Artificial Intelligence based on Knowledge Graphs

The growing need to integrate heterogeneous and distributed content, to interpret it and provide it with context, has led knowledge graphs to a prominent position. The cases of Google's Knowledge Graph or Microsoft Graph are examples of its increasing usefulness and popularity. Knowledge graphs are the means to enrich data with data related to, but not within the organization's systems.

A knowledge graph has four great properties: it unifies heterogeneous and distributed information, it makes it interrogated by machines and people, it is expressive, and this expressiveness can be easily extended or expanded.

Knowledge graphs are an ideal solution for storing data extracted from the analysis of unstructured sources, such as documents, using natural language processing (NLP). They are also capable of storing structured data, including metadata that implicitly provides structure and context to information. In short, knowledge graphs allow data to be stored, provided with structure and context, while offering interrogation systems, information retrieval, knowledge discovery and analysis that make them essential in a very large number of use cases.

Through knowledge graph-based analysis, organic and dynamic relationships between digital assets, data sources, processes, people and interactions can be automatically discovered and exploited. A key aspect in this regard is the extraction of entities. Entities (people, organizations, places, events, etc.) can be identified through natural language processing techniques prior to data ingestion and subsequent disambiguation. In this way, knowledge graphs silently accumulate "smart data", that is, expressive data that the appropriate Artificial Intelligence systems can easily read and "understand" and whose exploitation offers high quality results.