/INFACT

The repository contains the source code for the INFACT evaluation framework, which is developed to conduct user-centric evaluation studies for dialog systems, Q&A, conversational recommender systems, and related domains.

Primary LanguageHTMLMIT LicenseMIT

INFACT: An Online Human Evaluation Framework for Conversational Recommendation

This repository contains complete source code of the INFACT framework, which is developed to facilitate the user-centric evaluation studies of dialog systems, conversational recommender systems, and related domains.

Installation

Basically, To run the project on localhost, first you need to create and activate a virtual environment for your Django project. Therefore, please follow the below commands sequentially

cd CRS_Evaluation
python3 -m pip install --upgrade pip
pip3 install requirement.txt
pip3 install virtualenv

To verify the installation of virtual environment library, use

which virtualenv

Now, create a virtual environment for the INFACT framework using

virtualenv venv

Activate your venv

source ./venv/bin/activate

After this command, you should be able to see venv as a virtual environment created in your command promt.

Once you create and activate your virtual environment, to run the localhost, use

python3 manage.py runserv

Once your localhost server is ready, hit this URL in a browser and experience the wonders

 http://127.0.0.1:8000/

How to parse the JSON data collected with this study?

The Python script named 'Parse_study_data.py' can be used to parse the complete data into MS excel sheets.

INPUT: JSON study data

OUTPUT: Separate excel sheets for valid, invalid rating scores, demographic, and feedback questionnaires.

Helpful links

You may have a look at the below links in case of abbration in activating the virtual environment

Install Django using virtualenv

Configure a virtual environment in PyCharm

Create a virtual environment in PyCharm terminal

Citation

@InProceedings{manzoor2022infact,
  title={INFACT: An Online Human Evaluation Framework for Conversational Recommendation},
  author={Manzoor, Ahtsham and Jannach, Dietmar},
  booktitle={KaRS Workshop at RecSys '22},
  address = {Seattle, USA},
  year={2022},
  keywords = {Conversational Recommender Systems, user-centric studies, evaluation, dialog systems},
}