Meeting Transcript Post-Processing for Readable Summaries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems are inadequate in generating accurate and readable summaries from meeting transcripts due to challenges such as heterogeneous semantic styles, lengthy transcripts, and errors in transcription, which affect the efficiency and accuracy of meeting summarization.
Innovation Solution
A computing system that post-processes meeting transcripts using machine learning models to modify punctuation, grammar, and formatting, and attributes content to specific participants, enabling the generation of improved summaries and templates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If abstractive summarization is used to generate more readable summaries, then readability is improved, but system complexity increases due to complicated multi-stage machine learning pipelines
Solution Approach 1:
The patent segments the summarization task into distinct components: extractive summarization for identifying key sentences and abstractive summarization for generating paraphrased content. This segmentation allows each component to be optimized independently, reducing overall system complexity while maintaining readability improvements.
Solution Approach 2:
The patent merges extractive and abstractive summarization approaches into a unified system that leverages the strengths of both methods. By combining sentence extraction with targeted paraphrasing, the system achieves improved readability without requiring entirely complex multi-stage pipelines.
2Manufacturing precision
If multi-stage machine learning pipelines are used for abstractive summarization, then summarization quality is improved, but training optimization becomes intractable
Solution Approach 1:
The patent performs preliminary extraction of key sentences and identification of important content before applying abstractive summarization. This preliminary action reduces the input size and complexity for the abstractive model, making training optimization more tractable while maintaining summarization quality.
Solution Approach 2:
The patent applies abstractive summarization selectively to specific portions of the transcript rather than the entire text. By focusing computational resources on key sentences and important segments, the system achieves high summarization quality with reduced training complexity compared to processing entire transcripts.
3Manufacturing precision
If human involvement is increased in template generation, then summary accuracy is improved, but scalability decreases
Solution Approach 1:
The patent implements self-service mechanisms where the system automatically learns from meeting transcripts and improves its summarization accuracy over time without requiring continuous human intervention. The model performs self-training and adaptation, maintaining high accuracy while enabling scalable deployment.
Solution Approach 2:
The patent incorporates feedback loops where summarization outputs are evaluated and used to refine the model. This automated feedback mechanism allows the system to improve accuracy through learning from its own performance data, eliminating the need for extensive human involvement while maintaining scalability.
Data Source
AI summary
A transcription of audio speech included in electronic content associated with a meeting is created by an ASR model trained on speech-to-text data. The transcription is post-processed by modifying text included in the transcription, for example, by modifying punctuation, grammar, or formatting introduced by the ASR model and by changing or omitting one or more words that were included in both the audio speech and the transcription. After the transcription is post-processed, output based on the post-processed transcription is generated in the form of a meeting summary and/or template.