The Streaming Machine learning model, with classic NLP and statistical functions, classifies and segments intents. This gives a structure to score keywords and searches.
The Bidirectional Neural model converts words into vectors and simplifies it into a matrix graph. The neural classifier matches queries with intents in the database.
The Graph Neural Model combines vector and enriched data relationships in a matrix. This keeps the accuracy of answers high.
Our Entity Label Graph is an entity that identifies distinct words in a sentence. With answer deviation, the graph maps these words and filters metaphors in questions.
The Logical Graph represents NLP data flow through a series of computations in terms of relational and logical dependencies between individual hyponyms and hypernyms.
The Core Knowledge Graph is built with intent databases and external data. This helps data enrichment and collecting data behind keywords. This helps give accurate answers.