HOi thao ICT.rda'06 Proceedings ofICT.rda'06 Hanoi May 20-21,2006 DEFINING QUERY PROCESSING STEPS OVER j0:TEROGENEOUS LEARNING INFORMATION RESOURCES BY USING DESCRIPTION LOGIC KNOWLEDGE BASE Xiir ly truy van tren c^c nguon tai nguyen hoc tap da tap vci cor so tri thirc Logic mieu ta Hoang Thi Anh Duong, Nguyen Thanh Binh Abstract One of the most critical problems arising in E-leaming systems is accessing and integrating heterogeneous learning information resources, which often formatted in incompatible ways and even worse, represented using incompatible assumptions Characterized by high expressivity together with decidability DescrqUion Logics can be considered to ensure both the soundness and completeness of quay processing, one of the most challenging issues in relation lo effective information integration Motivated by the emerging need for effective query processing strategies in the field of E-leaming this paper represents a knowledge-based approach lo answering queries with ontology concepts in terms of Description Logics More specifically, we describe in detail query processing steps optimized on the basis of Description Logic knowledge base Thus, the role of Description Logics in facilitating the accessibility i^ a huge number of heterogeneous and autonomous learning repositories is emphasized Tir khda: E-learning, Knowledge Base, Ontology, Description Logics Query Processing Tom tdt Mgt Irong nhiing vdn di quan trgng dang dugc ddt cdc hf thdng E-learning Id khd ndng truy cgp vd tich hgp nhirng ngudn tdi nguyen thdng lin hgc tgp da Igp thu&ng dugrc dinh dgng v&i cdch thirc khdng luang thich Idn V&i khd ndng biiu diin cimg khd ndng cd thi tu quyil dinh cao Description Logics ddm bdo tinh chinh xdc vd tinh ddy dii eia qud trinh xu ly truy vdn, mgt nhiing thdch thic nhdm dgt dugc khd ndng tich hgp hiiu qud thdng lin Nhdm ddp ting yiu cdu phdi cd nhiing chiin hrgc xu ly truy vdn hiiu qud mdi tru&ng E-leaming, bdi bdo ndy biiu diin mgi hu&ng tiip cgn dua trin nin Idng tri thirc di Ird l&i nhihig cdu truy cdn v&i cdc khdi niim oiUology dugc biiu diin du&i hinh thirc Description Logics Ddc biit, chiing tdi mieu td chi tiit cdc bu&c xir ly truy vdn dugrc tdi uu hda dua trin nin tdng ca sd Iri thirc Description Logics Qua do, bdi bdo nhdn mgnh vai trd cua Description Logics hd trg khd ndng truy cgp mgt lugng l&n cdc kho tdi nguyen hgc Igp da tgp vd tutrf Keywords: E-learning Cas& tri thuc Ontology, Logic mieu Id Xafylruy vdn INTRODUCTION The Internet and more particularly the Web have been making widely accessible a huge range of digital resources The current state of affairs is that the task of integrating together relevant information involves searching for pieces of information from a wide range of heterogeneous resources and manually combining them into a whole Besides, among all other web-based information systems, E-leaming is one of the fastest growing and universally accepted [6, 8] Furthermore, die possibility to just in time, efficiently integrated access to multiple learning information resource has been the 237 K.y yeu HQi thao ICT.rda'06 ProceeditigsoiiCl.rda'06 Hanoi May focus of much research in the field of Eleaming, but still remains a significant challenge Heterogeneity may arise from syntactic, structural and semantic differences in the learning information resources Important structural and syntactic problems could be solved by various initiatives in the standardization of learning object metadata and the emergence of specifications towards the standardization of other aspects of learning objects and learning processes by organizations such as IMS [10] and ADL [9] However, we must deal with a situation where various metadata repositories, with standard and non-standard metadata, have been developed independently from each other, and in addition, most of metadata standards lack formal semantic [2] As the result, the main challenges in this respect are linked with the identification of semantically related information in different learning information resources and semantic heterogeneity arising from inconsistencies in terminology and meanings [1] Recognized as having considerable advantages, many recent information integration systems have used conceptual models expressed in expressive emd decidable Description Logics [4] The use of Description Logics as both high-level ontology language and query language has several advantages: providing an expressive global modeling language; enabling the use of Description Logic reasoning to assist in building consistent conceptual models and in accurately query reformulating; and using intensional knowledge captured in Description Logic ontology in query translation and optimization In this paper, we describe learning resources using ontology as a further step to clarify the educational domain and the content of the resources [6] Consequently, the main contribution of this paper is to introduce a general approach that focuses on processing steps of a query formulated based on Description Logics when the answer must be found in the underlying metadata r In this approach, we use onto semantic layer over schemas in transform queries to other semantic ones with respect to local leamint schemas We assume the pre-ex ontologies in term of Deseriptic associated with the schemas In thi Description Logics can be used extc support the mapping between global ontology-based schemas, which is al of answering queries in semantic age The rest of this writing is orgi follows: section introduces some ap related to our work; section focuse: of transformation rules that has beer and taken into account possible r between learning global and local c based schemas; in section 4, we pre overview of query processing steps I Description Logics knowledge base; section S gives a summary of what achieved and future works RELATED WORKS Our work is related to research within of query optimization, especially S( query optimization as well as the appl of Description Logics for query proi purposes Over the years, query optimizatic received a particular attention in web information systems Different appn have been proposed to improve the effii of query answering [1] However, the approaches are mostly based on form: that assume a closed world Sue assumption is clearly not adequate a always have to deal with incon knowledge With the rapid development of distril and heterogeneous repositories on the semantic query optimization has considered as a promising approach tc deployment of query answering as we data integration Semantic query optimizi techniques have emerged to provide effective enhancement to traditional qi 238 Proceedings of ICT.fda'06 Hanoi May 20-21.20O6 HOi thao ICT.rda°06 Optimization to reduce processing costs and to overcome the heterogeneity problem [5] In that context, a number of recent work has come up with advanced reasoning techniques for query evaluation and rewriting in which the global schema and queries are expressed in some form of Description Logics [I], (for example, TAMBIS, DWQ, Information Manifold, SIMS, etc.) The critical problem in characterizing the information is that different terms and : constraints used to characterize similar information Motivated by this problem, the key objective of our approach is to represent a general query processing, in which a global schema is used to provide a unified view of local metadata schemas for extracting Icaming object metadata Thus, the mapping between the global and local ontology-based schemas is the main aspect of query processing Interoperation across these ontology-based schemas is achieved by traversing semantic relationships defined between terms across ontologies, and then queries are rewritten in a semantics-preserving [7] manner by replacing them with synonym terms from different local ontologies QUERY ANSWERING BASED ON DESCRIPTION LOGIC KNOWLEDGE BASE Following our proposed ontology-based framework introduced in [7], this section describes an approach to answer queries based on an ontology modeled using Description Logics 3.1 Metadata repositories and ontologybased schema features such as where the object is stored, who created or updated the object, and how it can correctly be used, and structure represents the set of existing associations within or among individuals Learning Object Metadata formalized in different formats are stored in MDB MDB ={MDB,, MDB2, , MDB„}, ne N Where MDB; = , i = l,n, with OMSj is local ontology-based metadata schema of MDBj - OMSi = ,CSi: set of schema elements and CSi: set of relations of elements - And V i = Hii dom(MDBi) = {LoMy, j = l,t} with each LoM, is a metadata item in repository MDBj describing the content, context and structure of a learning object 3.2 Problem motivation In our approach, an ontology-based global schema, called GS, is used for providing a general declarative description of the concepts and their relations in the Elearning domain [5] The GS = , CQ: set of ontology concepts and RQ: set of ontology relations Since Protege OWL plugin provides a comfortable editor for displaying Description Logic concept and role expressions in standard Description Logic syntax, we use the Protege editor and its OWL plug-in to describe the example V H UndttgraduM* * - » m ^%m\ftm^ \! • Pregrvm In the context of Semantic Web, metadata can be represented in such a way that contains not only information, but also the semantics of the structures in order to enable interoperability at the semantic level Any learning objects have three describablc features using metadata: content, context and structure [3] Content is related to what the object contains or is about Context indicates 239 C It- T" m t w t i X Migii I J^ ~fW- I] C