TIGER: Temporally Improved Graph Entity Linker

TL;DR

By incorporating structural information between entities into the model, the core idea is to integrate graph-based information into text-based information, from which both distinct and shared embeddings are based on an entity's feature and structural relationships and their interaction.

Abstract

Knowledge graphs change over time, for example, when new entities are introduced or entity descriptions change. This impacts the performance of entity linking, a key task in many uses of knowledge graphs such as web search and recommendation. Specifically, entity linking models exhibit temporal degradation - their performance decreases the further a knowledge graph moves from its original state on which an entity linking model was trained. To tackle this challenge, we introduce \textbf{TIGER}: a \textbf{T}emporally \textbf{I}mproved \textbf{G}raph \textbf{E}ntity Linke\textbf{r}. By incorporating structural information between entities into the model, we enhance the learned representation, making entities more distinguishable over time. The core idea is to integrate graph-based information into text-based information, from which both distinct and shared embeddings are based on an entity's feature and structural relationships and their interaction. Experiments on three datasets show that our model can effectively prevent temporal degradation, demonstrating a 16.24\% performance boost over the state-of-the-art in a temporal setting when the time gap is one year and an improvement to 20.93\% as the gap expands to three years. The code and data are made available at \url{https://github.com/pengyu-zhang/TIGER-Temporally-Improved-Graph-Entity-Linker}.

Figures (6)

• Figure 1: In this example, the mention 'Wildcats' could refer to Arizona Wildcats football team, or Wildcats (film) entities. We can learn better entity representations by including information from the graph structure.
• Figure 2: The dataset construction process. We use Wikidata5M to extend TempEL with strutured graph representations. The green section represents the input to our model.
• Figure 3: The proposed TIGER model adaptively integrates text data (mention context and entity descriptions) with graph data (structural graphs, feature matrices, and feature graphs) to enhance temporal accuracy. The model employs a Shared Convolution Module to learn common features and two Distinct Convolution Modules to capture unique features. Additionally, loss functions are used to emphasize these distinctions.
• Figure 4: Recall performance (recall@1) of different models on testset. The solid and dashed lines represent models training on new and continual entities.
• Figure 5: Percentage improvement of the TIGER model compared to the SpEL model across evaluation metrics from recall@1 to recall@64.