System and method for generating weighted query representations for enhanced retrieval augmented generation

1 . A computer-implemented method, executed on a computing device, comprising:

receiving a query to be processed by a generative artificial intelligence (AI) model;

extracting a query topic from the query;

generating a weighted query topic embedding for the query topic;

generating a plurality of weighted chunk topic embeddings for a plurality of chunks based on a frequency of each term in a particular chunk from a plurality of chunks of a target document and each term's frequency across all of the plurality of chunks;

identifying a candidate chunk from the plurality of chunks by determining a similarity between the weighted query topic embedding and the plurality of weighted chunk topic embeddings for the plurality of chunks;

generating a prompt using the query and the candidate chunk; and

providing the prompt to the generative AI model.

2 . The computer-implemented method of claim 1 , wherein identifying a candidate chunk includes identifying a predefined number of most similar candidate chunks.

3 . The computer-implemented method of claim 1 , further comprising:

generating the plurality of chunks for a plurality of text portions of the target document;

generating a plurality of chunk topics by extracting a topic for each respective chunk of the plurality of chunks; and

generating a weighting for the topic for each respective chunk of the plurality of chunks.

4 . The computer-implemented method of claim 3 , further comprising:

generating a weighted chunk topic embedding for each topic of a plurality of topics for a respective chunk; and

generating a weighted average chunk topic embedding for the respective chunk from the weighted chunk topic embedding for each topic of the plurality of topics for the respective chunk.

5 . The computer-implemented method of claim 1 , wherein extracting the query topic includes extracting a plurality of topics for the query.

6 . The computer-implemented method of claim 5 , wherein generating the weighted query topic embedding includes generating a weighting for each topic of the plurality of topics for the query.

7 . The computer-implemented method of claim 6 , wherein generating the weighted query topic embedding includes:

generating a weighted query topic embedding for each topic; and

generating a weighted average query topic embedding from the weighted query topic embedding for each topic.

8 . A computer program product residing on a non-transitory computer readable medium having a plurality of instructions stored thereon which, when executed by a processor, cause the processor to perform operations comprising:

receiving a query to be processed by a generative artificial intelligence (AI) model;

extracting a query topic from the query;

generating a weighted query topic embedding for the query topic;

generating a plurality of weighted chunk topic embeddings for a plurality of chunks based on a frequency of each term in a particular chunk from a plurality of chunks of a target document and each term's frequency across all of the plurality of chunks;

identifying a candidate chunk from the plurality of chunks by determining a similarity between the weighted query topic embedding and the plurality of weighted chunk topic embeddings for the plurality of chunks;

generating a prompt using the query and the candidate chunk; and

providing the prompt to the generative AI model.

9 . The computer program product of claim 8 , wherein identifying a candidate chunk includes identifying a predefined number of most similar candidate chunks.

10 . The computer program product of claim 8 , wherein the operations further comprise:

generating the plurality of chunks for a plurality of text portions of the target document;

generating a plurality of chunk topics by extracting a topic for each respective chunk of the plurality of chunks; and

generating a weighting for the topic for each respective chunk of the plurality of chunks.

11 . The computer program product of claim 8 , wherein the operations further comprise:

generating a weighted chunk topic embedding for each topic of a plurality of topics for a respective chunk; and

generating a weighted average chunk topic embedding for the respective chunk from the weighted chunk topic embedding for each topic of the plurality of topics for the respective chunk.

12 . The computer program product of claim 8 , wherein extracting the query topic includes extracting a plurality of topics for the query.

13 . The computer program product of claim 12 , wherein generating the weighted query topic embedding includes generating a weighting for each topic of the plurality of topics for the query.

14 . The computer program product of claim 13 , wherein generating the weighted query topic embedding includes:

generating a weighted query topic embedding for each topic; and

generating a weighted average query topic embedding from the weighted query topic embedding for each topic.

15 . A computing system comprising:

a memory; and

a processor configured to:

receive a query to be processed by a generative artificial intelligence (AI) model,

extract a query topic from the query,

generate a weighted query topic embedding for the query topic,

generate a plurality of weighted chunk topic embeddings for a plurality of chunks based on a frequency of each term in a particular chunk from a plurality of chunks of a target document and each term's frequency across all of the plurality of chunks,

identify a candidate chunk from the plurality of chunks by determining a similarity between the weighted query topic embedding and the plurality of weighted chunk topic embeddings for the plurality of chunks,

generate a prompt using the query and the candidate chunk, and

provide the prompt to the generative AI model.

16 . The computing system of claim 15 , wherein identifying a candidate chunk includes identifying a predefined number of most similar candidate chunks.

17 . The computing system of claim 15 , wherein the processor is further configured to:

generate the plurality of chunks for a plurality of text portions of the target document;

generate a plurality of chunk topics by extracting a topic for each respective chunk of the plurality of chunks; and

generate a weighting for the topic for each respective chunk of the plurality of chunks.

18 . The computing system of claim 17 , wherein the processor is further configured to:

generate a weighted chunk topic embedding for each topic of a plurality of topics for a respective chunk; and

generate a weighted average chunk topic embedding for the respective chunk from the weighted chunk topic embedding for each topic of the plurality of topics for the respective chunk.

19 . The computing system of claim 15 , wherein extracting the topic includes extracting a plurality of topics for the query.

20 . The computing system of claim 19 , wherein generating the weighting for the topic includes generating a weighting for each topic of the plurality of topics for the query.