943-7862

Martín, Y., Cutter, S L., & Li, Z (2020) Bridging Twitter and Survey Data for Evacuation Assessment of Hurricane Matthew and Hurricane Irma Natural

Hazards Review, 21(2), 1–14 https://doi.org/10.1061/(ASCE)NH.1527-

Marzouk, M., & Enaba, M (2019) Text analytics to analyze and monitor construction project contract and correspondence Automation in Construction,

98(December 2017), 265–274 https://doi.org/10.1016/j.autcon.2018.11.018

Mild, P., & Salo, A (2009) Combining a Multiattribute Value Function with an Optimization Model: An Application to Dynamic Resource Allocation for Infrastructure Maintenance Decision Analysis, 6(3), 139–152 https://doi.org/10.1287/deca.1090.0143

Miner, G., Elder, J., Nisbet, R A., Thompson, J., & Foley, R (2012) Practical Text

Mining and Statistical Analysis for Non-structured Text Data Applications (1st

Mishalani, R G., & Gong, L (2009) Optimal sampling of infrastructure condition: Motivation, formulation, and evaluation Journal of Infrastructure Systems,

Nik-Bakht, M., & El-Diraby, T E (2020) Beyond Chatter: Profiling Community Discussion Networks in Urban Infrastructure Projects Journal of

Infrastructure Systems, 26(3), 1–15 https://doi.org/10.1061/(ASCE)IS.1943-

Nik-Bakht, Mazdak, & El-diraby, T E (2016) Communities of Interest-Interest of Communities: Social and Semantic Analysis of Communities in Infrastructure Discussion Networks Computer-Aided Civil and Infrastructure Engineering,

Qaiser, S., & Ali, R (2018) Text Mining: Use of TF-IDF to Examine the

Relevance of Words to Documents International Journal of Computer

Applications, 181(1), 25–29 https://doi.org/10.5120/ijca2018917395

Radopoulou, S C., & Brilakis, I (2017) Automated Detection of Multiple

Pavement Defects Journal of Computing in Civil Engineering, 31(2), 1–14 https://doi.org/10.1061/(ASCE)CP.1943-5487.0000623

Rashedi, R., & Hegazy, T (2016) Holistic analysis of infrastructure deterioration and rehabilitation using system dynamics Journal of Infrastructure Systems,

22(1), 1–10 https://doi.org/10.1061/(ASCE)IS.1943-555X.0000273

Salman, A., Moselhi, O., & Zayed, T (2013) Scheduling model for rehabilitation of distribution networks using MINLP Journal of Construction Engineering and Management, 139(5), 498–509 https://doi.org/10.1061/(ASCE)CO.1943-

Sánchez-Silva, M., Frangopol, D M., Padgett, J., & Soliman, M (2016)

Maintenance and operation of infrastructure systems: Review Journal of

Structural Engineering (United States), 142(9), 1–16 https://doi.org/10.1061/(ASCE)ST.1943-541X.0001543

Sánchez-Silva, M., & Klutke, G.-A (2016) Reliability and Life-Cycle Analysis of

Deteriorating Systems https://doi.org/10.1080/13854049708407039

Seto, T., & Sekimoto, Y (2019) Trends in citizen-generated and collaborative urban infrastructure feedback data: Toward citizen-oriented infrastructure management in Japan ISPRS International Journal of Geo-Information, 8(3) https://doi.org/10.3390/ijgi8030115

Sohail, M., Cavill, S., & Cotton, A P (2005) Sustainable operation and maintenance of urban infrastructure: Myth or reality? Journal of Urban

Planning and Development, 131(1), 39–49 https://doi.org/10.1061/(ASCE)0733-9488(2005)131:1(39)

Song, J., Kim, J., & Lee, J.-K (2018) NLP and Deep Learning-based Analysis of

Building Regulations to Support Automated Rule Checking System – The

International Association for Automation and Robotics in Construction Isarc https://www.iaarc.org/publications/2018_proceedings_of_the_35th_isarc/nlp_a nd_deep_learning_based_analysis_of_building_regulations_to_support_autom ated_rule_checking_system.html

Tafazzoli, M (2017) Strategizing sustainable infrastructure asset management in developing countries International Conference on Sustainable Infrastructure

