$ git clone https://github.com/feup-infolab/army-ant-install.git
$ cd army-ant-install
$ docker-compose up
This will create two containers, one for MongoDB, including preloaded document metadata, and another one for Army ANT server that depends on the MongoDB instance.
You might want to run different instances of Army ANT, for different projects. Let's say that you are, for example, working on two papers and would like to do several runs per paper without mixing them in the same instance.
You can do this by launching separate Army ANT instances, using:
$ docker-compose -p ijr2018 up
This will launch a separate instance that, unless you call docker-compose -p ijr2018 down
, you will be able to restore just like the default instance.
Each Army ANT instance creates a volume that maps the local directory data
to /army-ant/data
on the container, enabling indexing operations to run inside the container over a local collection. Please view Army ANT Examples to learn the syntax. An index
command for a collection in data/inex-2009-52t-nl
can be launched within the container using:
$ docker exec -i -t irj2018_army-ant_1 ./army-ant.py index \
--source-path ~/data/inex-2009-52t-nl/corpus \
--source-reader inex_dir \
--index-location /opt/army-ant/data/inex-2099-52t-nl/lucene \
--index-type lucene
Where ijr2018
is the name of the project assigned with docker-compose
.
We are also considering a way to backup your session for archival, but this is something that might be hard to do, as we do not want backward compatibility to be a requirement in Army ANT, in order to avoid it becoming a stagnant project. Freedom to change code is essential in a set for innovation.
Each milestone release of Army ANT should include a set of asset files with evaluation data for the preloaded indexes. For example, release 0.3 of Army ANT includes inex_3t_nl-documents.json.gz
, corresponding to the JSON preloaded in the MongoDB instance, as well as inex-2009-3t-nl.zip
, an archive with the indexed collection, which includes topics and relevance judgments to test the evaluation features provided by Army ANT via its web interface. Further information about the asset files attached to each release can be found in the DOCKER.md of the corresponding release.