Information extraction, enrichment, and caching framework for augmented reality applications

A system comprising:

• a datastore comprising data relating to items in a real-world;
• and
• a computing device configured to:
• receive real-world data from a client viewport;
• receive a client selection of an item of the real-world data that is associated with a first query in the client viewport;
• identify, using a first prediction model, a current user analysis phase;
• determine, using a second prediction model, a predicted next user interaction, wherein the second prediction model is trained based at least in part on user past interactions corresponding with one or more of a retrieval phase, fusion phase, sensemaking phase, and a navigation phase;
• while a user is analyzing a result-set of the first query in the current user analysis phase, precompute data from the datastore based at least in part on the predicted next user interaction by performing a fusion query operation to retrieve supplemental data related to the item that is not available in the client viewport;
• store the precomputed data in a middleware cache during the current user analysis phase;
• responsive to receiving a subsequent user interaction associated with a subsequent query, retrieve at least a portion of the precomputed data relating to the item from the middleware cache;
• interpret a subsequent query result using a relational data model;
• modify the result-set using the precomputed data in an additive fashion;
• render the modified result-set by translating the relational data model from a query space and encoding at least a portion of the precomputed data to generate to at least one visual artifact;
• and
• output the at least one visual artifact to the client viewport with at least some of the real-world data.

A computing device comprising:

• a middleware cache;
• a cache manager;
• and
• a prediction engine comprising a first prediction model and a second prediction model configured to:
• receive a client selection of an item of real-world data that is associated with a first query from a client running an augmented reality application
• identify, using the first prediction model, a current user analysis phase;
• determine, using the second prediction model, a predicted next user interaction, wherein the second prediction model is trained based at least in part on user past interactions corresponding with one or more of a retrieval phase, fusion phase, sensemaking phase, and a navigation phase;
• while a user is analyzing a result-set of the first query in the current user analysis phase, precompute data based at least in part on the predicted next user interaction by performing a fusion query operation to retrieve supplemental data related to the item that is not available in a client viewport of the computing device;
• store the precomputed data in a middleware cache during the current user analysis phase;
• and
• responsive to receiving a subsequent user interaction associated with a subsequent query, retrieve at least a portion of the precomputed data relating to the item from the middleware cache;
• interpret a subsequent query result using a relational data model;
• modify the result-set using the precomputed data in an additive fashion;
• render the modified result-set by translating the relational data model from a query space and encoding at least a portion of the precomputed data to generate to at least one visual artifact;
• and
• output the at least one visual artifact to the client viewport with at least some of the real-world data.