US 2026/0300775 A1Published application

Contextual search on a database based on dynamic reward function using relevancy count

AssigneeTata Consultancy Services Limited

InventorsJyoti BHAT, Amit KALELE, Rohan Dhanraj ATALE, Sudhanshu DIMRI

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method and system for contextual search on a database based on dynamic reward function using relevancy count is disclosed. Generative artificial intelligence has challenges in handling complex natural language queries for database transactions that require contextual rather than direct column mappings. The method disclosed extracts contextually relevant data from relational databases using a combination of Retrieval-Augmented Generation (RAG) and a contextual Monte Carlo Tree Search (CMCTS) with a dynamic reward function. The method disclosed uses the dynamic reward function during the simulation step of the CMCTS on a Monte Carlo Tree by calculating a contextually relevant score of a current node in the Monte Carlo Tree. The method also uses a modified upper confidence bound during selection step of the CMCTS. The method disclosed solves the real-world problem of searching for companies relevant to a given topic or capability.

Claims18

Dependent claims

A processor implemented method comprising:

selecting, via one or more hardware processors, a column comprising a plurality of attributes from a relational database based on a natural language query received from a user, using a pre-defined set of heuristics rules applied on a set of columns of the relational database;
selecting, via the one or more hardware processors, a set of attributes amongst the plurality of attributes based on a Retrieval-Augmented Generation (RAG), a large language model and the natural language query;
determining, via the one or more hardware processors, an optimal set of attributes from the set of attributes by performing a Contextual Monte-Carlo tree search (CMCTS), wherein the CMCTS constructs a Monte Carlo Tree comprising a set of nodes representing the set of attributes, based on (i) an upper confidence bound (UCB) in a selection step of the CMCTS, and (ii) a reward function calculated in a simulation step of the CMCTS based on a relevancy count and a total count;
and
determining, via the one or more hardware processors, a set of relevant rows corresponding to the optimal set of attributes from the relational database to generate a contextually relevant response to the natural language query.

2 more elements

A system, comprising:

a memory storing instructions;
one or more communication interfaces;
and
one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to:
select a column comprising a plurality of attributes from a relational database based on a natural language query received from a user, using a pre-defined set of heuristics rules applied on a set of columns of the relational database;
select a set of attributes amongst the plurality of attributes based on a Retrieval-Augmented Generation (RAG), a large language model and the natural language query;
determine an optimal set of attributes from the set of attributes by performing a Contextual Monte-Carlo tree search (CMCTS), wherein the CMCTS constructs a Monte Carlo Tree comprising a set of nodes representing the set of attributes, based on (i) an upper confidence bound (UCB) in a selection step of the CMCTS, and (ii) a reward function calculated in a simulation step of the CMCTS based on a relevancy count and a total count;
and
determine a set of relevant rows corresponding to the optimal set of attributes from the relational database to generate a contextually relevant response to the natural language query.

6 more elements

One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:

selecting a column comprising a plurality of attributes from a relational database based on a natural language query received from a user, using a pre-defined set of heuristics rules applied on a set of columns of the relational database;
selecting a set of attributes amongst the plurality of attributes based on a Retrieval-Augmented Generation (RAG), a large language model and the natural language query;
determining an optimal set of attributes from the set of attributes by performing a Contextual Monte-Carlo tree search (CMCTS), wherein the CMCTS constructs a Monte Carlo Tree comprising a set of nodes representing the set of attributes, based on (i) an upper confidence bound (UCB) in a selection step of the CMCTS, and (ii) a reward function calculated in a simulation step of the CMCTS based on a relevancy count and a total count;
and
determining a set of relevant rows corresponding to the optimal set of attributes from the relational database to generate a contextually relevant response to the natural language query.

2 more elements

Dependencies are inferred from the claim text.