Sentiment analysis using sentence minimization with natural language generation (NLG)

Abstract

The analysis of feeling is used to define the attitude of a writer in relation to a subject or the appropriate global polarity of a document. The proposed work is to provide a platform in order to visualize the relative analysis of feedback for some particular product. In doing so, instead of the basic truthful information, the analysis will be done based on comments and comments developed from various sources. In this approach, the analysis of feeling at the document level will be carried out taking into account all aspects in the same way using natural language processing techniques. The present unsupervised method is used for sentence minimization that relies on a Stanford-type dependency for extracting information elements and compressed sentences are generated via a Natural language generation engine (NLG). An automatic evaluation of the same is done and F-scores of about 87.51 is achieved.