Response Selection Apparatus Quantifying Question-Answer Pair Appropriateness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current dialogue systems face challenges in providing accurate and appropriate responses due to difficulties in calculating the closeness in meaning between user-input questions and available question-answer pairs, leading to suboptimal response selection.
Innovation Solution
A response selecting apparatus that includes a recording part, document searching part, information acquiring part, and ranking part to quantify the appropriateness of question-answer pairs based on search schemes and quantification information, ensuring that responses are selected based on their relevance and appropriateness to the input question.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If question-answer pairs are used for response selection, then the system can provide structured responses, but it is not easy to accurately calculate closeness in meaning between questions, leading to inappropriate responses
Solution Approach 1:
The patent transforms the qualitative concept of 'closeness in meaning' into quantitative parameters by extracting multiple features (lexical overlap, semantic similarity, structural similarity) and calculating composite scores. This parameter transformation enables accurate measurement of question similarity, directly resolving the contradiction between measurement precision and response reliability.
Solution Approach 2:
The patent segments the overall similarity assessment into multiple independent components: lexical similarity, semantic similarity, and structural similarity. Each component is calculated separately using specific algorithms, then combined to form a comprehensive similarity score. This segmentation allows precise measurement of different aspects of question closeness, improving both measurement precision and response appropriateness.
2Adaptability or versatility
If manual utterance conversion rules are created to provide personality to dialogue system, then individual answers can be realized, but it requires cost and time
Solution Approach 1:
The patent enables the dialogue system to automatically generate personalized responses by extracting user-specific information from dialogue history and context, without requiring manual creation of conversion rules. The system self-adapts to user preferences and characteristics, eliminating the time-consuming manual rule creation process while maintaining individual answer capability.
Solution Approach 2:
The patent performs preliminary extraction and storage of user characteristics, preferences, and dialogue patterns during the interaction process. This preliminary action prepares the system to quickly generate personalized responses without requiring time-consuming manual rule creation, resolving the contradiction between adaptability and time loss.
3Ease of manufacture
If neural network is used to automatically generate individual answers from dialogue data, then cost is reduced, but the system lacks explicit control over response selection
Solution Approach 1:
The patent segments the response selection process into distinct controllable stages: question similarity calculation, candidate answer retrieval, and final selection. Each stage uses explicit algorithms and criteria that can be independently adjusted and controlled, maintaining ease of manufacture through automated processing while preserving explicit control over response selection.
Solution Approach 2:
The patent implements feedback mechanisms where the system evaluates the effectiveness of selected responses and adjusts future selections based on this feedback. This explicit control loop allows the system to maintain automated generation capabilities while providing adjustable control over response selection through configurable parameters and evaluation criteria.
Data Source
AI summary
A response selecting apparatus includes a recording part, a document searching part, an information acquiring part, a score calculating part, and a ranking part. The document searching part searches for a question-answer pair from the question-answer pairs recorded in the recording part using the input question as input. The information acquiring part acquires information for quantifying appropriateness of the search-result-question-answer pair with respect to the input question using the input question and the search-result-question-answer pair as input. The score calculating part calculates a score with respect to the input question for each of the search-result-question-answer pairs from a numerical value indicating appropriateness based on a search scheme and a numerical value based on the quantification information. The ranking part selects the search-result-question-answer pairs in descending order of appropriateness indicated by the scores and outputs the answers of the selected question-answer pairs as responses.