Hybrid Question Answering System for Rankable and Probabilistic Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current question answering systems face challenges in reconciling simultaneous ranking and probabilistic criteria, particularly when dealing with rankable and non-rankable questions that require access to both structured and unstructured data sources.
Innovation Solution
A computer-implemented method that identifies lexical, rankable, and probabilistic criteria in a question, generates candidate answers from structured data, and assigns likelihood based on unstructured data, ranking them to produce a hybrid score that reconciles both types of criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the system uses only structured data sources for rankable questions, then ranking precision is improved, but adaptability to diverse query types deteriorates
Solution Approach 1:
The patent combines structured data sources (for rankable questions) and unstructured data sources (for probabilistic questions) into a unified QA system. The system merges two previously separate approaches by integrating structured data processing with unstructured text analysis, allowing it to handle both rankable and non-rankable questions through a single hybrid architecture.
Solution Approach 2:
The system achieves multi-functionality by enabling a single QA platform to handle multiple question types (rankable and non-rankable) that previously required different specialized systems. The hybrid architecture provides universal capability to process diverse query types by selecting appropriate data sources and processing methods based on question characteristics.
2Adaptability or versatility
If the system uses only unstructured data sources for probabilistic questions, then adaptability to diverse query types is improved, but measurement precision deteriorates
Solution Approach 1:
The patent combines structured data sources (for rankable questions) and unstructured data sources (for probabilistic questions) into a unified QA system. The system merges two previously separate approaches by integrating structured data processing with unstructured text analysis, allowing it to handle both rankable and non-rankable questions through a single hybrid architecture.
3Adaptability or versatility
If the system integrates both structured and unstructured data sources, then adaptability to diverse query types is improved, but device complexity increases
Solution Approach 1:
The patent segments the QA system into distinct modules: a structured data processing component, an unstructured data processing component, and a hybrid integration component. This segmentation allows each module to specialize in handling its specific data type while maintaining clear boundaries and interfaces, reducing overall system complexity through modular design.
Solution Approach 2:
The system introduces intermediary components that facilitate communication and coordination between the structured and unstructured data processing modules. These intermediaries manage data flow, coordinate processing steps, and integrate results from both data sources, reducing the complexity of direct integration between heterogeneous systems.
Data Source
AI summary
Providing answers to questions. A LAT, a rankable criterion, and a probabilistic criterion are identified in a received query. A set of candidate answers to the question that are instances of the LAT identified in a first data source is generated. Based on the rankable criterion, the candidate answers are ranked. Each candidate answer is assigned a likelihood that the candidate answer satisfies the probabilistic criterion, based on a statistic of occurrences of terms related to the candidate answer and the probabilistic criterion in text passages from a second data source. One or more candidate answers are selected based on the rank and the likelihood of the candidate answers. In another aspect of the invention, a Boolean criterion is also identified in the query and the set of candidate answers is reduced by applying the Boolean criterion.