MG4J is a highly customizable, high-performance, full-text Java search engine for large document collections. It provides state-of-the-art features (such as BM25/BM25F scoring) and new research algorithms.
| Tags | Internet Web Indexing/Search Text Processing Indexing Software Development Libraries Java Libraries |
|---|---|
| Licenses | LGPL |
| Operating Systems | OS Independent |
| Implementation | Java |
Recent releases


Changes: Major improvements were made to indexing. Lightweight compressed-collection construction was added. A skipping system with variable quanta was added. Memory mapping is used for large indices. Many bugs were fixed.


Changes: All new stemmers from Snowball were generating empty strings, causing major indexing problems. This has been fixed.


Changes: This release has a new, high-performance index format, several optimizations, indices with arbitrary payloads (dates, integers, etc.), faster minimal perfect hashing, new operators, and better algorithms.


Changes: Significant speed improvements in index writing and query resolution. A few bugfixes.


Changes: A few important bugs that appeared in 1.1, as well as a very old one, have been fixed.