Contextualization of generative language models based on entity resource identifiers
AssigneeMicrosoft Technology Licensing, LLC
InventorsSujay Kumar Jauhar, Silviu-Petru Cucerzan, Nirupama Chandrasekaran, Allen Herring, Jinheon Baek
Life of this patent
Estimating the term…
Abstract
The disclosed concepts relate to contextualization of generative language models. In some implementations, a linked entity database is populated with entity resource identifiers of entities extracted from a search log by an entity linker. A contextualized prompt data structure is generated based on the linked entity database, e.g., by including linked entity context information in the contextualized prompt data structure. A response to the contextualized prompt data structure is received, where the response is conditioned on the linked entity context information.
Claims20
A computer-implemented method comprising:
- inputting a search log to an entity linker, the search log including web search queries submitted by a particular user to obtain web search results from a search engine and clicked web pages selected by the particular user from the web search results;
- receiving, from the entity linker, first linked entity resource identifiers of first linked entities identified by the entity linker by processing the search log;
- populating a linked entity database with the first linked entity resource identifiers received from the entity linker;
- receiving a current natural language query from the particular user;
- generating a contextualized prompt data structure based at least on the current natural language query, the contextualized prompt data structure being personalized for the particular user based at least on linked entity context information derived from the linked entity database;
- inputting the contextualized prompt data structure to a generative language model;
- receiving a response to the contextualized prompt data structure generated by the generative language model, wherein the response is conditioned on the linked entity context information;
- and
- replying to the current natural language query based at least on the response.
A system comprising:
- a linked entity database storing first linked entity resource identifiers of linked entities identified by an entity linker by processing a search log for a particular user;
- a hardware processing unit;
- and
- a storage resource storing computer-readable instructions which, when executed by the hardware processing unit, cause the system to:
- receive a current natural language query from the particular user;
- generate a contextualized prompt data structure based at least on the current natural language query, the contextualized prompt data structure being personalized for the particular user based at least on linked entity context information derived from the linked entity database;
- input the contextualized prompt data structure to a generative language model;
- receive a response to the current natural language query generated by the generative language model, wherein the response is conditioned on the linked entity context information included in the contextualized prompt data structure;
- and
- reply to the current natural language query based at least on the response.
A computer-readable storage medium storing computer-readable instructions which, when executed by a processing unit, cause the processing unit to perform acts comprising:
- receiving a current natural language query from a particular user;
- accessing a linked entity database storing first linked entity resource identifiers of linked entities identified by an entity linker by processing a search log for the particular user;
- generating a contextualized prompt data structure based at least on the current natural language query, the contextualized prompt data structure being personalized for the particular user based at least on linked entity context information derived from the linked entity database;
- inputting the contextualized prompt to a generative machine learning model;
- receiving a response to the contextualized prompt generated by the generative machine learning model, wherein the response is conditioned on the linked entity context information;
- and
- replying to the current natural language query based at least on the response.
Dependencies are inferred from the claim text.