Towards sentiment analysis application in housing projects

In becoming a develop nation by 2020, Malaysia Government realized the need in providing affordable house to the public. Since Second Malaysia Plan, government has implemented various affordable housing projects and it continues until recent Malaysia Plan. To measure the effectiveness of the initiatives taken, public opinion is necessary. A social media platform has been seen as the most effective mechanism to get information on people’s thought and feeling towards certain issues. One of the best ways to extract emotions and thoughts from what people post in social media is through Sentiment Analysis (SA). There are three different levels of analysis: document level, sentence level and feature level. Most of previous research focused on the classification of sentiment at document or sentence level. Unfortunately, both document and sentence level does not discover what exactly people like or not. While the analysis based on feature, there exist accuracy problem when classifying the sentiment scores. This paper will propose a new framework that focuses on sentiment classification scores at feature level to overcome the uncertainty and accuracy issues on the result.