Elaborative Subtopic Query Reformulation for Broad and Indirect Queries in Travel Destination Recommendation

This repository provides the code and data accompanying our paper titled "Elaborative Subtopic Query Reformulation for Broad and Indirect Queries in Travel Destination Recommendation".

News

Our paper has been accepted to The 1st Workshop on Risks, Opportunities, and Evaluation of Generative Models in Recommender Systems (ROEGEN@RecSys 2024). You can read the paper on arXiv.

Dataset and Ground Truth

We introduce the TravelDest dataset, curated specifically to support our study. It is located in the data/ directory, which contains:

• data/ground_truth: A JSON file with the ground truth answers for each query listed in data/general_queries.txt.
• data/wikivoyage_data_clean: Cleaned metadata files sourced from Wikivoyage for reference and further analysis.

Setup

To get started, follow these steps:

1. Configure API Key: Copy the template configuration file and insert your OpenAI API key:

   cp config_template.py config.py
   # Open `config.py` and replace "YOUR_API_KEY" with your actual API key.

2. Environment Setup: Install the required Python packages:

   pip install -r requirements.txt

Running the Project

Execute the main script using the command below. The script processes data/general_queries.txt and saves the results in the output directory:

bash run.sh

Citation

@article{wen2024elaborative,
  title={Elaborative Subtopic Query Reformulation for Broad and Indirect Queries in Travel Destination Recommendation},
  author={Wen, Qianfeng and Liu, Yifan and Zhang, Joshua and Saad, George and Korikov, Anton and Sambale, Yury and Sanner, Scott},
  journal={arXiv preprint arXiv:2410.01598},
  year={2024}
}