	US 12,141,114 B2
	Semantic indices for accelerating semantic queries on databases
Rajesh Bordawekar, Westchester, NY (US)
Assigned to International Business Machines Corporation, Armonk, NY (US)
Filed by International Business Machines Corporation, Armonk, NY (US)
Filed on Dec. 9, 2021, as Appl. No. 17/546,343.
Prior Publication US 2023/0185788 A1, Jun. 15, 2023
Int. Cl. G06F 16/00 (2019.01); G06F 16/22 (2019.01); G06F 16/242 (2019.01); G06N 3/008 (2023.01)

CPC G06F 16/2228 (2019.01) [G06F 16/243 (2019.01); G06N 3/008 (2013.01)]

20 Claims

1. A method of using a computing device to enable and accelerate one or more cognitive intelligence queries, the method comprising:

receiving a database using a program for accelerating semantic queries on databases (PASQD) within the computing device, comprising an embedding table, wherein the embedding table comprises one or more rows, wherein the one or more rows each comprise one or more entities, wherein the one or more entities are each associated with a meaning vector through processing the entities through a trained database embedding model, and wherein the meaning vector comprises a semantic definition of its associated one or more entities;

converting the one or more meaning vectors into one or more semantic keys using locality sensing hashing;

building one or more semantic indexes using the semantic keys, wherein the semantic index comprises one or more tuples, and wherein each of the one or more tuples is a value that identifies one or more entries in the embedding table that share predefined features;

inserting one or more key-value pairs into the embedding table, wherein a key in the one or more key-value pairs is the one or more semantic keys, and wherein a value in the one or more key-value pairs is the one or more tuples; and

using the semantic indexes and an inputted entity to respond to a request to locate the inputted entity upon receiving one or more cognitive intelligence queries, wherein a cognitive intelligence query comprises the request.