Reproducibility in Research

This repository contains codes and files used during the subject course "Reproducibility in Research" at Unicamp. The main objective was to create an Executable Paper by providing all the necessary sources to reproduce the research.

Files Structure

The files are organized as follow:

• /data - Dataset Facebook metrics (original and processed)
• /deliver - The executable paper and experiment's codes
• /dev - Codes and notebooks used during the experiment
• /figures - Figures used on research
• /utils - Some recommendations about the process of research in a Best Practices file

Development Environment

Before to start the experiment, make sure your setup is according to what is describe here.

Languages and Libraries

For running the experiments in this research the following languages and libraries were selected:

1. Environment: Anaconda3 4.3.1
2. Programming Language: Python 3.3
3. Dataframe Library: Panda 0.19.2.
4. Neo4j-Python Driver: Py2neo v3

Graph Database

Option 1: Local

1. Install the Neo4j 3.2.0 Community Edition
2. Run the Neo4j
3. In the Neo4j's Interface select an empty directory for your Graph Database (The Server has a default, check out)
4. Start the Server
5. Open it in your Browser (http://127.0.0.1:7474/browser/)
6. In the first access, the user and password are neo4j, but you have to change the password(For the experiment used in this repository, the password used is neo4jresearch, but you can create yours)
7. Go to /deliver to run the codes according "Workflow".

Option 2: Docker

1. Install the Docker toolbox
2. Create your account in Docker Hub
3. In Docker Hub the Neo4j's image is available on: Neo4j in Docker Hub
4. Open your Docker Toolbox
5. Make the pull of the image: docker pull neo4j
6. Start an instance of Neo4j:

docker run \
    --publish=7474:7474 --publish=7687:7687 \
    --volume=$HOME/neo4j/data:/data \
    neo4j

7. Open in your Browser (http://localhost:7474) If you are using Windows, check out in the benning of your Docker shell the IP Address that it configures, and replace the localhost to the IP Address
8. In the first access, the user and password are neo4j, but you have to change the password(For the experiment used in this repository, the password used is neo4jresearch, but you can create yours) Note: The Graph Database will be create in the directory: $HOME/Neo4j/Data

Workflow

The workflow with the inputs and outputs of this research is organized as follows:

Execute the experiment

To execute the experiment, follow the steps:

1. Make sure you have did all installation steps (Development Environment).
2. Start your Neo4j Server (Local or Docker).
3. Verifiy the code connection_neo4j.ipynb. The host, user and password need to be corrects.
4. Follow the workflow image: first, execute the Prepreprocessing (preprocessing_data.ipynb).
5. Secondly, execute the indexing_data.ipynb.
6. Open the Neo4j Server in Browser.
7. The graph generated can be visualized by running the Cypher query:

MATCH (n) RETURN n

The Graph Database will return and display the complete graph.

Paper

The experiment results and paper of research is available in /deliver directory, or you can click here: Paper