Decoding Text Spans for Efficient and Accurate Named-Entity Recognition

TL;DR

SpanDec is an efficient span-based NER framework that reduces inference costs by computing span interactions at the final transformer layer and filtering unlikely candidates, achieving high accuracy and throughput.

Contribution

The paper introduces SpanDec, a novel span-based NER method that improves efficiency by computing span interactions late and filtering candidates early, enhancing scalability and speed.

Findings

SpanDec matches baseline accuracy on multiple benchmarks.

It significantly improves throughput and reduces computational cost.

The method is suitable for high-volume and on-device NER applications.

Abstract

Named Entity Recognition (NER) is a key component in industrial information extraction pipelines, where systems must satisfy strict latency and throughput constraints in addition to strong accuracy. State-of-the-art NER accuracy is often achieved by span-based frameworks, which construct span representations from token encodings and classify candidate spans. However, many span-based methods enumerate large numbers of candidates and process each candidate with marker-augmented inputs, substantially increasing inference cost and limiting scalability in large-scale deployments. In this work, we propose SpanDec, an efficient span-based NER framework that targets this bottleneck. Our main insight is that span representation interactions can be computed effectively at the final transformer stage, avoiding redundant computation in earlier layers via a lightweight decoder dedicated to span representations. We further introduce a span filtering mechanism during enumeration to prune unlikely candidates before expensive processing. Across multiple benchmarks, SpanDec matches competitive span-based baselines while improving throughput and reducing computational cost, yielding a better accuracy-efficiency trade-off suitable for high-volume serving and on-device applications.