Interactive conversation assistance using semantic search and generative AI
A method of generating content in association with an information search and retrieval system. It begins by receiving a query from a user. The query is semantically-searched to identify a context. A conversation history between the user and the system is identified. An enriched query is then generated by associating to the query both the context and at least a portion of the conversation history. The enriched query is then evaluated/processed by a generative-AI. In response, information associated with the enriched query is received from the generative-AI. A response to the query is then generated using the information, e.g., by passing the information back to the user, by modifying (e.g., editing or supplementing) the information to generate modified information and passing the modified information back to the user, or by dismissing the information. If sensitive information is identified in the utterance, it is masked prior to generating the enriched query.
1. A method of generating content in association with an information search and retrieval system that includes a conversational bot, comprising:
configuring an application programming interface (API);
responsive to receipt via the API of a query from a user, the query having been received by the information search and retrieval system:
semantically-searching the query to identify a context associated with the query;
generating an enriched query by associating to the query the context, at least a portion of a conversation history between the user and the conversational bot, and indexed or other website content;
receiving information associated with the enriched query, the information having been generated by evaluation of the enriched query using a transformer-based language model;
generating a response to the query using the information; and
returning the response to the information search and retrieval system.
2. The method as described in claim 1 , wherein receiving information associated with the enriched query comprises:
forwarding the enriched query to an external generative-AI service that includes the transformer-based language model; and
receiving the information from the external generative-AI service.
3. The method as described in claim 1 , further including masking personally-identifiable or other sensitive information identified in the query prior to generating the enriched query.
4. The method as described in claim 1 , wherein generating the response comprises one of:
passing the information back to the user, modifying the information to generate modified information and passing the modified information back to the user, and dismissing the information.
5. The method as described in claim 1 , wherein the conversational history includes a present conversation in which the query arises, and optionally one or more other conversations.
6. The method as described in claim 1 , wherein the portion of the conversation history is condensed prior to being associated with the query.
7. The method as described in claim 1 , further including updating a semantic search model utilizing the information.
8. The method as described in claim 1 , wherein the query is received by a live agent during a live chat session, and the response to the query is provided to the user by the live agent.
9. The method as described in claim 1 , wherein the conversation history is a current conversation dialogue between the user and a chatbot that includes the query.
10. A Software-As-A-Service (SaaS) computing platform associated with an information search and retrieval system, the computing platform comprising:
computer hardware comprising one or more hardware processors; and
computer memory holding computer software executed by the one or more hardware processors, wherein the computer software comprises program code configured as an application programming interface (API) to:
respond to receipt via the API of a query from a user, the query having been received by the information search and retrieval system;
semantically-search the query to identify a context associated with the query;
generate an enriched query by associating to the query the context, at least a portion of a conversation history between the user and a conversational bot associated with the information search and retrieval system, and indexed or other website content;
receive information associated with the enriched query, the information having been generated by evaluation of the enriched query using a transformer-based language model;
generate a response to the query using the information; and
return the response to the information search and retrieval system.