A method for detecting the profile of an author
| Autor: | Silvia García, Jesús Silva, Rosio Barrios, María Alejandra Binda, Fredy Marin Gonzalez, Bellanit Leon Castro, Ligia Castro |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Training set
Computer science business.industry Social software Gender PAN 2018 020206 networking & telecommunications 02 engineering and technology Supervised Classification computer.software_genre Random forest Age 0202 electrical engineering electronic engineering information engineering General Earth and Planetary Sciences Profiling (information science) 020201 artificial intelligence & image processing Artificial intelligence business computer Competence (human resources) Natural language processing General Environmental Science |
| Zdroj: | ANT/EDI40 Procedia Computer Science REDICUC-Repositorio CUC Corporación Universidad de la Costa instacron:Corporación Universidad de la Costa |
| ISSN: | 1877-0509 |
| DOI: | 10.1016/j.procs.2020.03.101 |
| Popis: | This paper presents a method for detecting an author’s profile using the following two elements: gender and age. This is based on a set of dialogues, written in two languages: English and Spanish, provided for Author Profiling competence within the evaluation forum "Uncovering Plagiarism, Authorship, and Social Software Misuse" (PAN2018). Counts of lexical, semantic, and syntactic characteristics are used to generate a two-phase classification system, which first classifies gender and then age. The results obtained show that, with the amount of data available, it is possible to characterize both the age and gender of an author with an accuracy greater than 50%. However, these values could be improved by having more evidence of information in the training data. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|