About Lun-Wei Ku's Previous and Current Research on Opinion Analysis
Opinion Analysis is a popular research topic and has drawn much attention because of its applicability. Many studies have been conducted on this topic, identifying variables that influence the results of determining opinion scores, opinion polarities, and text rating. Previous research reveals performance of adopting different algorithms on various datasets. Specifically, most studies agree that this task has different characteristics from other tasks of traditional natural language processing. Therefore, ways to process experimental materials, generate answers and evaluate the performance of algorithms are also different.
Opinion analysis is not only a research topic for computer science. As opinions are expressed by people, the techniques of analyzing opinions can be applied to social sciences and humanities, such as political science, marketing, web ethics, advertisement, law, etc., and are even helpful for government and national security.
Although these previously mentioned variables have been explored, little was directly investigated or discussed concerning the personal profile, the web behavior profile, and the real life environment. Furthermore, how to group people according to their opinions instead of interests, in order to provide better recommendations, has also not been investigated. It could be a progressive direction and a new aspect in building social networks.
In the previous related work, many experimental materials have been built and can be reused. In addition, several opinion dictionaries are available. However, collecting real users¡¦ profile and their web behaviors are still indispensable. Moreover, gathering more real data, which is also one of the project tasks, could benefit this project and future related research...
(More)
