Clustering Cantonese Words with Similar Meanings

Abstract

The technique of finding similar words is widely used in different areas, including biomedical informatics, geoinformatics, academic, and computational linguistics etc. A system will be designed, constructed and tested to cluster words with similar meanings based on user's input, which can be one or more than one Chinese keyword(s). A plot can also be generated to describe the relationship of the input and output. This tool should be able to improve the user searching experience with providing a certain of the accuracy of the result and respond to the user within an acceptable time. The system can support incremental updates. That is, user can update and modify the dictionary in the system. Also, the system should be able to recognize new Cantonese words and trains the model in the future when there is a new coming corpus. A responsive website, developed in python, is provided to the user to access the system. User can make a query, upload corpus and train the model via the website.