Demo Application for LETO: Learning Engine Through Ontologies

Proceedings of Recent Advances in Natural Language Processing, pages 276–284, Varna, Bulgaria, Sep 2–4, 2019. https://doi.org/10.26615/978-954-452-056-4_032 276 Demo Application for LETO: Learning Engine Through Ontologies Suilan Estevez-Velarde1, Andr´es Montoyo2, Yudivi´an Almeida-Cruz1, Yoan Guti´errez3, Alejandro Piad-Morffis1, and Rafael Mu˜noz2 1School of Math and Computer Science, University of Havana, Cuba {sestevez,yudy,apiad}@matcom.uh.cu 2Department of Languages and Computing Systems, University of Alicante, Spain 3U.I. for Computer Research (IUII), University of Alicante, Spain {montoyo,ygutierrez,rafael}@dlsi.ua.es Abstract The massive amount of multi-formatted information available on the Web neces- sitates the design of software systems that leverage this information to obtain knowledge that is valid and useful. The main challenge is to discover relevant in- formation and continuously update, en- rich and integrate knowledge from vari- ous sources of structured and unstructured data. This paper presents the Learning En- gine Through Ontologies (LETO) frame- work, an architecture for the continuous and incremental discovery of knowledge from multiple sources of unstructured and structured data. We justify the main design decision behind LETO’s architecture and evaluate the framework’s feasibility using the Internet Movie Data Base (IMDB) and Twitter as a practical application. 1 Introduction In recent years, research in machine learning, knowledge discovery, data mining and natural lan- guage processing, among others, have produced many approaches and techniques to deal with the large amount of information available on the Inter- net to carry out a variety of tasks, such as, for ex- ample building search (Brin and Page, 1998) and recommendation systems (Davidson et al., 2010) that could be used to improve business, health-care and political decisions (Ferrucci et al., 2013). The purpose of our proposal is to present LETO: Learning Engine Through Ontologies, a framework to automatically and gradually extract knowledge from different sources (both structured and unstructured), building internal representa- tions that can be adapted to and integrated in mul- tiple domains. The current state of LETO’s imple- mentation is a computational prototype that illus- trates the different components of its architecture and demonstrate its feasibility. Inspired by the dif- ferent processes that occur during human learning, we design the framework’s architecture as a learn- ing pipeline that gradually builds more complex knowledge. In a simplified view, the human learning pro- cess can be modeled as a continuous loop that transforms sensorial data into knowledge (see Fig- ure 1) (Gross, 2015). Humans collect information about the environment through senses, where the human brain attempts to detect relations between individual signals to form a more structured rep- resentation of reality. By relating as many sig- nals as possible, humans build a much richer se- mantic representation of the environment, which is unconsciously filtered storing only the most rel- evant part. In order to achieve this, the brain is able to access to stored experiences about what has been important before, and what is already known. This feedback loop also evaluates previ- ously known facts, and modifies them at the light of new experiences. In time, humans not only learn new facts, but also learn better ways of learn- ing. …

Content truncated to 500 words