2017: Policy, Finance, and Education - Proceedings of the International

Conference on Sustainable Infrastructure 2017, 375–387 https://doi.org/10.1061/9780784481202.036

Tang, L., Shen, G Q., Skitmore, M., & Wang, H (2015) Procurement-related critical factors for briefing in public-private partnership projects: Case of Hong Kong Journal of Management in Engineering, 31(6) https://doi.org/10.1061/(ASCE)ME.1943-5479.0000352

Tang, L., Zhang, Y., Dai, F., Yoon, Y., & Song, Y (2018) What Construction Topics Do They Discuss in Social Media? A Case Study of Weibo in China In

C Wang, C Harper, Y Lee, R Harris, & C Berryman (Eds.), Construction

Research Congress 2018: Construction Information Technology (pp 612–621)

American Society of Civil Engineers

Tang, L., Zhang, Y., Dai, F., Yoon, Y., Song, Y., & Sharma, R S (2017) Social Media Data Analytics for the U.S Construction Industry: Preliminary Study on Twitter Journal of Management in Engineering, 33(6), 1–15 https://doi.org/10.1061/(ASCE)ME.1943-5479.0000554

Torres-Machi, C., Osorio-Lird, A., Chamorro, A., Videla, C., Tighe, S L., &

Mourgues, C (2018) Impact of environmental assessment and budgetary restrictions in pavement maintenance decisions: Application to an urban network Transportation Research Part D: Transport and Environment, 59, 192–204 https://doi.org/10.1016/j.trd.2017.12.017

Tracy, A., Klucik, R., Javernick-Will, A., & Poleacovschi, C (2018) New Disasters in the Twittersphere: How Communities Utilize Social Media to Seek and Share Information in the Wake of Induced Seismicity In C Wang, C Harper,

Y Lee, R Harris, & C Berryman (Eds.), Construction Research Congress

2018: Construction Information Technology (pp 524–534) American Society of Civil Engineers

UN-Habitat (2015) 18 – Urban Infrastructure and Basic Services, Including

Energy In Habitat III Issue Papers (Vol 2015, Issue May) https://www.habitat3.org/the-new-urban-agenda/issue-papers

Valentin, V., Abraham, D., Mannering, F., & Mostafavi, A (2012) Assessment of public opposition to infrastructure developments: The case of nuclear power projects Construction Research Congress 2012: Construction Challenges in a

Flat World, Proceedings of the 2012 Construction Research Congress, 1,

Valentin, V., & Bogus, S M (2013) Public opinion as an indicator of the social sustainability of construction projects ICSDEC 2012: Developing the Frontier of Sustainable Design, Engineering, and Construction - Proceedings of the

2012 International Conference on Sustainable Design and Construction, 561–

Valentin, V., Naderpajouh, N., & Abraham, D (2018) Impact of Characteristics of Infrastructure Projects on Public Opinion Journal of Management in

Engineering, 34(1), 1–11 https://doi.org/10.1061/(ASCE)ME.1943-

Van Ryzin, G G., Immerwahr, S., & Altman, S (2008) Measuring Street

Cleanliness: A Comparison of New York City’s Scorecard and Results from a Citizen Survey Public Administration Review, 68(2), 295–303 https://doi.org/10.1111/j.1540-6210.2007.00863.x

59 von Hippel, E (2005) Democratizing Innvation (pp 1–220)

Walker, R S., Pettitt, J A., Scruggs, K T., & Mlakar, P F (2014) Data collection and organization by smartphone for infrastructure assessment Journal of

Infrastructure Systems, 20(1), 1–4 https://doi.org/10.1061/(ASCE)IS.1943-

Wang, P., & Shi, S (2018) Empirical Analysis on How Urban Infrastructure

Wang, Q., & Taylor, J E (2016) Process Map for Urban-Human Mobility and Civil Infrastructure Data Collection Using Geosocial Networking Platforms

Journal of Computing in Civil Engineering, 30(2), 1–11 https://doi.org/10.1061/(ASCE)CP.1943-5487

Wang, W J., Haase, T W., & Yang, C H (2020) Warning Message Elements and Retweet Counts: An Analysis of Tweets Sent during Hurricane Irma Natural

