Enriching the knowledge of a semantic net: extending two knowledge graphs in the financial domain

This thesis constitutes an example of applied research carried out with expert.ai’s R&D department. The project aims at enriching the company knowledge graph and at solving a text categorization task in the financial domain, focusing on the sub-domain of cryptocurrencies. Expert.ai is involved in the field of natural language understanding, offering companies and institutions professional services in this area. After illustrating the tools employed for the project and the main steps for its development, the reasons for creating a knowledge graph extension with concepts related to cryptocurrencies are discussed. To further enrich the company knowledge graph with concepts related to finance, another knowledge graph extension with the concepts of the Investopedia dictionary is created. The knowledge graph extension on cryptocurrencies focuses on the main criteria applied when choosing the terms to insert into it. The extension on the Investopedia dictionary focuses on the creation of the semantic chain, namely on how new concepts are linked to those already-existing and less specific concepts in the knowledge graph to create a hierarchy between them. The text categorization results highlight the positive outcome of creating the knowledge graph extension on cryptocurrencies. By enriching the knowledge graph, used as a tool containing the concepts used to write categorization rules for the text categorization task, precision, recall and F-measure, the three metrics used to evaluate the performance of text categorization, improve significantly. The applications of the knowledge graph extension on cryptocurrency and the one on the Investopedia dictionary are manyfold. They constitute a permanent tool that can be used by the company for further text categorization and entity extraction tasks related to the financial domain. The thesis concludes with some personal observations on carrying out applied research as an MA student in the field of languages and translation.