Temporal Knowledge Graph Generation with Time Interval Attributes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current knowledge graphs lack precision in expressing time information, only supporting discrete time states and failing to accurately represent the validity period of entity relationships, which limits their ability to perform time-based knowledge calculations.
Innovation Solution
A method and apparatus for generating a temporal knowledge graph by extracting and normalizing entity pairs, relationships, and target time intervals, using multivariate data extraction and fusion techniques to unify and accurately express the valid period of entity relationships, thereby enhancing the precision of time information representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If general knowledge triplet format is used, then the knowledge graph can be constructed simply, but the precision of time information expression deteriorates
Solution Approach 1:
The patent segments the time information into distinct components: start time, end time, and time interval. This segmentation allows the knowledge graph to maintain simple construction through standardized triplets while achieving precise time expression by explicitly representing temporal boundaries and durations.
Solution Approach 2:
The patent adds a temporal dimension to the traditional knowledge graph structure by introducing time-related attributes (start time, end time, time interval) to entity relationships. This dimensional expansion enables precise time information expression without complicating the fundamental triplet structure, as time becomes an additional attribute layer rather than a structural overhaul.
2Device complexity
If discrete time states are used, then the knowledge graph structure remains simple, but the ability to perform time-based calculations deteriorates
Solution Approach 1:
The patent transforms discrete time states into continuous time parameters by introducing start time, end time, and time interval attributes. This parameter transformation enables time-based calculations (such as duration computation, temporal overlap detection, and chronological ordering) while maintaining relatively simple graph structure through standardized attribute representation.
Solution Approach 2:
The patent introduces time interval as an intermediary parameter that mediates between discrete time points and calculation requirements. The time interval attribute serves as a computational bridge, enabling various time-based operations (duration calculation, temporal relationship analysis) without requiring complex temporal reasoning structures in the knowledge graph.
3Loss of information
If time information is not explicitly represented, then the data format remains simple, but the validity period of entity relationships cannot be accurately expressed
Solution Approach 1:
The patent applies preliminary action by pre-defining standardized time attribute formats (start time, end time, time interval) that can be directly populated during knowledge graph construction. This preliminary structuring ensures complete time information representation without increasing data format complexity, as the temporal framework is established in advance and systematically filled during data integration.
Data Source
AI summary
A method and apparatus for generating a temporal knowledge graph, a device and a medium. An embodiment comprises: acquiring corpus including time information; performing multivariate data extraction on the corpus, multivariate data including an entity pair, an entity relationship and a target time interval of the entity relationship, the target time interval being used to indicate a valid period of the entity relationship; and generating a temporal knowledge graph based on the entity pair, the entity relationship and the target time interval of the entity relationship.