Abstract
This paper examines the development of the Chinese Sentiment Lexicon for Internet (CSLI), a sentiment lexicon for capturing the valence and arousal in Chinese online social media texts. We first review the current sentiment lexicons and their building process, including the collection of words, judging the emotionality of words, and testing reliability and validity. In Study 1, we develop CSLI and test its initial reliability and validity. In Study 2, we further test the convergent validity of CSLI by examining its correlations with human judgment in 429 aggregated Weibo comments. In Study 3, the predictive validity of CSLI is examined by linking its results to personality traits among 52 undergraduates. Two replication studies are also conducted to verify the findings in Study 2 and 3. The results have generally supported the reliability and validity of CSLI. Therefore, CSLI can be used as a research tool to capture the degree of valence and arousal in Chinese online social media texts. Its potential to promote human well-being is also discussed.