SpeedRead: A Fast Named Entity Recognition Pipeline

TL;DR

SpeedRead is a high-performance named entity recognition pipeline that significantly outperforms existing systems in speed, enabling scalable analysis of large web-scale text corpora.

Contribution

The paper introduces SpeedRead, a fast NER pipeline combining efficient tokenization, POS tagging, and knowledge-based recognition, achieving at least 10 times faster performance.

Findings

SpeedRead runs at least 10 times faster than Stanford NLP pipeline.

It maintains high accuracy with a Penn Treebank-compliant tokenizer and near state-of-art POS tagging.

The pipeline enables scalable web-scale text analysis.

Abstract

Online content analysis employs algorithmic methods to identify entities in unstructured text. Both machine learning and knowledge-base approaches lie at the foundation of contemporary named entities extraction systems. However, the progress in deploying these approaches on web-scale has been been hampered by the computational cost of NLP over massive text corpora. We present SpeedRead (SR), a named entity recognition pipeline that runs at least 10 times faster than Stanford NLP pipeline. This pipeline consists of a high performance Penn Treebank- compliant tokenizer, close to state-of-art part-of-speech (POS) tagger and knowledge-based named entity recognizer.