CAR-RECOMMENDATION-SYSTEM-1-1 (2)

CAR RECOMMENDATIO N SYSTEM Analysis Duong Anh Quang Member Obijiaku Calisstus Chisom Nguyen Thi Phuong Hang MOTIVATION AND PURPOSE REPERTORY GRID Content ANALYSIS FUZZY COGNITIVE MAP Motivation and purpose • The number of vehicles in global market is increased • Many cars with similar features get into market  People will be confused on what to choose We built a car recommend system based on the elements and constructs Repertory grid Car Friend Youtube company recomme Personal Showroo videos drive.com nada.com telegraph edmunds website ndation taste m au co.uk com Detailed Surfaced Explanation 6 Explanation Used by most Used by few people 7 people Reliable Info Unreliable Source 4 info source Widely Not Widely Accepted 6 Accepted Satisfactory Not conclusion 6 Satisfactory Easy to find Hard to find Information 2 4 information Small Big Database 3 Database Not easily Easy Approach 1 1 1 Approach Able to Not able to recommend 5 5 recommend Feedback No Feedback functionality functionality Element Analysis (Pearson Correlation) youtube drive.com nada.com telegraph.co.u k edmunds.com car company website friend personal taste showroom 9 0.26 -0.02 0.38 -0.46 0.31 -0.28 -0.30 0.44 0.12 -0.12 0.54 -0.24 -0.04 -0.4 0.30 0.70 0.71 -0.19 0.02 -0.16 0.63 -0.1 0.65 0.16 -0.7 0.55 0.03 0.02 -0.32 -0.7 -0.02 0.21 -0.41 -0.32 Element Analysis (Distance Euclidean) youtube drive.com nada.com telegraph.co.u k edmunds.com car company website friend personal taste showroom 9 8.0 9.27 7.14 6.56 7.42 8.89 11.75 9.95 6.32 7.81 7.55 5.57 9.11 10.0 9.54 6.86 4.36 4.36 8.77 9.59 8.54 4.69 8.25 4.69 8.77 10.1 5.10 7.48 9.22 8.6 10.10 9.54 7.62 8.43 9.06 10.63 Contruct Analysis (Pearson Correlation) 0.44 10 Detailed Explaintion -0.05 0.04 0.44 0.40 -0.14 -0.09 0.05 Used by most people -0.48 0.54 0.25 0.20 0.11 0.05 0.15 0.40 Reliable Source 0.1 0.27 -0.36 -0.58 0.15 -0.61 -0.6 Widely Accepted 0.43 -0.17 -0.41 -0.16 -0.03 -0.21 Satisfactory conclusion 0.62 -0.02 0.05 -0.17 0.18 Easy to find Information 0.61 -0.01 0.31 0.69 Big Database 0.39 0.16 0.94 Easy Approach -0.81 0.42 Able to recommend 0.13 Fuzzy Cognitive Map User Cause/Effec Numbe Technolog t r y Service User Number 0 -0.2 Technology 0.3 0.7 Service 0.8 0 UX 0.6 0 UI 0.1 0.1 Car data 0.1 0 User data 0 0.1 Concept Value initializati on 0.3 0.4 0.6 UX -0.3 0.3 0.9 0.15 0.2 0.1 UI 0.2 0 0 0.7 0.1 Car data User data 0.8 0.4 0.5 0.2 0.2 0 0 0 0.6 0.3 FCM Diagram Technology UX Service User Number UI User data Car Data Formular • A t is the value of concept at timestep t • W is the matrix of casual relationship • K j is the weight of previous concept value at time t-1 • A t-1 is the value of concepts at timestep t-1 • K j = 0.5 Sigmoid Implementatio n • The code used for calculate the concepts value at different stages Concepts Value Step 1 User Number 0.3 0.776 0.835 0.845 0.847 0.848 Technolo gy 0.4 0.55 0.568 0.571 0.571 0.571 Service 0.6 0.636 0.662 0.669 0.67 0.67 UX 0.7 0.747 0.76 0.767 0.769 0.77 UI 0.1 0.532 0.593 0.601 0.602 0.602 Car data User data 0.6 0.3 0.613 0.701 0.629 0.821 0.632 0.838 0.633 0.841 0.633 0.842 => User number and user data is the most important concept in the car recommendation system