Learning to Recommend Questions Based on User Ratings Ke Sun, Yunbo Cao, Xinying Song, Young-In Song, Xiaolong Wang and Chin-Yew Lin. In Proceeding of the 18th ACM Conference on Information and Knowledge Management (Hong Kong, China, November , 2009). Prepared and Presented by Baichuan Li
Outline Introduction Problem Statement Algorithms Experiments Conclusion 9/4/2015 Paper Presentation 2/21
Introduction Community-Based Question-Answering (CQA) Services 9/4/2015 Paper Presentation 3/21
Finding Answers 9/4/2015 Paper Presentation Query Existed similar questions and their answers 4/21
Finding Questions 9/4/2015 Paper Presentation Sort by popularity 5/21
PROBLEM STATEMENT 9/4/2015 Paper Presentation 6/21
Recommendation 9/4/2015Paper Presentation 7/21
Preference Order 9/4/2015 Paper Presentation 8/21
Ordered Pairs 9/4/2015 Paper Presentation 9/21
Ranking Function 9/4/2015 Paper Presentation 10/21
Principle 9/4/2015 Paper Presentation 11/21
ALGORITHMS 9/4/2015Paper Presentation 12/21
The Perceptron Algorithm for Preference Learning (PAPL) 9/4/2015 Paper Presentation 13/21
The Majority-Based Perceptron Algorithm (MBPA) 9/4/2015 Paper Presentation 14/21
EXPERIMENTS 9/4/2015Paper Presentation 15/21
Dataset 297,919 questions under ‘travel’ category at Yahoo! Answers 9/4/2015 Paper Presentation 16/21
Dataset (Cont.) 9/4/2015 Paper Presentation 17/21
Dataset (Cont.) 9/4/2015 Paper Presentation 18/21
Results Evaluation Measure ◦ Error rate of preference pairs Result 9/4/2015 Paper Presentation 19/21
Results (Cont.) 9/4/2015 Paper Presentation 20/21
Conclusion Investigated the problem of learning to recommend questions based on user ratings ◦ Enlarged the size of available training data through adding questions without user rating ◦ Demonstrated the approach’s effectiveness through intensive experiments Q&A 9/4/2015 Paper Presentation 21/21