Hazards Review, 21(1), 1–11 https://doi.org/10.1061/(ASCE)NH.1527-

Ward, B., & Savić, D A (2012) A multi-objective optimisation model for sewer rehabilitation considering critical risk of failure Water Science and

Technology, 66(11), 2410–2417 https://doi.org/10.2166/wst.2012.393

World Bank (1994) World Development Report 1994: Infrastructure for

Development In World Development Indicators https://doi.org/10.2307/j.ctt18fsc3v

World Bank (2020) Infrastructure in Asia and the Pacific: Road Transport,

Electricity, and Water and Sanitation Services in East Asia, South Asia, and the Pacific Islands

Yuan, F., & Liu, R (2020) Mining Social Media Data for Rapid Damage

Assessment during Hurricane Matthew: Feasibility Study Journal of

Computing in Civil Engineering, 34(3), 1–14 https://doi.org/10.1061/(ASCE)CP.1943-5487.0000877

Zhang, D., Qiang, M., Jiang, H., Wen, Q., An, N., & Xia, B (2018) Social sensing system for water conservation project: A case study of the South-to-North Water Transfer Project in China Water Policy, 20(4), 667–691 https://doi.org/10.2166/wp.2018.141

Zhang, F., Fleyeh, H., Wang, X., & Lu, M (2019) Construction site accident analysis using text mining and natural language processing techniques

Automation in Construction, 99(June 2018), 238–248 https://doi.org/10.1016/j.autcon.2018.12.016

Zhang, Y., Li, D., & Li, C (2019) Public Transportation Analysis Based on Social Media Data CICTP 2019, 1517–1529 https://doi.org/10.1061/9780784482292.133

Appendix A – Pre-processing techniques and Stopwords dictionary applied

Lowercasing and Hyperlinks removal using python module “re” import re def preprocessor(text):

#pattern = re.compile() text = re.sub(']*>', '', text)

#question_mark = re.findall('[£$]',text)

#emoticons = re.findall('(?::|;|=)(?:-)?(?:\)|\(|D|P)', text) text = re.sub('http\S+\s*', ' ', text) text = re.sub('\r\n', ' ', text.lower()) #+ ' ' + '

'.join(emoticons).replace('-', '')#+' '.join(question_mark)

#text = re.sub('@[A-Za-z0-9_]+\s', '', text) return text

Removal of Twitter usernames and hashtags using python module “re” import re from IPython.display import clear_output def preprocessor(text): if type(text) != 'str': text = str(text) text = re.sub('@\s?[A-Za-z0-9_]+\S', ' ', text) text = re.sub(''', "'", text)

# text = re.sub('#[A-Za-z0-9_]+\S', ' ', text) text = re.sub('[^A-Za-z0-9]', ' ', text) text = re.sub('amp', " ", text) text = re.sub('quot', " ", text) return text

English stopwords dictionary used from NLTK library import nltk nltk.download('stopwords') from nltk.corpus import stopwords i me my myself we our ours ourselves you your yours yourself yourselves he him his himself she her hers herself it its itself they them their theirs themselves what which who whom this that these those am is are was were be been being have has had having do does did doing a an the and but if or because as until while of at by for with about against between into through during before after above below to from up down in out on off over under again further then once here there when where why how all any both each few more most other some such no nor not only own same so than too very s t can will just don should now

Filipino/Tagalog stopwords dictionary used from AdverTools library import spacy

!pip install advertools import advertools as adv

'ibaba', 'ibabaw', 'ibig', 'ikaw', 'ilagay', 'ilalim', 'ilan', 'inyong', 'isa', 'isang', 'itaas', 'ito', 'iyo', 'iyon', 'iyong', 'ka', 'kahit', 'kailangan', 'kailanman', 'kami', 'kanila', 'kanilang', 'kanino', 'kanya', 'kanyang', 'kapag', 'kapwa', 'karamihan', 'katiyakan', 'katulad', 'kaya', 'kaysa', 'ko', 'kong', 'kulang', 'kumuha', 'kung', 'laban',

