DFKI-LT - Factuality Drift Assessment by Lexical Markers in Resolved Rumors
Factuality Drift Assessment by Lexical Markers in Resolved Rumors
1 Proceedings of the 1st Int. Workshop on Semantic Change & Evolving Semantics, CEURS, 12/2016
Our study presents a social media content representation, visualization and assessment method for modeling the emer- gence and resolution of rumorous claims during crisis events. Interpreting the factuality with which a claim is expressed is typically context-dependent and can be subject to semantic drift. We identify and quantify temporally anchored distri- butional and typological patterns of lexical cues present in microposts in order to pinpoint the factuality trend in rumor timelines. The ndings on our English dataset are discussed with respect to provenance modeling for the social media do- main, and we also show how to port the method to content in German.
Files: BibTeX, crc_success16_submission_final.pdf