Active Temporal Knowledge Graph Alignment

Active Temporal Knowledge Graph Alignment

Jie Zhou, Weixin Zeng, Hao Xu, Xiang Zhao
Copyright: © 2023 |Pages: 17
DOI: 10.4018/IJSWIS.318339
Article PDF Download
Open access articles are freely available for download


Entity alignment aims to identify equivalent entity pairs from different knowledge graphs (KGs). Recently, aligning temporal knowledge graphs (TKGs) that contain time information has aroused increasingly more interest, as the time dimension is widely used in real-life applications. The matching between TKGs requires seed entity pairs, which are lacking in practice. Hence, it is of great significance to study TKG alignment under scarce supervision. In this work, the authors formally formulate the problem of TKG alignment with limited labeled data and propose to solve it under the active learning framework. As the core of active learning is to devise query strategies to select the most informative instances to label, the authors propose to make full use of time information and put forward novel time-aware strategies to meet the requirement of weakly supervised temporal entity alignment. Extensive experimental results on multiple real-world datasets show that it is important to study TKG alignment with scarce supervision, and the proposed time-aware strategy is effective.
Article Preview


Temporal knowledge graphs (TKGs) store entities, their relationships, and time information with a structured knowledge representation. They are widely used to facilitate downstream tasks in the field of artificial intelligence, such as recommender systems (Schall, 2015) and natural language understanding (Choudhury et al., 2022). A TKG stores knowledge in the form of IJSWIS.318339.m01, where IJSWIS.318339.m02 is the subject entity, IJSWIS.318339.m03 is the object entity, IJSWIS.318339.m04 denotes the relation between entities, and IJSWIS.318339.m05 represents a specific timestamp or a time interval with beginning time and ending time. Typical TKGs include ICEWS14 (García-Durán et al., 2018), DELT (Trivedi et al., 2017), and Wikidata (Erxleben et al., 2014) which contain temporal facts; for example, the triple IJSWIS.318339.m06 is valid only from January 2021.

Since most TKGs are developed independently, existing TKGs are often incomplete but complementary to each other. Thus, several TKG fusion approaches are put forward and attempt to combine several TKGs into a single and comprehensive one. As an important stage of TKG fusion, temporal entity alignment (TEA) is the task of detecting the equivalent entities (i.e., the entities that refer to the same object) from different TKGs.

The task of TEA is inherently challenging from at least the following aspects:

  • Usage of Time Information: In TKGs, most of the events have specific time stamps or time intervals, such as IJSWIS.318339.m07 and IJSWIS.318339.m08. If the observer neglecst the time information and only considers the relation IJSWIS.318339.m09 and the object entity IJSWIS.318339.m10, they may mistakenly match IJSWIS.318339.m11 and IJSWIS.318339.m12. Figure 1 provides a more specific example. Therefore, in TEA, it is critical to make good use of the time information.

  • Heterogeneity: Since most different TKGs are constructed individually and obtain source data from various channels, the same entities in different graphs may have different relations and time information with other entities. Different TKGs may cover the different parts of factual events, which makes matching more difficult. For instance, in Figure 2, the two entities IJSWIS.318339.m13 and IJSWIS.318339.m14 refer to the same entity in the real world, but it is hard to match them since they are associated with different events.

Complete Article List

Search this Journal:
Volume 20: 1 Issue (2024)
Volume 19: 1 Issue (2023)
Volume 18: 4 Issues (2022): 2 Released, 2 Forthcoming
Volume 17: 4 Issues (2021)
Volume 16: 4 Issues (2020)
Volume 15: 4 Issues (2019)
Volume 14: 4 Issues (2018)
Volume 13: 4 Issues (2017)
Volume 12: 4 Issues (2016)
Volume 11: 4 Issues (2015)
Volume 10: 4 Issues (2014)
Volume 9: 4 Issues (2013)
Volume 8: 4 Issues (2012)
Volume 7: 4 Issues (2011)
Volume 6: 4 Issues (2010)
Volume 5: 4 Issues (2009)
Volume 4: 4 Issues (2008)
Volume 3: 4 Issues (2007)
Volume 2: 4 Issues (2006)
Volume 1: 4 Issues (2005)
View Complete Journal Contents Listing