Graph Embeddings for Conversational Flow Analysis and Privacy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Developing natural language interfaces is complex due to the difficulty in determining conversation quality, and existing technologies lack tools for analyzing conversational data effectively, particularly in handling conversational outcomes and ensuring privacy in shared datasets.
Innovation Solution
A conversational analytics toolset that generates summary statistics and graphical representations of conversational data, uses graph embeddings to identify similar conversations, and ensures k-anonymity to protect user privacy, allowing for improved conversation steering and compliance with regulatory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conversational data is collected and analyzed to improve natural language interface quality, then conversation quality and user interaction improve, but user privacy and personally identifying information are exposed
Solution Approach 1:
The patent extracts and removes personally identifying information (PII) from conversational data before analysis. The system separates sensitive user information from the conversational content, allowing the interface to learn from conversation patterns without exposing individual user identities. This extraction process enables quality improvement while protecting privacy.
Solution Approach 2:
The patent introduces an intermediary processing layer that anonymizes conversational data before it is used for training and analysis. This intermediary system transforms raw conversational data into anonymized representations that preserve linguistic patterns and conversation quality metrics while removing identifiable user information, thus mediating between quality improvement and privacy protection.
2Adaptability or versatility
If conversational datasets are shared for research and development purposes, then collaboration and innovation improve, but regulatory compliance and privacy protection become more difficult
Solution Approach 1:
The patent applies anonymization and privacy protection measures before sharing conversational datasets. By performing preliminary processing to remove PII and ensure compliance with regulations like GDPR and HIPAA, the system enables safe sharing of data for research and development purposes. This preliminary action ensures that collaboration can proceed without compromising regulatory compliance.
Solution Approach 2:
The patent transforms conversational data by changing its parameters - specifically converting identifiable information into anonymized representations. This parameter transformation maintains the utility of the data for research and development while ensuring that shared datasets meet regulatory requirements. The system adjusts data parameters to balance collaboration needs with compliance obligations.
Data Source
AI summary
Systems and methods develop a natural language interface. Conversational data including user utterances is received for a plurality of conversations from a natural language interface. Each of the conversations is classified to determine intents for each user utterance, and for each of the conversations, a control flow diagram showing the intents and sequential flow of the conversation is generated. Each of the control flow diagrams is processed to generate a graph embedding representative of the conversation. A previous conversation that is similar to the current conversation is identified from a previous graph embedding that is nearest to a current graph embedding of a most recent utterance in a current conversation. A previous outcome of the previous conversation is used to predict an outcome of the current conversation, which, when not positive, may control response outputs of the natural language interface to steer the current conversation towards a positive result.