A Trio Neural Model for Dynamic Entity Relatedness Ranking

TL;DR

This paper introduces a neural network model that dynamically measures entity relatedness over time, utilizing collective attention as supervision, and outperforms existing static methods in large-scale experiments.

Contribution

The work presents a novel neural approach for dynamic entity relatedness ranking that captures temporal changes and leverages collective attention for supervision.

Findings

Achieves superior performance over baseline methods

Effectively models dynamic entity relationships

Demonstrates robustness on large-scale datasets

Abstract

Measuring entity relatedness is a fundamental task for many natural language processing and information retrieval applications. Prior work often studies entity relatedness in static settings and an unsupervised manner. However, entities in real-world are often involved in many different relationships, consequently entity-relations are very dynamic over time. In this work, we propose a neural networkbased approach for dynamic entity relatedness, leveraging the collective attention as supervision. Our model is capable of learning rich and different entity representations in a joint framework. Through extensive experiments on large-scale datasets, we demonstrate that our method achieves better results than competitive baselines.