WebCorpus is a Hadoop-based framework that enables you to calculate statistics on large web corpora extracted from web crawls.

Features

  • linguistic processing of text corpora with multiple GB or TB in size using Apache Hadoop
  • extracts and counts sentences, word n-grams (with or without POS-tags) and cooccurrences
  • reads popular web crawl formats (ARC and WARC)
  • filters input data by language, duplicate URL, duplicate content and encoding errors
  • can be extended by further linguistic counts based on custom UIMA annotations

Project Samples

WebCorpus processes different types of text archives (e.g. ARC as used by the Internet Archive) and extracts linguistic statistics (e.g. word n-gram frequencies, optionally with annotated Part-of-speech-tags). The processing pipeline consists of Hadoop jobs which can be run independently or altogether from a command-line tool. Illustrated example of last two steps of the WebCorpus pipeline. Each of these steps is realized by a specific Hadoop job. In this example, the result are counted and sorted word bigrams with annotated POS (Part-of-speech) tags.

Project Activity

See All Activity >

License

Apache License V2.0

Follow WebCorpus

WebCorpus Web Site

Other Useful Business Software
Go From AI Idea to AI App Fast Icon
Go From AI Idea to AI App Fast

One platform to build, fine-tune, and deploy ML models. No MLOps team required.

Access Gemini 3 and 200+ models. Build chatbots, agents, or custom models with built-in monitoring and scaling.
Try Free
Rate This Project
Login To Rate This Project

User Reviews

Be the first to post a review of WebCorpus!

Additional Project Details

Operating Systems

Linux

Programming Language

Java

Registered

2013-03-08
Report inappropriate content