The Sentiment Analysis of Unstructured Social Network Data Using the Extended Ontology SentiWordNet

Abstract

In this paper the features of semantic and sentiment analysis of textual data of social networks are presented, and an original model and algorithm for sentiment analysis of textual fragments of social networks using fuzzy linguistic ontology are proposed. This approach involves the use of various subgraphs of fuzzy ontology when considering texts of various subject areas with regard to contexts. In addition, the algorithm involves the assessment of the sentiment scores of individual syntagmatic structures into which the analyzed text fragments are divided. It also presents the results of experiments comparing the efficiency of the developed algorithm with a group of existing approaches in analyzing text fragments on the example of data from the social network VKontakte.