Introduction

Midas Loop is a web application for taking Universal Dependencies corpora and improving the quality of their annotations. For more information on motivation, functionality, and supported workflows, please see our paper.

See documentation and our paper for more details.

Quick Start

To quickly get a server up and running:

1. Clone the repository.

git clone https://github.com/gucorpling/midas-loop.git
cd midas-loop

2. You may now interact with Midas Loop by invoking the script scripts/midas-loop.sh. First, invoke it with no arguments to download the JAR and create a configuration file.

bash scripts/midas-loop.sh

3. Now, we will import a sample document into Midas Loop.

wget https://raw.githubusercontent.com/gucorpling/amalgum/master/amalgum/fiction/dep/AMALGUM_fiction_amontillado.conllu
bash scripts/midas-loop.sh import AMALGUM_fiction_amontillado.conllu

4. Now generate a token for yourself.

bash scripts/midas-loop.sh token add --name YOURNAME --email YOUREMAIL --quality gold

5. Start the server, and enter the token you generated for yourself.

bash scripts/midas-loop.sh run

6. If you wish to use Midas Loop with NLP services, refer to our instructions for setup. After the services are set up, modify your config.edn to look like below and reimport any documents you imported.

{:port 3000,
 :midas-loop.server.xtdb/config {:main-db-dir "xtdb_data"},
 :midas-loop.server.tokens/config {:token-db-dir "xtdb_token_data"},
 :nlp-services [{:type :http :anno-type :xpos :url "http://localhost:5555"}
                {:type :http :anno-type :sentence :url "http://localhost:5556"}
                {:type :http :anno-type :head :url "http://localhost:5557"}]}

Citation

If you use Midas Loop in your research, please cite it like so (URL to be added soon):

@InProceedings{gessler-etal-2022-midas-loop,
  author    = {Luke Gessler and Lauren Levine and Amir Zeldes},
  booktitle = {Proceedings of LREC 2022},
  title     = {{M}idas {L}oop: Prioritized {H}uman-in-the-{L}oop Annotation for Large Scale Multilayer Data},
  year      = {2022},
  address   = {Marseille, France}
}