Constitutional Cherry-Picking Replication Materials

Quick Start

1. Install Anaconda
2. Open a new terminal window and run: conda create -n pubcon python=3.9.21 pip jupyter
   conda activate pubcon
3. Clone or download this repository.
4. Install required packages: pip install -r required_packages.txt
5. Launch Jupyter Notebook: jupyter notebook
6. Open installer.ipynb and select "Run All Cells" from the "Run" tab.

---

First Steps

Download Anaconda

• Download and install Anaconda
• Accept the default installation settings and choose to initialize Conda when prompted.

Create Conda Environment

After installation, open a new terminal window and create the environment:

conda create -n pubcon python=3.9.21 pip jupyter conda activate pubcon

---

Installation

Obtain the Repository

The recommended method is to use GitHub Desktop to clone the repository. The directory includes:

• cython/: Prebuilt shared objects (angular_distance.so) for Mac Intel, Mac ARM, and Linux Intel. You may need to build your own version for other systems.
• installer.ipynb: Jupyter notebook for installing data and NLP model resources.
• es_core_news_lg-3.7.0/: spaCy model for Spanish text segmentation (compatible with spaCy v3.7.x).
• processing/: Python code for initializing the CCP model and processing Chilean data. This code is user-configurable.
• required_packages.txt: Required Python packages for your Conda environment.
• README.md: This file.

Install Required Packages

Make sure you’re in the same directory as required_packages.txt. You can set your working directory using the cd command in the terminal. For example:

cd "/Users/janedoe/Downloads/pubcon-main"

Then run:

pip install -r required_packages.txt

---

Open Jupyter Notebook

With the pubcon environment activated, run:

jupyter notebook

This will open a new browser tab with the Jupyter interface.

---

Run the Installer

Using Jupyter, navigate to the repository folder and open installer.ipynb. Select "Run All Cells" from the "Run" menu.

This will populate your top-level directory with:

• data/: Chilean data sources and benchmark data
• model/: Serialized objects from text and topic processing
• use_ml_3/: Multilingual sentence-level encoder

Depending on your machine and internet connection, this may take several minutes.

---

Platform-Specific Step: Install angular_distance.so

The processing/ directory includes angular_distance.so compiled for Mac Intel by default.

If you're on a different architecture, replace it with the correct file from the cython/ subdirectories.

To build your own shared object file, set your directory to the cython/ folder and run:

conda install cython python setup.py build_ext --inplace

Ensure this is run within the activated pubcon environment.

Rename angular_distance.cpython-39-darwin.so to angular_distance.so and move the file to analysis/_library and processing/. You can replace the old .so files.

---

Processing: Run pipeline.py

pipeline.py processes data sources located in ../data/ and is configured through the config data structure in the main() function.

Steps:

1. Switch to the processing/ directory in your terminal.

2. In pipeline.py, set the run flag to True for any data source you want to process.

3. Run pipeline.py. It will process all selected data sources based on your configuration.

   python pipeline.py

The configuration mirrors the data/ directory layout, so no changes are required beyond updating the run flags.

---

Run the

Using Jupyter, navigate to the analysis/ folder and open pubcon.ipynb.

Run the first cell complete initialization. Then you are good to go. Run other cells as needed.

Have fun!