Question Answering Information Completion Using Machine Reading Comprehension

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional machine reading comprehension question answering systems fail to provide accurate answers due to missing information in open-ended questions, leading to unclear queries and inadequate search results.

Innovation Solution

A feedback-type information completion technique that constructs a training set to detect missing information, generates rhetorical questions using natural language generation models, and combines responses to clarify the original question, enabling a more comprehensive search within a document library.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a knowledge base performs search using open-ended questions, then it can process user queries, but it cannot provide accurate answers due to missing information

Engineering Contradiction:

Improveability to process open-ended questionsVSAvoidaccuracy of answer

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system performs preliminary action by generating rhetorical questions and identifying missing information before conducting the actual search. The feedback-type information completion technique constructs training sets to detect missing information and generates rhetorical questions that clarify what additional information is needed, allowing the system to prepare more precise search queries in advance

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback by using generated rhetorical questions to identify what information is missing from the original query. The feedback loop involves generating rhetorical questions based on the original question, analyzing the differences to determine missing information, and then using this insight to refine the search strategy and improve answer accuracy

Inventive Principle:

Principle #23Feedback

2Measurement precision

If the system generates rhetorical questions to clarify missing information, then answer accuracy improves, but system complexity increases

Engineering Contradiction:

Improveaccuracy of answerVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the question analysis process into distinct components: generating rhetorical questions, comparing them with the original question, identifying missing information, and using this information to refine searches. This segmentation allows each component to be handled by specialized modules, making the overall complex system more manageable and maintainable

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Rhetorical questions serve as an intermediary mechanism between the original user query and the final search execution. Instead of directly searching with the original question, the system uses rhetorical questions as a mediating step to identify missing information, which then informs the actual search process, bridging the gap between incomplete user input and precise information retrieval

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS12112135B2Question answering information completion using machine reading comprehension-based process

Publication Date: 2024.10.08 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

An approach is provided for optimizing a feedback-type question answering process. A training set is constructed to detect missing information of a question. A natural language generation model is trained using the missing information. The natural language generation model is executed to generate a rhetorical question. A response to the rhetorical question is combined with the question to generate an input to a language processor. A new question is generated. The new question is applied to a document library. A final answer is generated.