SYSTEM AND METHOD FOR VISUALIZING AND INTERPRETING PROPERTY INCLUDED LEARNED ENTITY REPRESENTATION FROM NATURAL LANGUAGE TEXT

摘要

In the present invention, disclosed are a method for interpreting and visualizing properties included in an entity representation learned from a natural language text and a system thereof wherein interpreted properties and a representation vector are visualized and provided to enable a user to recognize the same in an easy manner at a glance. The method for interpreting and visualizing an entity representation comprises the following steps: generating a representation vector of an entity representation for entity representation learning for understanding properties representing meaning of an entity existing in the real world with respect to a natural language text; generating a vector activation pattern by performing filtering with respect to the representation vector of at least two entities among the representation vectors generated for each of multiple entities; and mapping target properties related to at least two entities to the vector activation pattern.