Patent Recommendation Engine Using Graph Database

Abstract

Accurate analysis of patents is an essential tool for modern companies. The idea behind the patent recommendation engine is to build solutions that enhance the quality as well the quantity of extractable data from a patent and discover meaningful relations, helping companies by spending fewer resources in terms of time and manpower. The recommendation engine is a concept of having the patent data transferred into a graph database and executing queries to answer questions specific to certain business use cases such that the task is significantly easier, less resource-intensive, and less complex when compared to the same being performed by a conventional relational database. The engine concept accepts a single input and forming clusters from the single starting point based on chain queries. It can then run the required algorithms on the clusters formed to select the best-fit data. The recommendation engine uses Neo4j as the database. Neo4j is a NoSQL graph database that focuses more on the relationship between various data rather than the data itself. We extract the data from our existing databases and then ingest them into Neo4j. Cypher queries power the engine that answers very specific questions within very little time.