graphANNIS

This is a new backend implementation of the ANNIS linguistic search and visualization system (http://corpus-tools.org/annis/).

While there are test cases, this project is still in a pre-release beta phase. Only a sub-set of the ANNIS Query Langugage (AQL) is supported yet (full support is planned). Integration with ANNIS is currently implemented in the development version and released as a beta version. There is a tutorial in the Developer Guide on how to embedd graphANNIS in your own application.

The basic design ideas and data models are described in detail in the PhD-thesis "ANNIS: A graph-based query system for deeply annotated text corpora". The thesis describes a prototype implementation in C++ and not Rust, but the design ideas are the same. Noteable differences/enhancements compared to the thesis are:

• Graph storages implement querying inverse edges and finding reachable nodes based on them: this allows to implement inverse operators (e.g. for precedence) and switching operands in situations where it was not possible before.
• The data model has been simplified: the inverse coverage component and inverse edges in the left-/right-most token component have been removed.
• Additional query language features are now supported.

How to compile

graphANNIS is written in the Rust programming language (https://www.rust-lang.org). You can install Rust from https://www.rust-lang.org/tools/install. After you have installed Rust, you can can build the complete project with

cargo build --release

Documentation

• Developer Guide (including descriptions of the data model and tutorials for the API)
• API documentation

3rd party dependencies

This software depends on several 3rd party libraries. These are documented in the "third-party-licenses.html" file in this folder.

Language bindings

• Java: https://github.com/korpling/graphANNIS-java
• Python 3: https://github.com/korpling/graphANNIS-python
• Rust (this repository)
• C (this repository)

Author(s)

• Thomas Krause ([email protected])

How to cite

@phdthesis{Krause2019ANNIS:,
author = {Krause, Thomas},
title = {ANNIS: A graph-based query system for deeply annotated text corpora},
school = {Humboldt-Universität zu Berlin, Mathematisch-Naturwissenschaftliche Fakultät},
year = {2019},
doi = {http://dx.doi.org/10.18452/19659}
}