Generative Answers With Citations for Cloud Document Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of cloud-based content management platforms face inefficiencies in locating relevant documents across disparate sources with different formats and permissions, leading to time-consuming searches and unnecessary resource consumption.
Innovation Solution
Implementing a generative machine learning model (MLM) that processes documents for query embeddings, provides personalized prompts, and anticipates user interest to generate answers with citations to source documents, reducing the need for manual searching and optimizing resource use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If users manually search for information across disparate documents and sources, then they can access the information they need, but it takes a significant amount of time and effort
Solution Approach 1:
The generative AI system performs the information gathering and synthesis work automatically. Users simply ask their question and the system autonomously searches through documents, retrieves relevant information, and generates answers without requiring manual user search through disparate sources
Solution Approach 2:
The generative AI model acts as an intermediary between the user's information need and the scattered documents in the system. It translates user queries into search operations, processes information from multiple sources, and presents synthesized answers, thereby mediating the information retrieval process
2Loss of information
If users search through documents in different locations and formats, then they can find relevant information, but the process is complex and time-consuming
Solution Approach 1:
The generative AI system provides a universal interface that handles multiple document formats and locations through a single query mechanism. Users don't need to navigate different search interfaces for different document types; the system uniformly processes all document formats and returns synthesized answers
Solution Approach 2:
The system automatically handles the complexity of searching across different locations and formats. The generative AI model manages document discovery, retrieval, and synthesis without requiring users to manually navigate complex search criteria or document structures
3Loss of information
If users access all documents in cloud storage, then they have complete access to information, but unnecessary resource consumption occurs
Solution Approach 1:
The system extracts only the necessary information needed to answer the user's specific question. Instead of loading or processing all documents in cloud storage, the generative AI model identifies and processes only the relevant portions of documents that directly contribute to answering the query, thereby reducing resource consumption
Solution Approach 2:
The system performs partial action by processing only the subset of documents required for the specific query rather than accessing all available documents. This selective processing approach maintains information availability for the user's needs while minimizing unnecessary resource consumption
Data Source
AI summary
A method is disclosed that includes obtaining a generative machine learning model (MLM) prompt that prompt includes an indication of a user request to generate content based on one or more of a plurality of documents stored in a cloud-based content management platform, selecting a subset of the plurality of documents based on the generative MLM prompt and a first query embedding corresponding to the generative MLM prompt, inputting the generative MLM prompt and the subset of the plurality of documents into a first generative MLM, and generating, using the first generative MLM, a response, wherein the response comprises content generated by the first generative MLM, and one or more citations to one or more documents of the subset.