'lahat', 'lamang', 'likod', 'lima', 'maaari', 'maaaring', 'maging', 'mahusay', 'makita', 'marami', 'marapat', 'masyado', 'may', 'mayroon', 'mga', 'minsan', 'mismo', 'mula', 'muli', 'na', 'nabanggit', 'naging', 'nagkaroon', 'nais', 'nakita', 'namin', 'napaka', 'narito', 'nasaan', 'ng', 'ngayon', 'ni', 'nila', 'nilang', 'nito', 'niya', 'niyang', 'noon',

'o', 'pa', 'paano', 'pababa', 'paggawa', 'pagitan',

'pamamagitan' , 'panahon', 'pangalawa', 'para', 'paraan', 'pareho', 'pataas', 'pero', 'pumunta', 'pumupunta', 'sa',

'saan', 'sabi', 'sabihin', 'sarili', 'sila', 'sino', 'siya', 'tatlo', 'tayo', 'tulad', 'tungkol', 'una', 'walang'

Thank you for participating in our study Through this interview form, you will be asked a series of questions (opinion-based) pertaining to the research study presented to you If you have yet to see the presentation, please do so before proceeding to the subsequent sections of this form

To facilitate completion of this form, please be noted of the following:

1 The questions or the research itself may not be directly aligned with your current professional role or the role of your organization In this light, we kindly ask you to answer based on your personal and technical judgment along with your general industry experience

2 The questions are presented in English However, please know that your responses may be free-form and need not be restricted to a certain tone, arrangement, or language You may answer in English, Filipino, or a mix of both (Taglish)

3 All of the information you will be providing here will be kept confidential and will only be used for the intended purposes of this research Your personal details such as name and email will only be used as reference to verify the legitimacy of your participation in this study We will not keep your information, nor we will send unrelated unsolicited emails without prior permission

4 Lastly, we would like to extend our sincerest gratitude for your participation in this study

5 How long have you been involved in urban infrastructure planning/development/assessment/management?

6 Please briefly describe your industry experience in relation to urban infrastructure planning/assessment/management

1 In your opinion, what are the benefits that can be derived in using social media data for infrastructure planning/assessment/management?

2 In your opinion, what are the disadvantages, drawbacks, or limitations of using social media data for this purpose?

3 Based on the potential of sample results presented, and your own opinion, what infrastructure management decisions can be done with these insights at hand? (For example, can we use it to decide a next infrastructure project? Decide on new contract designs? Decide on maintenance decisions? etc.) All elaborations are welcome

4 Do you think we can eventually fully integrate the use of social media data in infrastructure management? If not, why?

5 Do you have any comments, questions, or suggestions for the research?

Name UIS Component involvement Role/Position/Title Organization Involvement period/Industry experience

Respondent 1 Water Supply, Sewerage, and

Officer MWSS Regulatory Office 5-10 years

Respondent 2 Water Supply, Sewerage, and

Officer MWSS Regulatory Office more than 10 years

Respondent 3 Water Supply, Sewerage, and

AVP, Technical Planning, Water Supply Operations, MAYNILAD Maynilad Water Services Inc more than 10 years Respondent 4 Water Supply, Sewerage, and

Supervising Water Utilities Regulation Officer MWSS Regulatory Office more than 10 years Respondent 5 Water Supply, Sewerage, and

Sanitation Technical Regulation Head Manila Water Company, Inc more than 10 years Respondent 6 Water Supply, Sewerage, and

Sanitation Public Policy Department Head Manila Water Company, Inc 5-10 years

Respondent 7 Roads, Highways, and Bridges Engineer III DPWH 5-10 years

Respondent 8 Urban Space Planning Officer I MMDA 3-5 years

Respondent 9 Urban Space Engineer MMDA 3-5 years

Respondent 10 Urban Space Planning officer IV MMDA more than 10 years

Respondent 11 Roads, Highways, and Bridges Senior Technical Head DPWH more than 10 years

Respondent 12 Transportation Engineer III DOTr 5-10 years

Respondent 13 Transportation Traffic Engineering and

Management Head DOTr more than 10 years

Respondent 14 Urban Space Project Evaluation Officer II MMDA 3-5 years

Respondent 15 Roads, Highways, and Bridges Engineer IV DPWH more than 10 years

Respondent 16 Roads, Highways, and Bridges PPP and Special Projects Head DPWH more than 10 years