Probabilistic Inclusivity Model for Meeting Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional speech analysis methods fail to effectively evaluate communication inclusivity in virtual meetings, missing opportunities for participants to speak and complete their thoughts, leading to less inclusive discussions and a lack of data insights for users to improve their communication.
Innovation Solution
A system and method that model user communication as a probabilistic interaction, analyzing sequences of speaking states including active speech, silence, and contextual signals to predict communication inclusivity, generating data insights through a graphical user interface.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional speech analysis methods are used to quantify speaking time, then basic meeting metrics can be obtained, but deeper analytical aspects such as speaking opportunities and thought completion cannot be analyzed
Solution Approach 1:
The patent segments speech analysis into multiple distinct components: detecting speaking states (silence, overlap, monologue), analyzing state transitions, and evaluating inclusivity metrics. This segmentation allows the system to achieve deep analytical precision by breaking down the complex task of speech analysis into manageable, specialized sub-tasks that can be processed independently and then integrated.
Solution Approach 2:
The patent introduces an intermediary probabilistic model that bridges raw speech data and inclusivity insights. The model detects speaking states and transitions as intermediate representations, which then feed into the inclusivity evaluation. This intermediary layer enables the system to derive meaningful insights without requiring direct complex analysis of raw speech streams.
2Loss of information
If contextual analysis of user speech considering past communications is implemented, then communication patterns and inclusivity can be evaluated, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary detection of speaking states and transitions during the meeting, storing these intermediate results for later inclusivity evaluation. By pre-processing and segmenting the speech data into meaningful states during the meeting itself, the system reduces the computational burden of later contextual analysis and enables faster generation of inclusivity insights.
3Loss of information
If traditional speech analysis only quantifies speaking time percentage, then simple metrics are provided, but insights about speaking opportunities and thought completion are not generated
Solution Approach 1:
The patent implements feedback by detecting transitions between speaking states and using these transitions to evaluate inclusivity. The system continuously monitors state changes (e.g., from silence to speech, from one speaker to another) and feeds this information back into the inclusivity evaluation process, generating actionable insights about who had opportunities to speak and whether thoughts were completed.
Solution Approach 2:
The patent adds new dimensions to speech analysis by moving beyond simple time percentages to include state transition analysis and inclusivity metrics. Instead of only measuring how long someone spoke, the system analyzes the sequence of speaking states, transitions between speakers, and contextual patterns, thereby adding temporal and contextual dimensions to the analysis.
4Ease of operation
If real-time communication inclusivity analytics are provided, then users can adjust their communication style, but computational processing must be performed during the meeting
Solution Approach 1:
The patent segments the computational workload into real-time components (detecting speaking states and transitions as they occur) and post-meeting components (generating comprehensive inclusivity insights). By dividing the analysis into these temporal segments, the system provides real-time feedback with minimal processing while reserving more intensive computations for after the meeting when performance requirements are relaxed.
Data Source
AI summary
The present disclosure relates to determining communication inclusivity amongst speakers during a user communication. Communication inclusivity is a targeted analysis that collectively evaluates speaking opportunities (provided and taken by users) during a user communication and thought completion during speech associated with the user communication. To derive communication inclusivity, a user communication is modeled as a probabilistic interaction between speakers, where a sequence of speaking states of the user communication is identified and analyzed. Non-limiting examples of speaking states comprise: active user speech; periods of silence; overlapping speakers; icon indication; questions in corresponding chat windows; combination states; other contextual signals; and any combination thereof. With these observed sequences of speaking states, a probability distribution is modeled over transitions between states to predict inclusivity of a user communication. Data insights may be generated (and provided through a graphical user interface), thereby providing analytics that help users understand the concept of communication inclusivity.