NL2Query™: Natural Language to SPARQL

Enterprise knowledge graphs, or graph databases, are growing in popularity and replacing or complementing organizations' relational databases. Graph users have traditionally needed to learn SPARQL or other complex querying languages to get that information, which limited accessibility to technical team members and hampered adoption.

Now, use NL2Query™ to search that database with plain English phrases and increase the value of your knowledge graph.

Capabilities

Instead of a SPARQL query, a user can use natural language to search the database. Historically, Lymba's NLP pipeline has been used to extract knowledge out of large volumes of text and store it in a graph database.

We use the same technology for the query: a person’s question is sent through the Lymba pipeline just like any other text and the system establishes a semantic representation of the data. Then the system converts the plain English entry into SPARQL, queries the database, and displays the retrieved result.

The system can query any graph, even if Lymba did not originally extract the data. Knowledge Base Creation is accomplished by leveraging the Lymba K-Extractor™ NLP pipeline, which includes 86 entity types and 26 semantic relationships. We then layer on an ontology to understand the context of the search.

Because NL2Query™ uses the standard SPARQL language, it works with most knowledge graph databases, including Stardog, Anzograph, MarkLogic, Allegrograph, and Oracle.

Organizations can now provide employees a new knowledge base to make quicker, more thoughtful decisions about their business by giving them an easy way to access their company’s data resources.