Generative AI Content Search with Hybrid Result Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content searching systems using generative AI models are slow and expensive in live production environments, and traditional methods rely on keyword matching rather than understanding the meaning of queries, lacking transparency and efficiency.
Innovation Solution
A system utilizing both generative AI and traditional search sub-systems to generate and aggregate search results, leveraging the 'hallucination' feature of generative AI to create confabulated items that match query meaning, and employing a vector database for similarity retrieval, with transparency features for debugging and compliance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If generative AI models are used for content searching, then the system can understand the meaning of queries and generate relevant results, but the processing speed becomes slow and the cost increases
Solution Approach 1:
The search system is divided into two independent subsystems: a traditional search subsystem that handles fast keyword-based retrieval, and a generative AI subsystem that handles semantic understanding and confabulated result generation. Each subsystem operates independently and processes different aspects of the search query, allowing the system to maintain both speed and accuracy without requiring one subsystem to compensate for the other's weaknesses.
Solution Approach 2:
The patent combines the results from both the traditional search subsystem and the generative AI subsystem into a unified search result presentation. The aggregation layer merges fast keyword matches with semantically relevant confabulated results, providing a comprehensive answer that leverages the strengths of both approaches while mitigating their individual weaknesses.
2Adaptability or versatility
If generative AI models are used for content searching, then the system can generate confabulated search results, but the expense increases in live production environments
Solution Approach 1:
The system applies generative AI processing selectively rather than to all search queries. The traditional search subsystem handles routine keyword-based searches efficiently, while the generative AI subsystem is invoked for queries requiring deeper semantic understanding or when traditional search yields insufficient results. This partial application of the expensive generative process reduces overall computational costs while maintaining versatility where needed.
3Productivity
If traditional keyword matching is used for search, then the processing is fast and cheap, but the system cannot understand the meaning of queries
Solution Approach 1:
The patent introduces an intermediary layer that translates between keyword-based search results and semantically meaningful confabulated results. The generative AI model acts as a mediator that takes simple keyword queries and transforms them into comprehensive, contextually relevant search results by generating confabulated content that captures the intended meaning behind the keywords.
4Measurement precision
If human-reviewed items are used to map representative items to queries, then the search results are accurate, but the process is slow and expensive
Solution Approach 1:
The system enables the generative AI model to self-generate confabulated search results without requiring manual human review for each query. The model learns from training data to autonomously produce accurate and relevant search results, eliminating the time-consuming and expensive human-in-the-loop process while maintaining high relevance through the model's semantic understanding capabilities.
Data Source
AI summary
Methods and systems provide content searching and retrieval using generative artificial intelligence (AI) Models. The system is configured to receive a user search for content, media or item listings. The user search is provided to a generative AI based search sub-system and to a traditional search sub-system. A first search result listing is generated by the generative AI based subsystem, and a second search result listing is generated by the traditional search sub-system. The first search result listing and the second search result listing are aggregated together and provided for display to a user client device.