LocalizeAgent Reproducibility Package

This repository contains the necessary artifacts to replicate the results of the paper "An LLM-Based Agent-Oriented Approach for Automated Code Design Issue Localization." It includes Python scripts, evaluation results, and a shell script to automate the replication process. The results from the model are stored in the EvalResults directory. This artifact is provided to support reproducibility. Here is the link to our manuscript: LocalizeAgent

## Replication Instructions

To replicate the results presented in the paper, follow these steps:

1. **Prerequisites:**

- Python 3.12.2

- Required Python packages:

```bash

pip install anthropic openai google-generativeai tqdm javalang numpy python-dotenv

```

2. **Setting up LLM API:**

- Ensure you have the necessary API keys for the services used in the scripts (e.g., OpenAI, Anthropic, Google Generative AI). These keys should be set up in your environment before running the scripts.

- Create a .env file and place the API keys for each LLM provide there.

3. **Running the Scripts:**

- Execute the provided `.sh` file to run the replication scripts under `scripts/`:

```bash

./your_script.sh

```

4. **Results:**

- The results from the model are stored in the `EvalResults` directory.

## Directory Structure

- `REPLICATION/`: Contains scripts and files necessary for replication.

- `EvalResults/`: Contains the output and evaluation results from the model.

- `exal_lims.py`, `eval_multi.py`, `java_code_analysis.py`, `lim_interfaces.py`: Scripts used in the replication process.

For any further assistance, please refer to the documentation within the repository or contact the authors.