Assertion-Based Question Answering with Semi-Structured Output
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional question answering systems provide answers that are either too long, burdening users with excessive reading, or too short, failing to convey complete semantic meaning, thereby compromising user experience.
Innovation Solution
An assertion-based question answering method that determines a semi-structured answer with a predetermined structure, providing richer semantic content and more concise expression by analyzing the question and relevant passage, using a computing device to generate and output assertion answers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional question answering systems provide detailed long answers, then completeness of information is improved, but user reading burden increases
Solution Approach 1:
The patent segments the answer into two distinct parts: a concise summary answer and detailed supporting information. The summary provides the core response quickly, while the detailed information is available but separated, allowing users to access completeness only when needed. This segmentation resolves the contradiction by making the default view concise while preserving full information availability.
Solution Approach 2:
The system dynamically adjusts the presentation of information based on user interaction. Initially, only the concise summary is displayed to reduce reading burden. When users show interest or need more detail, the system then reveals the complete detailed information. This dynamic adaptation allows the same system to serve both concise and comprehensive information needs without forcing users to always read long answers.
2Ease of operation
If traditional question answering systems provide concise short answers, then user reading burden is reduced, but semantic completeness deteriorates
Solution Approach 1:
The answer is divided into a concise summary portion and a detailed supporting portion. The summary provides immediate semantic value in an easily digestible format, while the detailed portion contains the complete semantic information for users who need it. This segmentation ensures that concise presentation does not sacrifice semantic completeness, as both versions are available.
Solution Approach 2:
The system performs preliminary processing to extract and present the most essential semantic content in a concise summary format before the user even reads the full answer. This preliminary extraction of key semantic elements allows users to quickly grasp the core meaning without reading the entire detailed response, reducing reading burden while preserving semantic access.
3Adaptability or versatility
If open information extraction is used without predefined dictionaries, then adaptability to different questions is improved, but requirement for training data increases
Solution Approach 1:
The patent employs a universal question answering framework that can handle multiple types of questions and domains without requiring separate training for each. The system uses general-purpose information extraction and reasoning capabilities that adapt to different questions through the structure of the question-answer pair itself rather than through domain-specific training data. This universality reduces the need for extensive training data while maintaining high adaptability.
Solution Approach 2:
The system uses the question and answer context itself to guide the extraction and reasoning process, rather than relying heavily on pre-trained models. The question-answer pair structure provides the necessary guidance for information extraction, allowing the system to serve itself by using the input data to determine what information is relevant and how to structure the response, reducing dependence on external training data.
Data Source
AI summary
In embodiments of the present disclosure, there is provided an assertion-based question answering manner. After a question and the related passage are obtained, an assertion answer to the question is determined based on content of the passage, and the assertion answer has a predetermined structure and represents a complete semantic meaning. Then, the assertion answer to the question may be outputted to the user. In the embodiments of the present disclosure, the question and the relevant passage are used as input, and a semi-structured assertion answer is output. The assertion answer according to embodiments of the present disclosure can provide richer semantic content than the traditional short answer, and provide a more concise expression than the traditional long answer, thereby ensuring accuracy of the answer while improving the user experience.