Deep Question Answering Attribute Scoring via Source Code Influence

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Deep question answering systems often generate inaccurate responses due to missing information from users, as they lack the necessary attributes required for generating the most correct answer, and existing systems fail to identify and prompt users for essential attributes effectively.

Innovation Solution

The system identifies important attributes in its source code by computing influence and importance scores for each attribute, prompting users to provide values for attributes that exceed a predefined threshold, ensuring more accurate responses by accounting for the attributes' impact on confidence scores.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If the deep QA system processes cases with missing information, then it can provide responses without user input, but the accuracy and reliability of responses deteriorates

Engineering Contradiction:

Improvesystem ability to process cases without complete user inputVSAvoidaccuracy of generated responses

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The system performs preliminary analysis of the source code to identify important attributes and their influence scores before processing user cases. This allows the system to pre-determine which attributes are critical for accurate response generation, enabling it to prompt users for missing information in advance rather than discovering deficiencies during case processing

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements a feedback mechanism where it computes influence scores for attributes based on source code analysis, compares these scores against thresholds to determine importance, and then uses this information to guide user interactions. The system prompts users to provide missing important attributes and can re-process cases once additional information is supplied, creating a loop that continuously improves response accuracy

Inventive Principle:

Principle #23Feedback

2Loss of information

If the system prompts users for all possible attributes, then completeness of information improves, but user burden and interaction complexity increases

Engineering Contradiction:

Improvecompleteness of attribute informationVSAvoidcomplexity of user interaction process

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The system applies local quality by computing individual influence scores for each attribute based on its specific impact on response accuracy. Rather than treating all attributes uniformly, the system identifies and prompts only for attributes that exceed a predefined importance threshold, making the information collection process targeted and efficient rather than comprehensive and burdensome

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system changes the parameter of attribute selection from a static, predetermined list to a dynamic, computed set based on influence scores. By computing these scores from source code analysis and comparing against thresholds, the system adaptively determines which attributes to prompt for, reducing the number of prompts while maintaining information completeness

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If the system analyzes source code to identify important attributes, then response accuracy improves, but computational overhead and processing time increases

Engineering Contradiction:

Improveprecision of attribute importance identificationVSAvoidtime for source code analysis and score computation

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs source code analysis and attribute importance identification as a preliminary, one-time setup process before actual case processing. By computing influence scores and determining important attributes in advance, the system avoids repeating this computationally intensive analysis for each case, thereby reducing ongoing processing time while maintaining high precision in attribute identification

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system performs self-service by automatically analyzing its own source code to identify important attributes without requiring external intervention or manual configuration. This automated approach eliminates the need for domain experts to manually specify attribute importance, reducing setup time and computational overhead while maintaining high precision through systematic source code analysis

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS10127284B2Scoring attributes in deep question answering systems based on algorithmic source code influences

Publication Date: 2018.11.13 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Systems and computer program products to perform an operation comprising: identifying a first attribute of a source code in a deep question answering system, computing an influence score for the first attribute based on a rule in the source code used to compute a confidence score for each of a plurality of candidate answers generated by the deep question answering system, computing an importance score for the first attribute based at least in part on the computed influence score, and upon determining that the importance score exceeds a predefined threshold, storing an indication that the first attribute is an important attribute relative to other attributes specified in the source code.