TrajLLM

This repository contains the source code for the paper "TrajLLM: Explanation-Guided Semantic Trajectory Analysis".

Environment setup

conda create --name trajllm python=3.9 -y
conda activate trajllm

conda install pytorch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 pytorch-cuda=11.8 -c pytorch -c nvidia

pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.0.1+cu118.html

pip install wandb
pip install peft
pip install pandas
pip install transformers
pip install sentencepiece
pip install datasets
pip install pcst_fast

Data Preprocessing

Since extracting labels and explanations is expensive, we provide prepprocessed dataset GeoLife. There are file size limit, and other larger dataset can be constructed with prompts provided.

# generate data_info.pkl that records features and LLMs' supervison.
python -m data_generate.py

Training

Train SetFormer and explanation generator sperately.

# Training SeTFormer supervised by LLMs' explanations
python -m run.py
# A simplified version for fast reproducibility: Distilling the LLM's reasoning ability to the explanation generator
python -m distill.py

Generate your own dataset

You can use prompts in extrac_data to generate your own datasets.