Social-Media Users can be profiled by their Similarity with other Users

@inproceedings{LpezMonroy2017SocialMediaUC,
  title={Social-Media Users can be profiled by their Similarity with other Users},
  author={Adri{\'a}n Pastor L{\'o}pez-Monroy and Manuel Montes-y-G{\'o}mez and Hugo Jair Escalante and Luis Villase{\~n}or Pineda and Thamar Solorio},
  booktitle={CLEF},
  year={2017}
}
In this paper we describe the system we designed for participating at CLEF-PAN 2017. In this work, we addressed the Author Profiling (AP) task by exploiting the corpus as a knowledge base. The core idea is that profiles can be identified by exposing its relationship with other users. This strategy produces enriched representations, where fine grained user-document relationships are highlighted. For this, we use a non-sparse user-document representation, which captures distributional information… CONTINUE READING