Invention Application
- Patent Title: PRIMITIVE-BASED QUERY GENERATION FROM NATURAL LANGUAGE QUERIES
-
Application No.: US17072935Application Date: 2020-10-16
-
Publication No.: US20220121656A1Publication Date: 2022-04-21
- Inventor: Zuye Zheng , Ryoji Osawa , James Diefenderfer , Nadeem Marfani Aboobaker , Praveen Pandey , Sameer Sethi
- Applicant: salesforce.com, Inc.
- Applicant Address: US CA San Francisco
- Assignee: salesforce.com, Inc.
- Current Assignee: salesforce.com, Inc.
- Current Assignee Address: US CA San Francisco
- Main IPC: G06F16/242
- IPC: G06F16/242 ; G06N3/08 ; G06K9/62

Abstract:
Methods, systems, and devices supporting primitive-based query generation are described. A database system may receive a natural language query associated with a subset of data sets stored at the database system, determine a set of tokens included in the natural language query, and label one or more tokens of the set of tokens with one or more primitives from a set of supported primitives associated with the subset of data sets. The database system may determine a set of higher order concepts corresponding to query operations, query conditions, or both based on the one or more labeled tokens. The database system may train a neural network to learn a grammatical structure of the natural language query and generate a database query for querying the database system based on the one or more labeled tokens, the set of higher order concepts, and the neural network.
Public/Granted literature
- US11803541B2 Primitive-based query generation from natural language queries Public/Granted day:2023-10-31
Information query