Chinese
English
Assistant Research Fellow  |  Ku, Lun-Wei  
 
contact
vita
education
experience
interests
descriptions
activities
invited_talk
honors
publications
patents
others
supervised
lab (New window)
 
 
 
 
 
Research Descriptions
 

My research interests include natural language processing (NLP), computational linguistics, information retrieval, and in particular, sentiment analysis and opinion mining. My approaches focus on how to utilize linguistic rules, common senses and real world knowledge such as syntactic/semantic cues, thesaurus, ontology, and information acquired from explosive web data, to understand the natural languages. I have developed the first Chinese opinion analysis system, CopeOpi. Recently, I am interested in deep understanding, emotions detection and social media text analysis. Related research topics include:

Semantic Role Labeling: We try to enhance the performance of the semantic role labeling in the Sinica Parser. We utilize real world knowledge from E-Hownet together with contextual features on several models and their combinations. Techniques are also applied in sentiment analysis, entailment and question answering. 

Readers’ Interest Analysis: We are interested in predicting readers’ interest after their reading. From different dimensions, including semantic, physical and sentiment aspects, we pursue terms representing interests instead of topics.

Emotion Detection and Representation: We are interested in the emotion visualization of texts. We have analyzed a big quantity of social media posts and their emotion tags. For deep understanding, we develop an automatic emotion pattern extraction approach to detect emotions. Then we use emotion colors defined by psychologists to represent the emotion state of texts.

Emotion Expression Assistance: Solving real world problems with techniques of sentiment analysis is always attractive, and we find an interesting one: Helping ESL (English as a Second Language) learners to express their emotions with precise wording. We capture the contextual semantics in learners’ writing by identifying emotion events. The goal is to suggest appropriate emotion words according to the context. 

Other related NLP research topics I have been working on include opinion mining, medical NLP, and recommendation systems.

 
 
bg