Filtering and ranking for social media monitoring

Abstract

Social media monitoring is fast becoming a staple activity for public relations and communications staff, who have a growing mandate to track mentions of organisational entities, projects, or products in social media. However, this task is not trivial because: 1) the mentions may be found across a variety of social media and 2) the keywords used for detecting mentions are often ambiguous. We address here classification and ranking of search results collected from a range of social media. Experiments with a labelled corpus demonstrate system effectiveness in filtering out non-relevant results and gains in efficiency by improving result ranking, in both regards outperforming commonly-used alternatives. Show more