A web browser-based search and visualization architecture for complex multilayer linguistic corpora with diverse types of annotation.

ANNIS is an open source project distributed under the Apache Public License, Version 2.0. The system requires a PostgreSQL installation to run, and the server version also requires a running web-server (e.g. Apache Tomcat or Jetty). Both versions require the Installation of Java OpenJDK 8. If your operating system does already include Java. we recommend the installation of the open-source and free of charge installers provided at

Current stable ANNIS distribution: Version 3.6.0

For installation instructions see the Installation section in the User Guide. There are two editions of ANNIS, one for the local use on a standard computer (ANNIS Kickstarter) and one which is only needed when running a web-server.

See the change log for detailled information on what was changed in this version.

Experimental Beta Version 4.0.0-beta.4

This beta pre-release is a complete overhaul of the ANNIS service backend. Instead of using the relational database PostgreSQL, a custom AQL implementation based on graphs called graphANNIS is used.

ANNIS 4 currently only supports a sub-set of the ANNIS Query Langugage (AQL) compared to ANNIS 3. Full support is planned, but some backward-compatible features of AQL will only be available in a compatibility mode. See the chapter “Differences in Compatibility Mode” of the User Guide in in the Tutorial for more information. For adminstrators, there are instructions on how to migrate from ANNIS 3 to 4 in the User Guide.

Since the whole backend has been replaced, we expect more bugs that need to be fixed than in usual feature-releases. Please report any issues in the GitHub issue tracker.

Source code

Source code (including the latest experimental developments) is available together with our bugtracker/feature requests (‘issues’) at our GitHub project.

Older distributions

Older versions of ANNIS can be downloaded from


For corpora to be imported into the system, they must be converted into it’s native relational database format using the Pepper converter.

Other Converters:

Further documentation on the PAULA XML format may be found here.