Ontology Driven Sentiment Analysis on Social Web for Government Intelligence

The proposed tool SentIndiGov--O proffers a sentiment analysis application model based on ontologies, where the ontology IndiGov-O is created for "Ministries of Government of India" to reinforce the feature selection task in a Naïve Bayesian classification paradigm.

Abstract

The World Wide Web as a global information medium has evolved radically from the first 'read only web' generation to the next participative- collaborative 'read and write' social web to the intelligent generation of web referred to as Semantic Web, the 'Read Write and Execute Web'. As a novel facet, digital governance intends to ensure and assure that stakeholders (government and citizens) have greater access and control over the governance mechanism which leads to a more transparent, accountable and efficient governance. Social media can be used to facilitate interaction between people and can offer a substantial, unparallel platform for extensive involvement of citizens in the governance measures. We propose using an ontology-based analytics on social media that expounds an intelligent governance model where sentiment can be mined for extracting views of citizens towards government practices, policies, rules and monitoring performance. The idea is to deploy ontology based techniques to determine the subjects/topics discussed in the tweets, which are related to Ministries of India: their policies, rules, schemes, and analyzing the sentiment (positive, negative and neutral) for tweets retrieved (based on the concepts defined in Ontology). The proposed tool SentIndiGov--O proffers a sentiment analysis application model based on ontologies, where the ontology IndiGov-O is created for "Ministries of Government of India" to reinforce the feature selection task in a Naïve Bayesian classification paradigm. The preliminary results with performance accuracy of ~77% are encouraging and foster the need to exploit hybrid of concept based techniques and machine learning for sentiment analysis.