Extracting Features for Opinion Mining using Pointwise Mutual Information
Abstract
Opinion mining has been developing rapidly in the past years mainly because of its huge set of applications and the scientific barriers it poses. The opinions of tourists are useful for the stakekolders and for public when making some decisions. Opinion Mining is a way to extract data through search engines, social networks and web blogs. The essential part in opinion mining is to analyze and retrierve the feedback of tourists to disover their opinions on different hotels. Now users can actively use IT to search others opinions based on tourism. This study proposes point wise mutual information for summarizing and extracting opinions denoted by tourists in tourism associated internet platforms. The pointwise mutual information is used to decide the semantic orientation of opinions by providing queries to a search engine where poor and excellent words have been considered as a boundary for negative and positive inference words. In this study the PointWise Mutual Information has been used to identify the adjectives and noun phrases related with the target. Retrieving opinions from user generated reviews regarding the perspectives specific to hotel services are helpful for both the clients who are viewing for accomodation and also hotels attempting to develop their services. The proposed system retrieves the hotel reviews from online and categorize them using opinion mining techique. This study explains the importance of opinion mining using pointwise mutual information in the growth of tourism. This study focuses mainly on the opinions of travel behavior of tourists. The datasets collect reviews from the following cities namely Chicago, Beijing, New York, Dubai, San Francisco, London, New Delhi, Las Vegas and Shanghai.