Abstract
Sentiment is a combination of word choice, tone of voice and writing style, which allows the same news to be described as either positive or negative. With the assistance of modern machine learning technology, Natural Language Processing (NLP) is the most effective method for sentiment analysis from natural language. Using NLP, positiveness of a news can be analyzed and provide an overview of tone and manner used. Using Sentiment Intensity Analyzer (SIA) of Python’s Natural Language Toolkit to evaluate newspaper headlines and contents, the positiveness of the news can be uncovered. While the initial accuracy of the analyzer with default language database is approximately 80%, the accuracy of the sentiment analysis can be enhanced by supplying more training data, which are the news. By cross-referencing the view count of the news, it greatly facilitates journalists to learn about the likes and dislikes of readers. It is also particularly useful for webmasters of news portal to arrange personalized news feed for each reader by rearranging the news layout and display order according to their sentiment analysis results. While the initial work is limited to English news due to limitations of SIA, it is hoped that the system can be extended to other languages in future and offer opportunities for further work on reader preference analysis.
Digital Media
This presenter hasn’t added media.
Request media and follow this presentation.
