Implementación de la Técnica de los K-Vecinos en un Algoritmo Recomendador para un Sistema de Compras Utilizando NFC y Android

Abstract

Introduction: This paper aims to present the design of a mobile application involving NFC technology and a collaborative recommendation algorithm under the K-neighbors technique, allowing to observe personalized suggestions for each client.

Objective: Design and develop a mobile application, using NFC technologies and K-Neighbors Technique in a recommendation algorithm, for a Procurement System.

Methodology: The process followed for the design and development of the application focuses on:

• Review of the state of the art in mobile shopping systems.

• State-of-the-art construction in the use of NFC technology and AI techniques for recommending systems focused on K-Neighbors Algorithms

• Proposed system design

• Parameterization and implementation of the K-Neighbors Technique and integration of NFC Technology

• Proposed System Implementation and Testing.

Results:  Among the results obtained are detailed:

• Mobile application that integrates Android, NFC Technologies and a Technique of Algorithm Recommendation

• Parameterization of the K-Neighbors Technique, to be used within the recommended algorithm.

• Implementation of functional requirements that allow the generation of personalized recommendations for purchase to the user, user ratings

Conclusions: The k-neighbors technique in a recommendation algorithm allows the client to provide a series of recommendations with a level of security, since this algorithm performs calculations taking into account multiple parameters and contrasts the results obtained for other users, finding the articles with a Greater degree of similarity with the customer profile. This algorithm starts from a sample of similar, complementary and other unrelated products, applying its respective formulation, we obtain that the recommendation is made only with the complementary products that obtained higher qualification; Making a big difference with most recommending systems on the market, which are limited to suggest the best-selling, best qualified or in the same category.