Meeting Summary Relevance Thresholds for User-Specific Accuracy

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing meeting summary generation technologies fail to provide summaries tailored to the individual user's level of understanding, often resulting in summaries that are either too detailed or not detailed enough for the user's needs.

Innovation Solution

A system that assigns relevance scores to meeting sentences based on user-specific preferences, using a relevance threshold to generate summaries that match the user's desired level of detail, incorporating machine learning models to enhance relevance scoring.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If a transcript of the meeting is provided to users, then users can understand what occurred during the meeting, but users have to read the entire transcript which is time consuming

Engineering Contradiction:

Improvecompleteness of meeting informationVSAvoidtime to review meeting content

Core Design Contradiction:

Loss of informationVSLoss of time

Solution Approach 1:

The patent extracts only the most relevant sentences from the complete meeting transcript based on relevance scoring. Instead of presenting the entire transcript, the system identifies and extracts key sentences that capture essential meeting information, thereby reducing review time while maintaining information completeness for the user's needs

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the parameter of information presentation by generating summaries at different levels of detail. The system adjusts the amount and depth of information provided based on user characteristics such as expertise level, transforming the fixed transcript into a variable-summary format that optimizes both information retention and time efficiency

Inventive Principle:

Principle #35Parameter changes

2Loss of time

If meeting summary software provides a summary paragraph, then review time is reduced, but the summary is either too detailed or not detailed enough for the user's needs

Engineering Contradiction:

Improvetime to review meeting contentVSAvoidadaptability to user's level of understanding

Core Design Contradiction:

Loss of timeVSAdaptability or versatility

Solution Approach 1:

The patent applies local quality by tailoring the summary content to match the specific user's level of understanding and information needs. Different users receive different summary compositions based on their expertise - experts receive high-level summaries while novices receive more detailed summaries with additional context, making the same meeting summary adaptive to individual user characteristics

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent makes the summary generation dynamic by adjusting the level of detail and content selection based on user profiles. The system dynamically modifies summary characteristics such as sentence selection criteria, detail depth, and information density according to the requesting user's attributes, allowing the summary to adapt rather than remain static

Inventive Principle:

Principle #15Dynamics

3Loss of information

If notes are provided from other users, then summary information is available, but the notes may not match the requesting user's level of understanding

Engineering Contradiction:

Improveavailability of meeting summaryVSAvoidmatch with user's knowledge level

Core Design Contradiction:

Loss of informationVSAdaptability or versatility

Solution Approach 1:

The patent incorporates feedback mechanisms by analyzing user characteristics and using this information to adjust summary generation. The system receives feedback about the user's expertise level, role, and information needs, then uses this feedback to customize the summary content, ensuring it matches the user's knowledge level rather than providing generic notes

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS12475304B2Summarizing meeting content with accuracy control

Publication Date: 2025.11.18 RINGCENTRAL INC

AI summary

A computer-implemented method for generating a meeting summary of a meeting session is provided. The method comprises identifying a plurality of sentences spoken during a meeting session. The method further comprises assigning a relevance score to each sentence in the plurality of sentences. The relevance score represents how important each sentence is to the meeting session. The method further comprises generating a set of relevant sentences from the plurality of sentences based upon the relevance score assigned to each sentence and a relevance threshold. The relevance threshold represents a desired level of understanding of content from the meeting session. The method further comprises generating the meeting summary based on the set of relevant sentences and sending the meeting summary to a device associated with a user.