Publicación: Clasificación automática de textos sobre Trastornos de Conducta Alimentaria (TCA) obtenidos de Twitter
Cargando...
Fecha
2021-06-25
Autores
Editor/a
Director/a
Tutor/a
Coordinador/a
Prologuista
Revisor/a
Ilustrador/a
Derechos de acceso
Atribución-NoComercial-SinDerivadas 4.0 Internacional
info:eu-repo/semantics/openAccess
info:eu-repo/semantics/openAccess
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Nacional de Educación a Distancia (España). Escuela Técnica Superior de Ingeniería Informática. Departamento de Inteligencia Artificial
Resumen
Gran parte de nuestra sociedad da mucha importancia al aspecto físico. Tener un aspecto físico bonito, o ideal, no tiene por qué ser sinónimo de padecer una buena salud. Hay personas que pueden poseer un cuerpo delgado, fuerte y con una buena definición muscular, pero no mantener una vida muy saludable debido al uso de sustancias que no son precisamente sanas. En la búsqueda de un aspecto físico delgado, muchos piensan que también encontrarán la felicidad y, sin embargo, en muchas ocasiones esto no es así. La presión que imponen los cánones de belleza extendidos en la sociedad han provocado que gran parte de la sociedad padezca enfermedades mentales relacionados con la alimentación y más conocidos como trastornos de conducta alimentaria (TCA). Algunos ejemplos son la anorexia, la bulimia, la bulimarexia, trastornos por atracón, etc. Los medios sociales y, más concretamente, las redes sociales, son herramientas tecnológicas que cada vez tienen un mayor número de usuarios y mediante los cuales es sencillo difundir y divulgar información entre la sociedad de forma rápida. A pesar de las bondades de las redes sociales, cuando una persona padece un TCA, pueden fomentar o provocar un empeoramiento en la propia enfermedad del paciente. Twitter es una de las redes sociales más utilizadas en el ámbito de la recopilación de datos para su posterior estudio y, en el ámbito de la salud, es la herramienta más utilizada en el ámbito de la investigación en lo que a redes sociales se refiere. Son muchos los estudios que están centrándose en obtener tuits y generar modelos capaces de clasificar textos o analizar el sentimiento en distintas temática sanitarias. Por todo lo anteriormente expuesto, en esta investigación se ha propuesto aplicar distintas técnicas de minería de datos recopilando distintos tuits relacionados con TCA. Posteriormente, se han aplicado técnicas de minería de textos y procesamiento de lenguaje natural que han permitido generar modelos predictivos haciendo uso de distintas técnicas de aprendizaje automático supervisado como bosques aleatorios, redes neuronales recurrentes e incluso modelos conocidos como Bidirectional Long Short-Term Memory (Bi-LSTM). Tras la aplicación de estos modelos, se ha puesto en valor la exactitud obtenida de los diferentes modelos de clasificación para clasificar tuits relacionados con TCA dentro de cuatro categorías diferentes que pueden ser de interés para la comunidad científica: (i) mensajes escritos por personas que padecen, o no, TCA, (ii) mensajes que fomentan, o no, el padecer un TCA, (iii) mensajes de carácter informativo, o de opinión y (iv) mensajes de carácter científico, o no.
Much of our society attaches great importance to physical appearance. Having a beautiful, or ideal, physical appearance is not necessarily synonymous with good health. There are people who may have a slim, strong body with good muscle definition, but do not maintain a very healthy lifestyle due to the use of substances that are not exactly healthy. In the pursuit of a slim physique, many think that they will also find happiness, and yet this is often not the case. The pressure imposed by the canons of beauty prevalent in society has led to a large part of society suffering from mental illnesses related to eating, better known as eating disorders (ED). Some examples are anorexia, bulimia, bulimarexia, binge eating disorders, etc. Social media and, more specifically, social networks, are technological tools that have an increasing number of users and through which it is easy to disseminate and spread information among society quickly. Despite the benefits of social networks, when a person suffers from an ED, they can promote or provoke a worsening of the patient’s own illness. Twitter is one of the most widely used social networks in the field of data collection for subsequent study and, in the field of health, it is the most widely used tool in the field of research as far as social networks are concerned. Many studies are focusing on obtaining tweets and generating models capable of classifying texts or analysing sentiment in different health topics. For all of the above reasons, this research has proposed to apply different data mining techniques by collecting different tweets related to ATT. Subsequently, text mining and natural language processing techniques have been applied to generate predictive models using different supervised machine learning techniques such as random forests, recurrent neural networks and even models known as Bidirectional Long Short-Term Memory (Bi- LSTM). After the application of these models, the accuracy obtained from the different classification models has been assessed to classify ATT-related tweets into four different categories that may be of interest to the scientific community: (i) messages written by people with or without ED, (ii) messages that do or do not encourage ED, (iii) messages of an informative or opinionated nature, and (iv) messages of a scientific or non-scientific nature.
Much of our society attaches great importance to physical appearance. Having a beautiful, or ideal, physical appearance is not necessarily synonymous with good health. There are people who may have a slim, strong body with good muscle definition, but do not maintain a very healthy lifestyle due to the use of substances that are not exactly healthy. In the pursuit of a slim physique, many think that they will also find happiness, and yet this is often not the case. The pressure imposed by the canons of beauty prevalent in society has led to a large part of society suffering from mental illnesses related to eating, better known as eating disorders (ED). Some examples are anorexia, bulimia, bulimarexia, binge eating disorders, etc. Social media and, more specifically, social networks, are technological tools that have an increasing number of users and through which it is easy to disseminate and spread information among society quickly. Despite the benefits of social networks, when a person suffers from an ED, they can promote or provoke a worsening of the patient’s own illness. Twitter is one of the most widely used social networks in the field of data collection for subsequent study and, in the field of health, it is the most widely used tool in the field of research as far as social networks are concerned. Many studies are focusing on obtaining tweets and generating models capable of classifying texts or analysing sentiment in different health topics. For all of the above reasons, this research has proposed to apply different data mining techniques by collecting different tweets related to ATT. Subsequently, text mining and natural language processing techniques have been applied to generate predictive models using different supervised machine learning techniques such as random forests, recurrent neural networks and even models known as Bidirectional Long Short-Term Memory (Bi- LSTM). After the application of these models, the accuracy obtained from the different classification models has been assessed to classify ATT-related tweets into four different categories that may be of interest to the scientific community: (i) messages written by people with or without ED, (ii) messages that do or do not encourage ED, (iii) messages of an informative or opinionated nature, and (iv) messages of a scientific or non-scientific nature.
Descripción
Categorías UNESCO
Palabras clave
minería de datos, minería de textos, procesamiento de lenguaje natural, redes neuronales, BERT, data mining, text mining, natural language processing, neural networks
Citación
Centro
Facultades y escuelas::E.T.S. de Ingeniería Informática
Departamento
Inteligencia Artificial