Survey Paper on Pattern-Enhanced Topic Model for Data Filtering

Abstract

The machine learning & text mining area topic modeling has been extensively accepted etc. To generate statistical model to classify various topics in a collection of documents topic modelling was proposed. A elementary presumption for those approaches is that the documents in the collection are all about one topic. To represent number of topics in a collection of documents, Latent Dirichlet Allocation (LDA) topic modelling technique was proposed, it is also used in the fields of information retrieval. But its effectiveness in information filtering has not been well evaluated. Patterns are usually thought to be more discriminating than single terms for demonstrating documents. To discovered pattern become crucial when selection of the most representative and discriminating patterns from the huge amount. To overcome limitations and problems, a new information model approach is proposed. Proposed model includes user information important to generate in terms of various topics where each topic is represented by patterns. Patterns are generated from topic models and are organized in terms of their statistical and taxonomic features and the most discriminating and representative patterns are proposed to estimate the document relevant to the user?s information needs in order to filter out irrelevant documents. To access the propose model TREC data collection and Reuters Corpus vol. 1 are used for performance