Using Intra-Profile Information for Author Profiling Notebook for PAN at CLEF 2014

摘要: In this paper we describe the participation of Laboratory Lan- guage Technologies INAOE at PAN 2014. We address Author Profiling (AP) task finding and exploiting relationships among terms, documents, profiles subprofiles. Our approach uses idea second order attributes (a low- dimensional dense document representation) (4), but goes beyond incorpo- rating information each target profile. The proposed representation deepen analysis incorporating texts in same profile, is, focus For this, automatically find subprofiles build docu- ment vectors that represent more detailed documents subpro- files. compare with standard Bag-of-Terms best method PAN13 using 2014 corpora for AP task. Results show evidence usefulness intra-profile to determine gender age profiles. According official results, was one three approaches most social media domains. Particularly, it achieved performance predicting blogs tweets English.