Desarrollo de un sistema recomendador de ubicaciones para sitios turísticos del Centro Histórico de Quito basado en Procesamiento de Lenguaje Natural

Abstract

Recommendation systems that use traditional methods for classifying and generating recommendations can be improved by using deep learning techniques in conjunction with machine learning techniques. The purpose of this work is to provide a solution to improve the effectiveness of the recommendations of these systems, using natural language processing. To achieve this goal, a natural language processing model using deep learning approaches was trained.

The development of the model is made up of two phases: a) the data collection phase and b) the development phase. In the data collection phase, the training data set was obtained from surveys carried out on consultations of tourist sites with their respective category. In the development phase, the model was trained using the data set obtained through the implementation of a training event offered by the IBM service called Watson Natural Language Classifier (NLC). The developed model allows classifying text sentences in 62 categories obtained through the training event.

To test the aforementioned model, a system was developed that provides recommendations for tourist sites for the Historic Center of Quito. This system uses the NLC service as its main source for processing text statements. As a result, recommendations of points of interest associated with tourist sites that fit the user's preferences were obtained.