HOW ARE WE DOING? OPINION MINING CUSTOMER SENTIMENT IN US TRANSIT AGENCIES AND AIRLINES VIA TWITTER

摘要

This manuscript examines the unsolicted comments on the social media site Twitter usinga simple opinion mining method. The data are scored using a commonly available,experimentally tested lexicon of words associated with both positive and negativeexperiences. Analysis of the scoring performance finds that even though the analysis usesa simple algorithm, the algorithm performs fairly well in clustering similar services,recognizing non-opinions, and marking both strongly positive and strongly negativeassessments. Among transit agencies, Translink in Vancouver and TriMet in Portlandhave the least negative commentary in their feeds. Alaska Airlines and Southwest werethe airlines with the most positive comments appearing on Twitter.