Abstract | ||
---|---|---|
Tagging systems enable users to organise their online entities with distinct tags. Exploiting these user generated content and underlying bilingual information have become more and more important in recommendation system. Probabilistic topic model has been widely used in document management and social network mining. In this paper, we propose a new method to do tag-based recommendation with topic model. Some existing methods are based on mining association rules and similarity measures. In these cases, tags serve as the essential intermediates for statistical computation, but they have the drawbacks that results are sensitive to parameter setup. Even though they are popular in some real application situations, they are simply lack of scalability as the computational procedure differs over distinguished platforms. It's natural to take tags as words, from which topics can be effectively extracted by using topic model. Under the assumption of the generating process in topic model, user's topic distribution parameter implies his or her topic preference. Recommendation results are obtained according to the final probability calculated by summing over topics. Our experiments show that the proposed model is effective to do both tags and items recommendation on two sparse datasets. |
Year | DOI | Venue |
---|---|---|
2013 | 10.1109/IRI.2013.6642450 | IRI |
Keywords | Field | DocType |
tag-based top-n recommendation system,social network mining,document management,bilingual information,statistical computation,statistical analysis,recommender systems,topic preference,online entities,user topic distribution parameter,content management,item recommendation,user generated content,similarity measures,pairwise topic model,association rule mining,data mining,sparse datasets,social networking (online),tagging systems,probability,probabilistic topic model | Data mining,Computer science,Document management system,Artificial intelligence,Content management,Probabilistic logic,Recommender system,Pairwise comparison,Information retrieval,Association rule learning,Topic model,Machine learning,Scalability | Conference |
Volume | Issue | Citations |
null | null | 0 |
PageRank | References | Authors |
0.34 | 15 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Zhengyang Li | 1 | 0 | 1.01 |
Congfu Xu | 2 | 131 | 15.71 |