ESTRATEGIAS BASADAS EN APRENDIZAJE COMPUTACIONAL PARA DETECTAR REPRESIÓN EN USUARIOS DE REDES SOCIALES.

FARÍAS ANZALDÚA, ALÁN ALÉXIS (2018) ESTRATEGIAS BASADAS EN APRENDIZAJE COMPUTACIONAL PARA DETECTAR REPRESIÓN EN USUARIOS DE REDES SOCIALES. Maestría thesis, Universidad Autónoma de Chihuahua.

[img]
Vista Previa
Text
Tesis - Estrategias Basadas en Aprendizaje Computacional para Detectar Depresión en Usuarios de Redes Sociales.pdf

Download (1101Kb) | Vista Previa

Resumen

The goal of this work is to analyze publications found on social networks looking for textual and temporal characteristics that allow discerning users with or without depression. For these means, an experimental exploration of textual and temporal characteristics was perform in order to find the best parameter combination. A method combining both kinds of characteristics was obtained as a result, which processes users in two steps. First step treats publications individually, whereas second step performs a meta-analysis of said publications. This method was utilized as a proposal for participation in eRisk forum in CLEF 2017, in Dublin, Ireland. Specifically, participation occurred in “Pilot Task: Early Detection of Depression”, which distributed the “A Test Collection for Research on Depression and Language Use” dataset, which was utilized for the experiments and development of the proposed method. This document details the methodology proposed for CLEF, as well as the expedition that lead to the creation of the proposal. It also contains the information acquired along the way, in hopes of being of utility for those seeking for answers to the human psyche in the sea of information known as the internet.

Tipo de Documento: Tesis (Maestría)
Palabras Clave: Machine Learning, Natural Language Processing, Social Networks, Depression
Clasificación temática: Q Science > QA Mathematics > QA76 Computer software
Usuario Remitente: Admin Administrador del Respositorio
Depositado: 13 Ago 2020 16:34
Ultima Modificación: 13 Ago 2020 16:34
URI: http://repositorio.uach.mx/id/eprint/294

Actions (login required)

Ver Objeto Ver Objeto

Universidad Autonoma de Chihuahua