pdf2Speech and speechTranslate contain the configurations for running the workflows. The workflows itself are in another Git (described later).
data and outlierdata contain the gathered data, which is visualized in the paper.
*.ipynb files contain code to analyse the numbers and create the figures, which are shown in the paper.
- Launch the docker-compose, starting a MongoDB instance
- Check the configuration of your MongoDB instance in the following files:
In order to run the experiments, in the first step, you need to deploy both workflows, i.e., https://github.com/AFCLWorkflows/pdf2Speech and https://github.com/AFCLWorkflows/speechTranslate
When all functions are deployed, copy the ARNs and Google Function URLs to the following files:
- pdf2Speech/fc-amazon.yaml
- pdf2Speech/fc-google.yaml
- pdf2Speech/fc-amazon.yaml
- pdf2Speech/fc-google.yaml
For AWS you must replace all occurances of arn:aws:lambda:<fill here>,
for GCP you must replace all occurances of <fill your cloudfunctions.net url here> with your corresponding function's identifiers.
Moreover, you need to copy your credentials to those files: pdf2Speech/credentials.properties speechTranslate/credentials.properties
Lastly, input and output for the workflow needs to be configured.
Therefore replace <your aws bucket url (https:*.s3.amazonaws.com)> with your AWS Bucket URL and
<your gcp bucket url (https:storage.cloud.google.com/...)> with your GCP Bucket name and upload the respective input files to those buckets.
Disclaimer: This changes need to be done in a lot of files, therefore we suggest to do it via a capable editor which allows you to search and replace
text in multiple files.
Within each workflow directory (e.g. pdf2Speech and speechTranslate) there are execute-*.py
files which help you execute a given kind of configuration.
After execution the execution results are written to your Mongodatabase.
In order to run the experiments, java needs to be installed and available in $PATH.
For this replication package the AFCL jar file is pre-packaged.
If you want to compile it from source, the source is available on the website: https://github.com/sashkoristov/enactmentengine.
In order to run across multiple regions feel free to exchange function URLs in yaml files, or service regions in JSON files.
We recommend to create a virtual environment (e.g. using python -m venv .venv && source .venv/bin/activate && pip install -r requirements), then its easy to execute the respective *.ipynb and execute-*.py files.