Domain Specific Sentiment Dictionary for Opinion Mining of Vietnamese Text.

Knowing public opinions from subjective text messages vastly available on the Web is very useful for many different purposes. Technically, extracting efficiently and accurately the opinions from a huge amount of unstructured text messages is challenging. For English language, a common approach to this problem is using sentiment dictionaries. However, building a sentiment dictionary for less popular languages, such as Vietnamese, is difficult and time consuming. This paper proposes an approach to mining public opinions from Vietnamese text using a domain specific sentiment dictionary in order to improve the accuracy. The sentiment dictionary is built incrementally using statistical methods for a specific domain. The efficiency of the approach is demonstrated through an application which is built to extract public opinions on online products and services. Even though this approach is designed initially for Vietnamese text, we believe that it is also applicable to other languages.