On Entity Identification in Language Models

TL;DR

This paper investigates how well language models internally recognize and differentiate named entities, revealing that they effectively cluster entity mentions and represent entity information compactly in early layers.

Contribution

It introduces a novel framework for analyzing entity mention clustering in LM representations and provides empirical evidence of effective entity identification across multiple models.

Findings

LMs cluster entity mentions with high precision and recall (0.66 to 0.9)

Entity information is stored in a low-dimensional space in early layers

Representation of entities influences word prediction performance

Abstract

We analyze the extent to which internal representations of language models (LMs) identify and distinguish mentions of named entities, focusing on the many-to-many correspondence between entities and their mentions. We first formulate two problems of entity mentions -- ambiguity and variability -- and propose a framework analogous to clustering quality metrics. Specifically, we quantify through cluster analysis of LM internal representations the extent to which mentions of the same entity cluster together and mentions of different entities remain separated. Our experiments examine five Transformer-based autoregressive models, showing that they effectively identify and distinguish entities with metrics analogous to precision and recall ranging from 0.66 to 0.9. Further analysis reveals that entity-related information is compactly represented in a low-dimensional linear subspace at early LM layers. Additionally, we clarify how the characteristics of entity representations influence word prediction performance. These findings are interpreted through the lens of isomorphism between LM representations and entity-centric knowledge structures in the real world, providing insights into how LMs internally organize and use entity information.