Web-derived Emotional Word Detection in social media using Latent Semantic information
Sentiment (or opinionated) lexicon plays an important part in sentiment analysis (or opinion mining) tasks. With the popularity of Social Network platforms such as tweeter and microblog, net neologisms welcome the spring of development and popularity. With the blooming of net neologisms, it is quite necessary to identify new sentiment words and improve current sentiment lexicon. This paper proposes a pattern based bootstrapping method which extracts sentiment words from microblogs. Consider that frequently used sentiment words always appear after degree adverbs, we enhance the importance of degree adverbs in patterns. The experimental results validate the effectiveness of our method and large quantity of unrecorded sentiment words are extracted with reasonable precision.