Topic detection with recursive consensus clustering and semantic enrichment

Abstract Extracting meaningful information from short texts like tweets has proved to be a challenging task. Literature on topic detection focuses mostly methods that try guess the plausible words describe topics whose number been decided in advance. Topics change according initial setup of algorithms and show consistent instability with moving one another one. In this paper we propose an iterative procedure for searches most stable solutions terms describing topic. We use based clustering consensus matrix, traditional detection, find both set optimal topics. observe however several cases does not converge unique value but oscillates. further enhance methodology using semantic enrichment via Word Embedding aim reducing noise improving separation. foresee application techniques automatic discovery noisy channels such as Twitter or social media.