Opinion mining using modified ant colony optimization

Abstract

With the technological developments in the fields of Natural Language Processing (NLP) and Opinion Mining (OM), many real-time applications are concentrating on analyzing the opinions of the people. The opinions or reviews given by the people through the internet are collected for summarization or classification based on the need. OM obtains the feelings of people in positive or negative Comm ents. OM is very useful to track the opinions of people about a particular product, topic or service. It also has some difficulties. The opinionated word considered positive in one condition may be negative in another case. Secondly, all the individuals are not conveying their opinions in similar way. The important process of OM are feature extraction, feature selection and classification. Feature extraction is the initial task to retrieve the attributes from the opinionated document. Most popular techniques used for feature extraction are Term Frequency (TF), Inverse Document Frequency (IDF), Term Frequency-Inverse Document Frequency (TF-IDF), Opinion Words, Negations and Syntactic dependency. In this research work, TF-IDF is used for feature extraction. The feature extraction produces a large number of features and it leads to high computational task. To overcome this problem, feature selection techniques are used to reduce the number of features. The feature selection typically saves the operating time, eliminates irrelevant features and redundancy. Feature selection is NP Hard and therefore statistical techniques have been used to produce sub optimal solutions. newline