Creación automatizada de grafos de conocimiento (Automated Knowledge Graph Creation)

Abstract

Abstract: This research proposes a framework for the automatic creation of a knowledge graph from text in natural language, considering the difficulty of machines in understanding it. To achieve this goal, a systematic literature review was carried out to search for information about natural language processing (NLP) techniques, the framework was proposed, two use cases are proposed in the areas of Electronics and Covid-19, for the tests a text is entered in natural language and as a result a knowledge graph is obtained. The results of the tests indicate the importance of the NLP techniques used, and their contribution in the information extraction for the graph. Therefore, it is recognized that one of the most important steps at the beginning of the analysis is the text preprocessing, and in our case another of important process after the previous one was the "Semantic role labeling", thus it was possible to recognize the techniques that allow to start from a text in natural language and obtaining the triplets of the knowledge graph.