High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

A Survey on Topic Modelling in Text Mining

Author(s):

A.SARANYA , K.S.R.COLLEGE OF ENGINEERING; A.SARANYA, K.S.R.COLLEGE OF ENGINEERING; V.VENNILA, K.S.R.COLLEGE OF ENGINEERING; S.SUGANYA, K.S.R.COLLEGE OF ENGINEERING; G.S.RIZWANA BANU, K.S.R.COLLEGE OF ENGINEERING

Keywords:

Data Mining, Text Mining, Topic Modeling, Document Relevance, Information Filtering, Latent Dirichlet Allocation, Supervised Learning

Abstract

Text Mining has become an essential research area. Text Mining is the innovation by computer of new, earlier unknown information, by spontaneously extracting information from dissimilar written resources. In this paper, a Survey of TextMining technique for Selective supervised Latent Dirichlet Allocation have been presented. Although ssLDA receives the universal framework of sLDA where many forms of response (such as real and categorical responses) variables can be shown, we focus on the case where the response variable is definite in this paper. In this paper, according to the following two considerations: First, most of other topic models, ssLDA views documents as discrete data which consist of word counts, while documents are treated as directional data in STM. Second, the topics in ssLDA and STM are generated from Dirichlet and vMF distributions, respectively. Due to Dirichlet distributions, ssLDA can set a topic assignment (topic indicator) and adjust the weight for each individual word that STM fails to do. This paper explores existing research highlights and provides various needs of significant research in these topics.

Other Details

Paper ID: IJSRDV3I70266
Published in: Volume : 3, Issue : 7
Publication Date: 01/10/2015
Page(s): 774-777

Article Preview

Download Article