A Survey on Documentation Modelling in Information Filtering using Pattern Recognition

In the field of information filtering many full-fledged term-based or pattern-based approaches have been used by assuming the documents in a collection are all about one topic. However, user’s interest can be varied and generally multiple topics are involved in the collection of documents. Topic modeling, such as Latent Dirichlet allocation (LDA), has been widely exploited in the fields of machine learning and information retrieval etc. but its effectiveness in information filtering has not been so well explored. A novel information filtering model, Maximum matched Pattern-based Topic Model (MPBTM), is proposed. Following features makes the proposed model peculiar: (1) In terms of multiple topics the user information need are generated and each topic is represented by patterns; (2)the most discriminate and representative patterns, called Maximum Matched Patterns are generated from topic models and are systematized in terms of their statistical and taxonomic features and proposed to estimate the documents relevance in order to filter out irrelevant documents suited to user to user’s information needs.