Topological Relationship Graph for Knowledge Graph Link Completion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge lies in accurately acquiring entity webpage links from knowledge graphs, as existing methods often result in incomplete data due to missing value links during webpage data extraction, leading to incomplete Subject-Predicate-Object (SPO) triples, which hinders the effectiveness of knowledge-based services.
Innovation Solution
A technique utilizing topological relationship graphs to determine weight values of candidate entity webpages, leveraging forward and backward topological relationship graphs to establish association relationships and calculate weights through random walking, thereby improving the accuracy of finding value links and supplementing missing links in the knowledge graph.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional webpage data extraction methods are used, then the extraction process is simple and fast, but the completeness of knowledge graph data deteriorates due to missing value links
Solution Approach 1:
The patent introduces a topological relationship graph as an intermediary structure between webpage extraction and knowledge graph construction. This graph captures the structural relationships among webpages, enabling the system to infer missing value links by analyzing topological patterns rather than relying solely on direct extraction, thus improving data completeness without fundamentally changing the extraction process
Solution Approach 2:
The system implements a feedback mechanism where the topological relationship graph is continuously updated based on extracted data, and this updated graph is used to guide further data extraction and link completion. The weight values calculated from the graph provide feedback on the reliability of inferred links, allowing the system to iteratively improve knowledge graph completeness
2Measurement precision
If more candidate entity webpages are considered to improve accuracy, then the accuracy of finding value links improves, but the computational complexity increases
Solution Approach 1:
The patent transforms the complex problem of evaluating many candidate webpages into a parameter-based solution by calculating weight values that quantify the association strength between webpages. These weight values serve as parameters that rank candidates, allowing the system to efficiently identify the most relevant value links without exhaustively analyzing all possible candidates
Solution Approach 2:
The system segments the set of candidate entity webpages into different levels based on their weight values and topological relationships. By processing candidates in segments rather than as a single large set, the computational complexity is reduced while still maintaining high accuracy through systematic evaluation of the most promising candidates first
3Manufacturing precision
If topological relationship graphs with weight values are used, then the accuracy of knowledge graph construction improves, but the processing time and computational resources increase
Solution Approach 1:
The patent applies partial action by calculating weight values only for the most relevant candidate webpages rather than all possible candidates. The system identifies a subset of high-priority candidates based on initial topological analysis and focuses computational resources on these, achieving high quality knowledge graph construction without the full computational cost of exhaustive processing
Solution Approach 2:
The system performs preliminary action by pre-calculating topological relationships and weight values for candidate webpages before the actual knowledge graph construction. This preliminary processing organizes the data in a way that accelerates the main construction process, reducing the time required during the critical knowledge graph generation phase
Data Source
AI summary
The present disclosure discloses a chatbot with a stance taking. Opinions may be extracted from candidate replies during a processing of conversation, and the extracted opinions may be compared with the stance taken by the chatbot so as to perform selection on the candidate replies and remove the candidate replies which have conflicts with the stance taken by the chatbot. With such technical solutions, the stance taken by the chatbot may be exhibited.