Semantic Similarity-based Recommendation in Big Data repositories

Abstract

The measurement and evaluation projects are defined using a formal measurement and evaluation framework, which allows us to consider the entity under monitoring, which is part of our analysis. It may be that the entity under monitoring has no previous experience or knowledge, which means that no suggestions can be recommended in certain situations. In this sense, and to solve this situation, the idea is to recommend according to the semantic similarity presented by the entities under monitoring, considering the measurements arising from the project of measurement and evaluation previously defined. The objective is precisely to be able to detect similar entities under monitoring by establishing a similarity score that allows for the provision of experiences and/or recommendations by analogy with another entity in the absence of these in the entity under analysis. Here we present a partial advance of this line of research, indicating the results obtained in terms of semantic similarity, similarity coefficients, recommendation strategies, and updating of different components of the data flow processing architecture.