Honours’ project topics (2017-2018)
Topic 1. Extracting and visualising the key information of articles for assisting fast reading
In many occasions, people only need to do speed reading, i.e., understanding the main ideas of an article rather than all the details. This project aims to develop a prototype system which can assist fast reading by extracting and highlighting the text which describes the main points (or gist) of the article, so that readers can easily figure out which sentences they should read and which ones they can skip. The outcome of this project will be two-fold: methodologies for automatically extracting key information from text and novel visualisation strategies. To make the system more useful, when the reader clicks a sentence, the system will highlight the correlative sentences, permitting readers to quickly gain insights of the specific details that they are interested in. Some of the technologies required for this project include: text summarisation, key sentence extraction, clustering (e.g., LDA).
Topic2. Understand your friends from social media
This project aims to develop models for user profiling, which can help one understand their friends in terms of their characteristics based on the information they post on social media sites (e.g., Facebook, Twitter, Weibo, etc.). For privacy protection, you need to anonymise the personal information (e.g. name, DoB) in the data. This project is challenging. You will develop analytic models based on sentiment analysis and other natural language processing techniques, which can help find answers to the following questions:
— Who is usually in a good mood and who is not?
— Who change their mood patterns regularly?
— What are major factors that affect your friends’ mood? E.g., affected by friends, study, family or something else?
Finally, you will need to design how to evaluate the performance of your model.
Note: The above examples are only indicative of the kind of work I am interested in. Final project allocation will be based on discussion with students.