Inter-thread Anaphora Resolution for QA Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Anaphora resolution in threaded discussions, such as online forums, poses a challenge for Question Answering (QA) systems as existing technologies struggle to accurately identify and resolve anaphors across posts, leading to ambiguity and poor integration of referential elements into QA systems.
Innovation Solution
An approach that analyzes posts in a threaded discussion, identifies anaphors in child posts referencing terms from parent posts, and resolves these anaphors by associating them with their parent terms, storing the resolved information for use by QA systems to enhance the understanding and processing of language.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing anaphora resolution technologies are used for threaded discussions, then the system can process individual posts, but it fails to accurately identify and resolve anaphors across multiple posts, leading to ambiguity
Solution Approach 1:
The system segments the threaded discussion into individual posts while maintaining hierarchical relationships (parent-child connections). Each post is processed separately to identify anaphors, but the segmentation preserves the structural context needed for cross-post resolution. This allows the system to manage complexity by breaking down the overall task while maintaining the necessary contextual connections for accurate anaphor resolution.
Solution Approach 2:
The system transitions from traditional intra-paragraph anaphora resolution to inter-post anaphora resolution by adding a new dimension of analysis across multiple posts in a thread. This dimensional expansion allows the system to track referential elements vertically through the thread structure, resolving anaphors that span multiple posts while maintaining computational feasibility through structured approach.
2Reliability
If the system expands analysis to cover multiple posts for anaphor resolution, then anaphor resolution accuracy improves, but processing complexity and computational resources increase
Solution Approach 1:
The system performs preliminary processing of each post to identify and extract potential referential elements before conducting full anaphora resolution across the thread. This preliminary action prepares the data in advance, organizing terms and potential anaphors in a structured format that facilitates efficient cross-post matching and reduces the complexity of the subsequent resolution process.
Solution Approach 2:
The system introduces an intermediary layer that maps anaphors to their corresponding terms across posts. This intermediary mechanism acts as a bridge between scattered referential elements in different posts, enabling reliable reference identification without requiring direct complex analysis of all possible post combinations, thus managing processing complexity.
3Loss of information
If traditional QA systems process forum content without inter-post anaphora resolution, then processing speed is maintained, but the system fails to capture interconnected meanings across posts
Solution Approach 1:
The system merges the processing of multiple posts by establishing and utilizing parent-child relationships between posts in a thread. This combining approach allows the system to capture interconnected meanings and referential information across posts while maintaining a unified processing framework that preserves content processing efficiency through structured integration rather than isolated analysis.
Solution Approach 2:
The system maintains continuous tracking of referential elements throughout the thread, ensuring that anaphor resolution information is preserved and utilized across all posts. This continuous action prevents loss of referential information by maintaining an ongoing context awareness throughout the processing of the entire thread, while the systematic approach keeps processing efficiency intact.
Data Source
AI summary
An approach is provided to resolve anaphors between posts, or threads, in a threaded discussion, for example an online forum. The approach analyzes a number of posts that are included in threads of an online forum. During the analysis, the approach identifies terms in parent posts, detects anaphors in child posts that reference the terms in the parent posts, and resolves the anaphor found in the child post with the term. The parent post with the identified term and the child post with the resolved anaphor are then stored in the memory for use by information handling systems, such as question answering (QA) systems.