Controllable LLM Summarization for Abstraction and Key Data Retention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems lack precision and user configurability in controlling the level of abstraction and brevity when generating abstractive summaries, and fail to preserve key data during summarization.
Innovation Solution
A system that enables users to define a target level of abstraction, employs iterative prompt engineering to refine summaries, and integrates external data to maintain data integrity during summarization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If abstractive summarization is generated without controlled parameters, then the summarization process is simple and fast, but the level of abstraction and brevity cannot be precisely controlled
Solution Approach 1:
The patent applies parameter changes by introducing controllable parameters such as target length, maximum length reduction percentage, and minimum length to precisely control the level of abstraction and brevity in summarization outputs, transforming the uncontrolled generative process into a parameter-driven controlled process
Solution Approach 2:
The system dynamically adjusts summarization parameters based on user inputs and content characteristics, allowing flexible control over abstraction levels and summary lengths in real-time, making the system adaptable to different summarization needs
2Reliability
If iterative refinement is applied to improve summary quality, then the abstraction capabilities and data retention are improved, but the processing time and computational resources increase
Solution Approach 1:
The patent implements feedback mechanisms where the system evaluates generated summaries against target parameters and iteratively refines them by providing feedback to the generation process, ensuring data retention and quality while controlling processing time through structured iteration limits
Solution Approach 2:
The system performs preliminary actions by pre-processing content to identify key information and structure before iterative summarization begins, reducing the number of iterations needed and thus decreasing overall processing time while maintaining data retention
3Loss of information
If key data preservation is enforced during summarization, then information integrity is maintained, but the brevity and conciseness of the summary are reduced
Solution Approach 1:
The patent applies local quality by differentiating between critical information that must be preserved and non-critical information that can be condensed or removed, allowing the summary to maintain key data integrity while achieving desired brevity through selective retention
Data Source
AI summary
A method of generating summaries of content items using one or more large language models (LLMs) is disclosed. A first content item is identified. The first content item includes a set of sub-content items. A level of abstraction is determined for the content item. A prompt is automatically engineered for providing to the one or more LLMs. The prompt includes a reference to the first content item and the level of the abstraction for the first content item. A response to the prompt is received from the LLM. The response includes a second content item. The second content item includes a representation of the first content item that is generated by the LLM. The representation omits or simplifies one or more of the set of sub-content items based on the level of abstraction. The representation is used to control an output that is communicated to a target device.