Text Mining and Others

Opinion Mining

With the growing availability of opinion-rich online resources such as product reviews, social issue forums and personal blogs, automatic opinion analysis has surged as a hot research field. This research can be generalized as who (opinion holder) comments on what (opinion target) with what feelings (opinion polarity) for what (arguments) [where and when]. Our research focuses on techniques of fine-grained opinion mining and opinion polling and their application.

Patent Mining

Currently, we focus on three applications in patent text processing: patent translation, patent mining, and cross-lingual patent retrieval.

1) Patent translation: The purpose of patent translation research is to translate patent text automatically. Nowadays, we focus on the following research aspects: technological terms collecting and translation, long sentences analysis and translation.

2) Paten mining: The purpose of the patent mining is to create technical trend maps from a set of research papers and patents. It has two subtasks; one is research papers classification, which is requested to classify research papers written in either Chinese or English in terms of the International Patent Classification (IPC) system. The other is technical trend map generation, which is requested to extract expressions of element technologies and their effects from research papers and patents.

3) Cross-lingual patent retrieval: The purpose of patent retrieval is for technology survey and invalidity search, which are two important tasks for someone to overview the state-of-the-art of the target technology or some relative technology.

Others

We are also focusing on document clustering, document classification, text topic analysis, document summarization, ontology learning and active learning for data annotation.