Integrated system based on the automation of the extraction and characterization, using natural processing language, of user ratings in sales platforms

Abstract

[Abstract] At present, more and more importance is given to what people think, what they think and what are your preferences. With the rise of social networks and online stores, these data are accessible more than ever in a simple and global way, so it is increasingly important to know manage and analyze all this information. Thus, it is not surprising that over time more companies are focusing their interest on sentiment mining, which allows them to identify possible business opportunities, maintain a good reputation of the company in the networks or improve the marketing your products. For all these reasons, we are interested in building a tool capable of extracting from all the reviews that exists throughout the Internet, what is the opinion of a product/service (its most praised characteristics, its greatest shortcomings…), so that, through a good analysis, we can extract important information in an automated way. This paper will discuss how to build a tool capable of analyzing and retrieving reviews from the World Wide Web. Building two main systems, one for extracting reviews with the use of scraping tools and another for analyzing sentiment in texts with the use of machine learning and natural language processing.[Resumo] Na actualidade cada vez dáselle máis importancia a qué opinan as persoas, qué é o que pensan e cales son as suas preferencias. Coa subida da popularidade das redes sociais e as tendas online, estes datos estan máis accesibles que nunca, de manera sinxela e global, co que cada vez é máis importante saber xestionar e analizar toda esa información. Así, non é de extrañar que co tempo máis compañías estén centrando o seu interés na minaría de sentimentos, que lles permite posibles oportunidades de negocio, mantener unha boa reputación de empresa nas redes e mellorar o marketing. Por todo isto, interésanos construir unha ferramenta capaz de extraer de todas as reseñas que existen ao longo de internet, cal é a opinión que se ten dun produto/servizo, de forma que, a través dunha boa análise podamos obter información de importancia de maneira automatizada. Neste traballo falarase de como construir unha ferramenta capaz de analizar e obter as reseñas da World Wide Web. Construindo un sistema de extracción de reviews co uso de ferramentas de scraping e outro de análise de sentimentos en textos co uso de aprendizaxe automático e procesamento da linguaxe natural.