Knowledge Graph

Our Knowledge Graph provides a domain-independent representation of the real world through concepts and their related meanings and the different relationships that exist among concepts.

The Knowledge Graph is designed to interact with our semantic engine to resolve ambiguity in the meaning of each word, a fundamental step in the NLU process.

In the same way that human knowledge can be improved by learning new things,’s knowledge may also be expanded through the acquisition of new knowledge from subject matter experts via tools like Studio or by machine learning through the analysis of tagged content.

The Knowledge Graph is an open box, meaning its structure is created by humans and is understandable by humans, the opposite  of what happens with pure machine learning systems.

Every item in the Knowledge Graph contains a set of attributes (grammar type, semantic link, definition/meaning, domain, frequency) that establish the characteristics of words and concepts.

It contains words gathered in groups of words and concepts that express the same meaning and are connected to one other by millions of logical and language-related links describing the relationship between concepts.

In the Knowledge Graph, each syncon is a node that is linked to other nodes through semantic relationships in a hierarchical structure. In this way, each node, in addition to its meaning and attributes, is enriched by the characteristics and meaning of the nodes above or below it. Each of these links identifies a kind of relationship that links the concepts in a language, and the links are principles used to organize the concepts in the Knowledge Graph. For example, concepts are organized starting from less specific to more specific (vehicle / car / SUV), or as potential subjects or objects of a verb, etc.