Using microblogs for crowdsourcing and public opinion mining
Akcora, Cuneyt Gurcan
MetadataShow full item record
With the advance of social networking sites, citizen publishing has taken the lead in providing a large part of information we daily consume. In this study, we present data mining and crowdsourcing strategies that can efficiently mine, classify and use this noisy information. In the RainRadar project, active mining that requires user contribution upon querying is presented for weather forecasts. We crowdsource the weather forecast for cities in USA and Canada, and show the results on a web site. In the Upinion project, we use Twitter user posts and employ sentiment analysis to detect changes in public opinion. We created a system that can map emotions to eight classes, and built a corpus to populate those classes.