Sentiment Lexicon for Chinese College Students to Build and Apply

Reviews from social media are considered as a significant information resource, which is useful for analyzing college students' sentiment and views. Students are eager to express and share their views on web regarding day-to-day activities. That information can be used to understand Chinese college students and their preference. Our research is meaningful in analyzing the psychological processes of students. In order to extract the fundamental student' associated sentiments from those reviews of plain texts, sentiment analysis has emerged and is regarded as a promising technology. In this regard, this paper proposes a novel lexicon model used in college students' domain to exploit semantic relationships between words in natural language text. The sentiment words are initially extracted through lexicon to describe the orientation and the polarity of the attitudes (positive, neutral or negative). Finally, sentiment strength of opinion is calculated and cause event is identified according to the finegrained lexicon. Comparison experimental results on real data from QQ speaking data validate the effectiveness and feasibility of the proposed lexicon model, providing high accuracy levels and low false positive rates. The described lexicon model gives the valuable knowledge to college manager who should detect the students' opinion trends.