© 2016, IJARCSMS All Rights Reserved 156 | P age ISSN: 2321-7782 (Online) Impact Factor: 6.047 Volume 4, Issue 11, November 2016 International Journal of Advance Research in Computer Science and Management Studies Research Article / Survey Paper / Case Study Available online at: www.ijarcsms.com Pattern Based Document Recommendation using Maximum Matched Equivalence Classes Smita K. Thakare 1 PG student Department of Computer Engineering, Late G.N.Sapkal college of Engineering Savitribai Phule, Pune University – India Prof. J. V. Shinde 2 Assistant Professor Department of Computer Engineering Late G.N.Sapkal college of Engineering, Savitribai Phule Pune University – India Abstract: Topicmodelling has been widely accepted in the areas of machine learning and text mining, etc. It was proposed to generate statistical models to classify multiple topics in a collection of documents.Existing model I.e. pattern based model, term based model suffered with polysemy and synonymy ,noise generated by this model . All this model only consider that user interested in in only one topic but in situation user are interested in at time many topic in the filled on information filtering Patterns are always thought to be more discriminative than single terms for describing documents. Selection of the most representative and discriminative patterns from the huge amount of discovered patterns becomes essential. To deal with the above mentioned limitations a novel information filtering model is proposed. Proposed model includes user information needs are generated in terms of multiple 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 discriminative and representative patterns are proposed to estimate the document relevance to the user’s information needs in order to filter out irrelevant documents. To evaluate the effectiveness of the proposed model TREC data collection and Reuters Corpus Volume 1 are used. Keywords: Topic model, information filtering, and pattern based model, term based model, maximum matched pattern. I. INTRODUCTION All data mining and text mining techniques assume that the user’s interest is only related to a single topic. Actually, this is not necessary in the case. When a user asks for information about a product like “CAR”, the user not able to typically mean to find documents which consistently mention t he word “CAR”. The user probably wants to find documents that contain information about different aspects of the product, such as location, price, and servicing. This means that a user’s interest usually involves multiple aspects relating to multiple topics. The most inspiring contribution of topic modeling is that it automatically classifies documents in the collection by a no. of topic which represent every document with multiple topics and their corresponding distribution. When we are comparing with pattern-based model and word-based model, pattern-based model generate most meaningful and useful content as per the use requirement. But some time pattern are small in size or large in size and that pattern is not carry the meaning related to the particular topic.so to avoid this Problem related to pattern we have to find out The topic-based representation generated by using topic modeling can conquer the problem of semantic confusion compared with the traditional text mining techniques. Topic modeling needs improved modeling users interests in terms of topics’ interpretations. Hence we proposed the innovate system i.e,A Maximum matched Pattern-based Topic Model which generates pattern enhanced topic representations to model user’s interests across multiple topics. Model selects maximum matched patterns, instead of using all discovered patterns, for estimating the relevance of incoming documents. To find out most meaningful pattern we using ranking method and most ranked pattern is most useful