Systems and methods for data-driven query and response generation

A system, comprising:

• one or more processors;
• and
• a non-transitory memory storing instructions that, when executed, cause the processor to obtaining a first query;
• identify, by the one or more processors, a plurality of documents relevant to the first query;
• generate, using a machine learning model, one or more additional queries that are semantically similar to the first query, wherein generating the additional queries includes vectorizing text passages using a large language model, calculating semantic similarity for the vectorized text passages, and using an artificial intelligence generative model to obtain a summary for the one or more additional queries, and wherein the plurality of documents includes a first set of documents selected in response to the first query and a second set of documents selected in response to the one or more additional queries;
• present, via a user device, a user interface listing identifiers of the plurality of documents;
• in response to receiving a user selection of one or more documents from the plurality of documents:
• for each selected document:
• generate by the one or more processors, a plurality of responses based on the document, using the generative model;
• generate, using the generative model, at least one link to a source portion of the document used by the generative model to generate at least a portion of the plurality of responses;
• and
• receive feedback data from the user device indicating a user selection of one or more of the plurality of responses, wherein the feedback is based at least in part on the one or more of the plurality of responses;
• associate the at least one response with a set of queries, including the first query, in a memory;
• and
• in response to receiving a second query in the set of queries, select a preferred response associated with the set of queries, based on the feedback, and display the preferred response via the user interface.

A non-transitory computer-readable medium having instructions stored thereon, wherein the instructions, when executed by at least one processor, cause at least one device to perform operations comprising:

• obtaining a first query;
• identifying, by the one or more processors, a plurality of documents relevant to the first query;
• generating, using a machine learning model, one or more additional queries that are semantically similar to the first query, wherein generating the additional queries includes vectorizing text passages using a large language model, calculating semantic similarity for the vectorized text passages, and using an artificial intelligence generative model to obtain a summary for the one or more additional queries, and wherein the plurality of documents includes a first set of documents selected in response to the first query and a second set of documents selected in response to the one or more additional queries;
• presenting, via a user device, a user interface listing identifiers of the plurality of documents;
• in response to receiving a user selection of one or more documents from the plurality of documents:
• for each selected document:
• generating by the one or more processors, a plurality of responses based on the document, using the generative model;
• generating, using the generative model, at least one link to a source portion of the document used by the generative model to generate at least a portion of the plurality of responses;
• and
• receiving feedback data from the user device indicating a user selection of one or more of the plurality of responses, wherein the feedback is based at least in part on the one or more of the plurality of responses;
• associating the at least one response with a set of queries, including the first query, in a memory;
• and
• in response to receiving a second query in the set of queries, selecting a preferred response associated with the set of queries, based on the feedback, and displaying the preferred response via the user interface.