Social Media Clustering Based on Semantic Contents and Entities

Overview

Topic status: We're looking for students to study this topic.

Every day a large amount of text is contributed to the Web via social media, including Twitter, Facebook and blogs. From these social media, important information can be revealed and utilized. For governments, early warnings can be set off for disaster expansion; for news media, important events can be detected at early stage before officially released to the public; for commercial corporate, public interests can be identified so that advertisements can be put on more effectively. For all these, understanding the content of social media text is essential. However, currently obtaining such an understanding suffers from some severe problems: social media is usually very short text and contains limited information; informal language is usually used in social media and makes text analysis difficult; and many duplicate social media contains the same content. Because of these drawbacks, traditional text analysis techniques cannot work well and require urgent enhancement in order to meet the requirements of modern social media analysis. This proposed project aims to develop an alternative text clustering methodology by considering both semantic contents and entities. Semantic contents are discovered from the social media text by using text-mining techniques, e.g. pattern mining and association rules mining. Entities are extracted from social media via natural language techniques, e.g. part-of-speech detection and subject cross-reference identification. The social media text will be clustered based on both semantic contents and entities. Better understanding of social media is expected to be obtained from analysing the clustering results. This research will promote the traditional text clustering methods and have potential contributions to smart information use by news media and commercial corporate. Hardworking, self-motivated, capable of working under stress and willing to take challenges, applicants from Information Technology, Information Sciences and Computer Science backgrounds, with Distinction-level academic results and/or research experience, are expected to join us in this project. Any enquiries are welcome.

This project is part of the Smart Services CRC and has the option to apply for scholarship.

Study level
PhD
Supervisors
QUT
Organisational unit

Science and Engineering Faculty

Research area

Information Systems

Contact

Please contact the supervisor.