Query Clarification System Using Dependency Tree Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ambiguous user queries in search engines often result in mixed and irrelevant search results, leading to reduced retrieval efficiency as users must sift through numerous webpages to find relevant information.
Innovation Solution
A method and device for clarifying questions based on deep question and answer technology, which involves receiving a query sentence, recalling and analyzing answer titles and history query sentences to generate dependency trees, clustering similar queries, and generalizing them to create candidate and clarified questions for more precise search results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If search engine returns abundant internet resources for a query, then the quantity of information is improved, but the precision and relevance of results deteriorate when query intention is ambiguous
Solution Approach 1:
The patent segments the search process into two stages: first generating multiple candidate questions from the ambiguous query, then allowing users to select or refine their intended meaning. This segmentation transforms a single imprecise search into multiple targeted searches, resolving the contradiction between quantity and precision of results.
Solution Approach 2:
The patent performs preliminary action by generating candidate clarified questions before the actual search execution. By pre-processing the ambiguous query into multiple possible interpretations and presenting them to users for selection, the system ensures that the subsequent search operates on precise, user-confirmed intentions, thereby improving result precision while maintaining adequate quantity.
2Loss of information
If search engine provides comprehensive results for ambiguous queries, then information coverage is improved, but user time to find relevant information increases
Solution Approach 1:
The system performs preliminary action by generating candidate clarified questions before search execution. This pre-processing step organizes potential search directions in advance, allowing users to quickly select their intended meaning without sifting through irrelevant results, thereby reducing time loss while maintaining information coverage.
Solution Approach 2:
The patent implements feedback by presenting generated candidate questions to users for selection or correction. This feedback loop ensures that the search system aligns with user intent, reducing the time users spend searching through irrelevant information while maintaining comprehensive coverage of relevant topics.
3Ease of operation
If search engine returns mixed results without clarification, then ease of operation is improved, but retrieval efficiency deteriorates
Solution Approach 1:
The patent performs preliminary action by automatically generating candidate clarified questions before the user initiates the search. This pre-processing step maintains ease of operation as users simply select from generated options, while simultaneously improving retrieval efficiency by ensuring searches are based on clarified, precise intentions rather than ambiguous queries.
Data Source
AI summary
The present disclosure discloses a method and a device for clarifying questions based on deep question and answer. The method includes: receiving a query sentence; recalling corresponding answer titles and/or history query sentences according to the query sentence; analyzing the answer titles and/or the history query sentences to obtain corresponding dependency trees; clustering the answer titles and/or the history query sentences according to the dependency trees, to generate at least one cluster of questions; generalizing the at least one cluster of questions to generate candidate and clarified questions; and displaying the candidate and clarified questions.