Sentiment analysis from travellers’ reviews using enhanced conjunction rule based approach for feature-specific evaluation of hotels

Abstract

The evolution of the internet has steered an enormous amount of travel reviews of hotels on the web. People referring to these reviews are often overloaded and confused by the sheer amount of information available. Sentiment analysis techniques have been successful in aggregating the reviews, extracting their sentiments and thereby minimizing the information overload. But lacking in specific feature-based sentiment analysis has restricted customers in getting the actual scenario of hotels entirely. This paper presents a prospective design on lexicon-based approach for feature-based sentiment analysis of travel reviews on hotels or resorts. In particular, an enhanced form of conjuncture-based approach is proposed to segregate sentences into relevant clauses, identifying the feature and the sentiment value associated with it. Overall sentiment score for features like food, service, and location of a hotel is being calculated. The experiment results show significantly better accuracy and precision than the conventional text segregation and sentiment analysis methods, namely trigram and conjunction rule based approach.

Subject Classification:

• 68
• 68T09
• 68T50

Keywords:

• Sentiment analysis
• Opinion mining
• Online review
• Aspect-based opinion mining
• Aspect-based sentiment analysis
• Machine learning
• Sentiment orientation
• Tourism reviews
• Lexicon