Aplicação do algoritmo k-means para geração de um sistema de recomendação de produções de uma plataforma de streaming

Abstract

With the popularization of internet use in recent years, the film industry needed to reinvent itself, giving rise to the creation of new technological products, the main one being streaming platforms. The lifetime of subscribers to streaming platforms is directly related to the number of occasions they have watched a video and enjoyed it. If subscribers cannot find movies and series that interest and engage them, they tend to abandon the platform. Given this, this monograph presents the development of a recommendation system for movies and series from the Amazon Prime Video streaming service, using the K-means algorithm to group the productions according to the similarity of their synopses. After grouping the titles, a manual evaluation of the 37 clusters created was performed, based on the similarity of the genres of the movies and series that were in the same cluster and their synopses. The clusters in its great majority contained productions with similar genres and synopses, it can be cited for example the cluster that contains all the films of the Evangelion anime saga, with only one group where it was concentrated the titles that do not have a similarity of genre and synopses between them.