Granted Patent B1

US 12,694,227 · App. 19/291,376 · Granted Jul 28, 2026

Inventor: Zeev Barkan (Jerusalem, IL)

G06F40/35G06F40/279G06F40/289G10L15/22

Abstract

Claims (33)

1 . A system of correcting inaccuracies in chatbot responses, the system comprising a processing circuitry (PC) configured to:

receive at least part of a chatbot's response to a user prompt, and receive data indicative of data sources, the data sources being indicated by the chatbot as relevant to the chatbot's response;

present, to the user, data derivative of the chatbot response, and receive, from the user, data indicative of one or more text segments,

the one or more text segments being associated with a user-suspected hallucination or inaccuracy of the chatbot response;

perform, on one or more of the data sources, at least one of:

a text search, and

a semantic search

of one or more of the text segments; and

present, to the user, data derivative of the text search and/or semantic search, thereby providing confirmation or correction of the chatbot response.

2 . The system of claim 1 , wherein at least one of the one or more data sources is provided by the chatbot.

3 . The system of claim 1 , wherein the received data indicative of one or more text segments is derivative of, at least, a user utilization of a device's user interface while viewing the chatbot response.

4 . The system of claim 1 , wherein the data derivative of the text search and/or semantic search is null, thereby being indicative of a hallucination in the chatbot response.

5 . The system of claim 1 , wherein the data derivative of the text search and/or semantic search is a text segment of a data source, thereby providing a clarification, confirmation, or correction of the response.

6 . The system of claim 1 , wherein the chatbot is a large language mode (LLM).

7 . The system of claim 1 , the data derivative of the text search and/or semantic search is: a corrected chatbot response.

8 . A processing circuitry-based method of correcting inaccuracies in chatbot responses, the method comprising:

receiving at least part of a chatbot's response to a user prompt, and receive data indicative of data sources, the data sources being indicated by the chatbot as relevant to the chatbot's response;

presenting, to the user, data derivative of the chatbot response, and receive, from the user, data indicative of one or more text segments,

the one or more text segments being associated with a user-suspected hallucination or inaccuracy of the chatbot response;

performing, on one or more of the data sources, at least one of:

a text search, and

a semantic search

of one or more of the text segments; and

presenting, to the user, data derivative of the text search and/or semantic search, thereby providing confirmation or correction of the chatbot response.

9 . A computer program product comprising a computer readable non-transitory storage medium containing program instructions, which program instructions when read by a processing circuitry, cause the processing circuitry to perform a method of correcting inaccuracies in chatbot responses, the method comprising:

receiving at least part of a chatbot's response to a user prompt, and receive data indicative of data sources, the data sources being indicated by the chatbot as relevant to the chatbot's response;

presenting, to the user, data derivative of the chatbot response, and receive, from the user, data indicative of one or more text segments,

the one or more text segments being associated with a user-suspected hallucination or inaccuracy of the chatbot response;

performing, on one or more of the data sources, at least one of:

a text search, and

a semantic search

of one or more of the text segments; and

presenting, to the user, data derivative of the text search and/or semantic search, thereby providing confirmation or correction of the chatbot response.

References Cited (33)