Lexalytics, Inc., a software and services company specializing in text and sentiment analysis, announced the availability of enhanced reporting on the conversations occurring around, about, and between different accounts on Twitter based on the sentiment analysis of commonly used emoticons and acronyms.
With the use of emoticons, abbreviations, and confusing "social speak" grammar, micro-blog services such as Twitter present a difficult task for natural language processing systems. For acronyms, Lexalytics parsed thousands of tweets to get to hundreds of common acronyms and emoticons. The team then made decisions on whether each acronym (such as LOL for Laugh Out Loud) was sentiment-bearing, needed to be expanded, or should be treated as simply an interjection.
With emoticons, Lexalytics found that some are obviously positive (such as :D) or negative (:<) while others are considered more neutral. For the @ sign, Salience part-of-speech tags the @ tagged string as a "MENTION" which can be used for further reporting. In particular, @ tagged strings will return as people entities, with the associated sentiment, themes, etc.
Additionally, # sign (hashtags) are part-of-speech tagged as @hashtag. These do not report back as any sort of entity type. Hashtags are typically used as a lightweight "tag" for the content of the tweet. This information can be used by Salience for further processing as a tag.