Calibrated Seq2seq Models for Efficient and Generalizable Ultra-fine Entity Typing

TL;DR

This paper introduces CASENT, a calibrated seq2seq model for ultra-fine entity typing that improves accuracy, calibration, and efficiency, and demonstrates strong zero-shot and few-shot generalization capabilities across diverse datasets.

Contribution

The paper presents CASENT, a novel seq2seq approach with calibration for ultra-fine entity typing, outperforming prior methods in accuracy, speed, and generalization.

Findings

Outperforms previous state-of-the-art in F1 score and calibration error.

Achieves over 50 times inference speedup.

Excels in zero-shot and few-shot settings, surpassing larger language models.

Abstract

Ultra-fine entity typing plays a crucial role in information extraction by predicting fine-grained semantic types for entity mentions in text. However, this task poses significant challenges due to the massive number of entity types in the output space. The current state-of-the-art approaches, based on standard multi-label classifiers or cross-encoder models, suffer from poor generalization performance or inefficient inference. In this paper, we present CASENT, a seq2seq model designed for ultra-fine entity typing that predicts ultra-fine types with calibrated confidence scores. Our model takes an entity mention as input and employs constrained beam search to generate multiple types autoregressively. The raw sequence probabilities associated with the predicted types are then transformed into confidence scores using a novel calibration method. We conduct extensive experiments on the UFET dataset which contains over 10k types. Our method outperforms the previous state-of-the-art in terms of F1 score and calibration error, while achieving an inference speedup of over 50 times. Additionally, we demonstrate the generalization capabilities of our model by evaluating it in zero-shot and few-shot settings on five specialized domain entity typing datasets that are unseen during training. Remarkably, our model outperforms large language models with 10 times more parameters in the zero-shot setting, and when fine-tuned on 50 examples, it significantly outperforms ChatGPT on all datasets. Our code, models and demo are available at https://github.com/yanlinf/CASENT.