Query Graph Coherence Scoring for Multi-Part Virtual Agent Responses
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems lack the ability to effectively extract and utilize relationships between chunks of support documents, particularly in unstructured text and images, to maximize answer coherence for multi-part queries in virtual agent responses.
Innovation Solution
A computer-implemented method that uses natural language processing and relationship extraction to create a query graph, where answer nodes are connected based on relationships between chunks, allowing for the calculation of an answer coherence score to select the most coherent response pathway.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional NLP chunking is used to extract phrases from unstructured text, then text processing is simplified, but the ability to identify relationships between chunks is lost
Solution Approach 1:
The patent embeds relationship extraction functionality within the existing NLP chunking pipeline. Chunk relationships are nested within the chunk data structure, allowing the system to maintain simple text processing while simultaneously capturing relationship information between chunks through hierarchical data organization.
Solution Approach 2:
The patent introduces an intermediary relationship extraction module that sits between traditional NLP chunking and downstream processing. This mediator captures relationships between chunks without disrupting the simplicity of the original chunking process, enabling relationship information to be extracted as a byproduct of normal text processing.
2Measurement precision
If multiple candidate answers are considered for each part of a multi-part query, then answer accuracy is improved, but answer coherence becomes difficult to maintain
Solution Approach 1:
The patent implements a feedback mechanism where the coherence evaluation of previously selected answers influences the selection of subsequent answers. The system continuously monitors how well new candidate answers integrate with the existing answer set and adjusts selections to maintain overall coherence, creating a self-correcting process that balances accuracy and consistency.
Solution Approach 2:
The patent adds a coherence dimension to the traditional accuracy-based answer selection process. By evaluating answers not only on their individual accuracy but also on their coherence with the overall answer set, the system transforms a single-dimensional selection problem into a multi-dimensional optimization that simultaneously considers both accuracy and coherence.
3Reliability
If relationships between chunks are extracted and utilized, then query answer coherence is maximized, but system complexity increases
Solution Approach 1:
The patent divides the complex task of relationship extraction and coherence evaluation into distinct modular components: chunk relationship extraction, candidate answer generation, coherence evaluation, and answer selection. Each module handles a specific aspect of the process, making the overall system more manageable and maintainable while achieving high answer coherence.
Data Source
AI summary
Systems, methods and/or computer program products maximizing answer coherence of virtual agents responding to multi-part user queries, using relationships among extracted chunks corresponding to support document(s). Relationships are extracted by mining entities from text and images of support document(s) and mapping entities to an otology. Induced subgraphs for chunks, based on extracted entities are created and compared. Depending on amounts of overlap and/or relationships between subgraphs, relationships among chunks are identified. Chunks relationships can be complementary, contained within one another, instances of one another, or not related at all. For a given multi-part query, a query answer graph is created comprising answer nodes corresponding to potential answer candidates associated with each chunk. Using the edges between nodes and relationships identified between chunks, coherence scores are calculated along answer pathways, whereby answer pathways with the highest coherence score are selected to respond to the query.