Mostrar el registro sencillo del ítem
Opinion mining of online product reviews using a lexicon-based algorithm
dc.contributor.author | Martín-Borregón Musso, Ignacio | es-ES |
dc.contributor.other | Universidad Pontificia Comillas, | es_ES |
dc.date.accessioned | 2018-05-02T10:29:41Z | |
dc.date.available | es_ES | |
dc.date.issued | 2018 | es_ES |
dc.identifier.uri | http://hdl.handle.net/11531/26716 | |
dc.description | The worldwide social media is a rich resource of user-generated data, which can help organizations to formulate their business strategies, or affect the process of decision making in product or service design and implementation. The focus of this thesis is on extraction and analysis of unstructured product reviews for training predictive models, which recognize a specific range of human affective states. These affective states include emotions, moods, opinions, attitudes, as well as continuous dimensions for sentiment characterization, such as valence or intensity. Based on the sentiment analysis, the task is to build a domain-specific sentiment dictionary of English words form the product posts and comments, and to propose and test of lexicon-based algorithm to predict the user opinions. | es_ES |
dc.description.abstract | Este trabajo de fin de grado, realizado en la Universidad de Zagreb durante un programa de intercambio, se centra en extraer y analizar las críticas que los usuarios de un tipo de producto comentan en Facebook. El primer análisis a realizar se trata de un "análisis de sentimientos", una técnica en desarrollo que mostrara los sentimientos expresados por los usuarios en los comentarios a las publicaciones relacionadas con el producto. Este análisis se usará para dar una idea general del alcance de esta técnica, aunque como se podrá comprobar también tiene limitaciones que se deben mejorar. Con todo ello, se tratará de predecir las reacciones de los usuarios a las publicaciones de Facebook relacionadas con el producto (reacciones que están vinculadas a la opinión que los usuarios tienen del producto). Para ello, la tarea en concreto a realizar será recopilar información, construir con ella un diccionario que asigne a cada palabra una "cantidad de sentimiento" y por último desarrollar un algoritmo que permita la predicción de estas reacciones de Facebook. Por último, se analizarán los resultados de esta predicción y se buscará una manera de mejorar aún más los mismos. Con todo ello, se presentarán conclusiones de la técnica del "análisis de sentimientos" en la red social de Facebook, de las hipótesis planteadas a lo largo del proyecto y de cómo se podría continuar la mejoría del trabajo. | es-ES |
dc.description.abstract | This bachelor thesis, which was done from the University of Zagreb during an exchange program, is focused on extraction and analysis of unstructured product reviews that users of a certain type of product comment on Facebook. The first analysis that will be made is a "sentiment analysis", which is a technique being developed that will retrieve the sentiments expressed by users of the platform in the comments to the posts related to this type of product. This analysis will be used in order to give a general idea of the scope of this technique, although it will be shown that it has its own limitations that could be improved. With all that, the thesis present a model for the prediction of the user's reactions to the posts related to the product (these reactions are related to the opinion of the users on the product). In order to achieve this, the task is to extract a huge and representative dataset, build a domain-specific sentiment dictionary that will match every word appearing in the dataset to a sentiment value and build an algorithm for prediction of this Facebook reactions. Finally, the results of this prediction will be analyzed and an improvement of the algorithm will be proposed. With all that, conclusions will be presented about the sentiment analysis technique applied to Facebook data, the hypothesis raised throughout the project and about how the results of this thesis could still be improved. | en-GB |
dc.format.mimetype | application/pdf | es_ES |
dc.language.iso | en-GB | es_ES |
dc.subject.other | IEM-N (KL0-electronica) | es_ES |
dc.title | Opinion mining of online product reviews using a lexicon-based algorithm | es_ES |
dc.type | info:eu-repo/semantics/bachelorThesis | es_ES |
dc.rights.accessRights | info:eu-repo/semantics/openAccess | es_ES |
dc.keywords | redes sociales, gestión de operaciones, análisis de sentimientos, diccionario de sentimientos, críticas de productos, modelos predictivos | es-ES |
dc.keywords | social networks, operations management, sentiment analysis, sentiment dictionary, product reviews, opinion mining, training predictive models | en-GB |