VERSES AI Inc. announced the filing of a provisional patent application representing a new method for Predictive Querying on vector graph document databases. Probabilistic querying is an approach to database queries that seeks to provide a user with additional information "pred predicted" to be of interest to the user, given the context implicit in the query and around its prompter. VERSES' novel Predictive Query method addresses limitations in prior arts by providing a system to perform probabilistic queries on the most advanced class of databases: vector graph document databases.

Predictive Querying operate on vector graph document databases by implementing Hyperspatial Modeling Language (HSML) and an inference algorithm to generate a probabilistic and contextualized result. The Predictive Querying method is the first querying method that allows probabilistic querying on vector graph document databases, that enables an engine to generate rich predictions about the information being searched for by the user based on comparative, relationship and similarity information. Knowledge graphs represent entities - any physical or conceptual "thing" one can have information about in the real world (e.g., a robot, a sofa, a waypoint in space, a specification of an activity) - and the relationships between them.

HSML is a modeling language for qualifying the relationships between entities in a knowledge graph. An HSML vector graph document database is structured as an HSML knowledge graph and allows for informationretrieval using complex queries that can simultaneously involve the comparison of entities (e.g., "find people older than Steve"), the identification of cause-effect relations (e.g., " who is Steve's manager"?), as well as the evaluation of similarity between entities (e.g., 'which employees have an educational background closest to Steve's"?). Compared to vector graph document databases, other classes of databases are limited to either comparative, relationship or similarity search.

Now, because of this new method for Predictive Querspatial Querying on vector graph documents databases, VERSES returns the most probable and relevant match to a user's rich implicit goal (e.g., inferredring and returning the most probable brand, model, and location to a search for "cheap sunglasses" along with the best deals on cycling clothes matching the style of the sunglasses).