NLP Algorithms

Babelfish NLP algorithms are designed specifically for business use cases, allowing the system to understand business related topics as well as identify specific roles within the topic to understand meaning and context.

Babelfish NLP engine is embedded with proprietary NER and SRL algorithms, along with the standard dependency parsing algorithm, to extract contextual keywords and find relationship in data.

Named Entity Recognition:
Babelfish proprietary NER algorithm extracts and builds an entity list over existing business data.

Unlike NER APIs available for detecting generic named entities like person, product etc, Babelfish NER extracts data from existing business data sources and creates a list for performing named entity recogntion, based on topical classification available in the graph.

Semantic Role Labelling:
The Semantic Role Labeling helps in setting roles to words in a sentence. Semantic Roles can be explained as any of a set of semantic roles that a noun phrase may have in relation to a verb, for example agent, patient, location, source, or goal.

Babelfish SRL uses a set of proprietary roles that fit well into business contexts that takes basis from the defining nodes of the graph.

Dependency Parsing:
This algorithm extracts the grammatical relationships between different roles occurring in a given sentence. Most roles within business topics are nouns (people, place, product, digital assets, money and more) and with limited verbs (transaction, purchase, delivery and more), the primary role-defined noun is taken to be the structural center of clause structure.

How it works
A high level animation of the Symbolic AI in play

Illustrating how the NLP engine converts text inputs to database query and its associated visualization.