PublicRelay Case Study - Toyota and CES
Tech Talk by Bill Mitchell ⚫ Feb. 9, 2017
Our colleague, Bill Mitchell (Vice President, Technology at PublicRelay), presents a case study on identifying topics and relevance for a large data set of tweets. In the case study, a team of MIT students analyzed 241K tweets relating to Toyota hybrid vehicles. This team performed topic analysis to identify relevant tags; and, reach analysis to predict the value of specific tweets. the key results from this exercise had some interesting results. Tweets are a terse medium that requires significant cleanup of the raw text. However, too much cleansing can remove important contextual information. In addition, the algorithms used to model topics and extract tags can be effective in reducing the need for manual tagging and identifying implicit topics. The conclusion is that both automated and human-assisted machine learning are important elements in content analysis.