Journal of Computer Science

Online Forums Hotspot Prediction Based on Sentiment Analysis

K. Nirmala Devi and V. Murali Bhaskarn

DOI : 10.3844/jcssp.2012.1219.1224

Journal of Computer Science

Volume 8, Issue 8

Pages 1219-1224

Abstract

Problem statement: Online forums hotspot prediction is one of the significant research areas in web mining, which can help people make proper decision in daily life. Online forums, news reports and blogs, are containing large volume of public opinion information. Rapid growth of network arouses much attention on public opinion, it is important to analyse the public opinion in time and understands the trends of their opinion correctly. Approach: The sentiment analysis and text mining are important key elements for forecasting the hotspots in online forums. Most of the traditional text mining work on static data sets, while the online hotspot forecasts works on the web information dynamically and timely. The earlier work on text information processing focuses in the factual domain rather than opinion domain. Due to the semi structured or unstructured characteristics of online public opinion, we introduce traditional Vector Space Model (VSM) to express them and then use K-means to perform hotspot detection, then we use J48 classifier to perform hotspot forecast. Results: The experimentation is conducted by Rapid Miner tool and performance of proposed method J48 is compared with other method, such as Naive Bayes. The consistency between K-means and J48 is validated using three metrics. They are accuracy, sensitivity and specificity. Conclusion: The experiment helps to identify that K-means and J48 together to predict forums hotspot. The results that have been obtained using J48 present a noticeable consistency with the results achieved by K-means clustering.

Copyright

© 2012 K. Nirmala Devi and V. Murali Bhaskarn. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.