Deep learning aplicado al resumen de texto
Loading...
Download
Official URL
Full text at PDC
Publication date
2017
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
Estudiamos la aplicación del deep learning a la tarea del resumen automático y abstractivo de textos. Para ello, hemos realizado un extenso estudio de la bibliografía y, siguiendo las últimas líneas de investigación, diseñamos nuestro propio modelo, que cuenta con una arquitectura encoder-decoder y un mecanismo de atención. Este modelo lo entrenamos sobre una tarjeta gráfica donada por NVIDIA. Finalmente, evaluamos los resultados obtenidos: aunque muchos de los resúmenes son buenos, identificamos algunos problemas como la repetición de frases y la falta de vocabulario.
We study the application of deep learning techniques to the task of automatic abstractive summarization. For this purpose, we have done a broad study of the literature and, following in the line of recent research, we design our own model, with an encoder-decoder architecture and a mechanism of attention. We train this model using a GPU that was granted to us by NVIDIA. Lastly, we evaluate the results we obtained: although plenty of summaries are good, we identify some problems like the repetition of phrases and the lack of vocabulary.
We study the application of deep learning techniques to the task of automatic abstractive summarization. For this purpose, we have done a broad study of the literature and, following in the line of recent research, we design our own model, with an encoder-decoder architecture and a mechanism of attention. We train this model using a GPU that was granted to us by NVIDIA. Lastly, we evaluate the results we obtained: although plenty of summaries are good, we identify some problems like the repetition of phrases and the lack of vocabulary.
Description
Trabajo de Fin de Grado en Ingeniería Informática y Matemáticas (Universidad Complutense, Facultad de Informática, curso 2016/2017)