jcseg

Why we included this project

For teams doing Chinese-language search or text analysis in Java, Jcseg is a self-contained segmentation engine worth a serious look. It handles both Chinese and English text with the MMSEG algorithm, and its seven segmentation modes run from a simple, latency-friendly one to a fine-grained mode built for retrieval, so you can tune the accuracy-versus-speed balance to fit your workload. Beyond tokenization, it adds keyword, key phrase, and key sentence extraction and article summarization, all built on the TextRank algorithm, plus a dictionary-based entity layer that catches URLs, email addresses, phone numbers, dates, and person names. A bundled REST API and ready-made analyzers for Lucene, Solr, Elasticsearch, and OpenSearch let you drop it into an existing search stack with little glue code. Custom dictionaries are plain files that reload automatically, which matters when your content needs domain vocabulary a generic segmenter would miss.