Continual Entity Alignment for Growing Knowledge Graphs

Entity alignment across two knowledge graphs

A New Scenario and Task

Landscapes of number of games released every week on Google Play (left) and App Store (right). [Source: Internal Data]

A growing knowledge graph G is a list of consecutive snapshots {G⁰, G¹, …, Gᵗ}, where each snapshot is a static graph. And there is Gᵗ⁻¹ ⊆ Gᵗ.

Given two growing knowledge graphs and a set of pre-aligned entity pairs as anchors, continual entity alignment is about discovering evolving potential alignments over time.

ContEA: A Framework For Continual Entity Alignment

Overview of ContEA: A Framework for Continual Entity Alignment
L = Lₐ + α·Lᵣ  # α is a weight to measure relative importance of Lᵣ
Process of Subgraph-based Entity Alignment Module
L = Lₐ(APA) + α·Lᵣ + β·Lₐ(STA) # α and β are both weights
Process of Embedding and Alignment Update Module

Performances of ContEA

Results of entity alignment on DBPZH-EN. NA stands for Not Applicable
Time cost comparison of ContEA and retraining baselines
Performance of ContEA when given different values to α and β

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store