Knowledge graph element organization method oriented to time sequence tangent plane

Abstract

The invention discloses a knowledge graph element organization method oriented to a time sequence tangent plane. The method comprises the steps of defining a maximum effective time range of a system;organizing entities and relationships, and constructing a time sequence map; obtaining a key point set; and obtaining the tangent plane of the time sequence map, and determining the entity, the relationship and all change processes of the attributes of the entity and the relationship in the effective time range, or the attribute values and states of the entity and the relationship in the key points. The method can reflect the storage time ranges of the vertexes and the edges, the attributes of the vertexes and the edges in different time ranges and the change conditions of the vertexes and theedges, and unified version division is not needed; the state of the atlas at any moment within the effective time range of the atlas can be obtained, all states of a certain graph element are directly obtained, and compared with an incremental graph method, calculation steps are greatly reduced.

Description

technical field [0001] The invention relates to database technology, in particular to a method for organizing knowledge map elements oriented to time series sections. Background technique [0002] The knowledge graph is a semantic network that includes entities and relationships. The knowledge graph technology can be used to model entities, relationships and their attributes. The usual knowledge graph is static, and only maintains information about a certain state of entities and relationships, that is, it only reflects the status and attributes of entities and their related relationships at a point in time. However, the real world is in a process of constant movement and development. If a static knowledge map is used to express the evolution process of the attributes, states, and relationships of things over time, there will be the following shortcomings: the attributes of things are changing In the process, for example, a person may change their nationality, and the stati...

AI Technical Summary

Problems solved by technology

However, the real world is in a process of constant movement and development. If a static knowledge map is used to express the evolution process of the attributes, states, and relationships of things over time, there will be the following shortcomings: the attributes of things are changing In the process, for example, a person may change their nationality, and the static knowledge map can only maintain one version of information, and cannot reflect the change of the nationality attribute of the person; the relationship between things is also constantly changing, for example, two people can After entering into a marriage relationship, you may also get divorced for various reasons after marriage, and you may remarry or remarry after divorce. Static knowledge graphs generally only show the relationship between husband and wife between two people, and do not reflect the duration of the marriage relationship, nor can it reflect well. Interruption of marital relationship between two people due to divorce or remarriage

[0003] At present, there are the following graph-related technologies that can reflect changes in entities or relationships over time, for example: 1) Document 1 (A.G.Labouseur, P.W.Olsen, and J.-H.Hwang. Scalable and robust management of dynamic graph data.In BD3@VLDB, 2013.) By storing all snapshots of all graphs, that is, storing all versions of graphs, to reflect changes in entities or relationships over time, but this method has large data redundancy, and the complexity of managing and retrieving a certain snapshot is cumbersome

2) Literature 2 (Tasnim, Mayesha&Collarana, Diego&Graux, Damien&Orlandi, Fabrizio&Vidal, Maria-Esther.(2019). Summarizing Entity Temporal Evolution in Knowledge Graphs.10.1145 / 3308560.3316521) Divide the graph by year, that is, generate a data version every year. Store the attributes of entities in different years. The minimum time granularity for recording entity attribute changes in this way is one year, and changes lower than one year cannot be reflected; it is difficult to deal with entity attribute changes that are not divided by year

3) Document 3 (Khurana U, Deshpande A. Efficient Snapshot Retrieval over Historical Graph Data [J]. 2012) adopts incremental graph technology, and a certain reference time of the graph (denoted as t 0 ) state as a reference map, since the reference time t 0 All operations that occur on the subsequent graph are recorded, when it is necessary to obtain t 1 When the state of the time graph is used, it is necessary to start from the state of the reference graph to transfer t 0 to t 1 All operations that occur at any time are executed once, and the graph obtained after that is t 1 The graph of the time state, this method requires a large amount of calculation

Images