Open AccessCROSS DOMAIN OPINION MINING USING MAXIMUM ENTROPY BASED CLASSIFIER
Author(s) -
V. Manimekalai,
S. Gomathi,
Rohini
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1362/1/012065
Subject(s) - sentiment analysis , computer science , artificial intelligence , natural language processing , principle of maximum entropy , negation , bipartite graph , classifier (uml) , domain (mathematical analysis) , cluster analysis , graph , mathematics , theoretical computer science , mathematical analysis , programming language
The current analysis is play a dominant role in opinion mining is additionally referred to as sentiment analysis because of clear volume of opinion made net resource like discussion type, review sites, blogs, and tweets area unit on the market in digital type. Sentiment analysis is that the field of study that analyzes client opinion, feedback, sentiment analysis, attitudes and feeling from communication. At intervals fraction second, we have a tendency to classify the text in several manner in several seconds. It’s one in all the active analysis areas in linguistic communication process [NLP]. There are a unit range of techniques we want to classify the Opinion reviews. The main problematic in the sentiment analysis is to understand the usage of negation and the taxonomy of positive and negative sentiments recorded by the users in the social group. The proposed approach that extracts and classifies opinion words from one domain called source domain and predicts opinion words of another domain called target domain using a semi-supervised approach, its combines modified maximum entropy and bipartite graph clustering. A comparison of opinionclassification of reviews on four different product domains are presented. The results demonstrate that the proposed method perform relatively well in comparison to the other methods. Comparison of SentiWordNet of domain-specific and domain-independent words reveals that on an average 72.6% and 88.4% words, respectively, are correctly classified.