Apache developers' Big-Five personality profiler

Content and scripts from this repository can be freely reused for academic purposes, provided that you cite the following paper in your work:

F. Calefato, F. Lanubile, and B. Vasilescu (2019) “A large-scale, in-depth analysis of developers’
personalities in the Apache ecosystem.” Information and Software Technology, Vol. 114, Oct., 
pp. 1-20, DOI: 10.1016/j.infsof.2019.05.012.

0. Dataset

The final results of the scripts (i.e., developers' monthly scores per project) are stored here (see the files in CSV format). Instead, the entire MySQL database, containing the data scraped from the Apache website, the email archives, and the code metadata obtained from GitHub, is stored here.

The dump can be imported into a pre-existing db named apache as follows:

$ mysql -u <username> -p<PlainPassword> apache < apachebig5.sql

Repeat the instruction above for all the .sql files provided.

1. Cloning

$ git clone https://github.com/collab-uniba/personality.git --recursive

2. Configuration

Edit the following configuration files:

• src/python/db/cfg/setup.yml - MySQL database configuration

mysql:
    host: 127.0.0.1
    user: root
    passwd: *******
    db: apache

• src/python/big5_personality/personality_insights/cfg/watson.yml - IBM Watson Personality Insights (you will need to register and get your personal username and password)

personality:
    username: secret-user
    password: secret-password
    version: 2017-10-13

• src/python/big5_personality/liwc/cfg/receptiviti.yaml - Receptiviti (you will need to register and get your personal api key and api secret key)

receptiviti:
    baseurl: https://api-v3.receptiviti.com
    api_key: *****
    api_secret_key: *****

3. Crawl Apache projects

• Setup: First, install the library libgit2 on your system. Then, use a Python 3 environment and install the required packages from src/python/requirements.txt
• Execution: From directory src/python/apache_crawler run:

$ scrapy apache_crawler -t (json|csv) -o apache-projects.(json|csv) [-L DEBUG --logfile apache.log]

4. Mine mailing lists (for Git projects only)

• Setup: Use Python 2 environment and install packages from src/python/ml_downloader/requirements.txt. Then, recreate database schema as follows:

$ mysql -u<user> -p<password> apache < submodules/mlminer/db/data_model_mysql.sql

• Execution: From directory src/python/ml_downloader run:

$ sh run.sh

5. Clone Git projects

• Setup: Use Python 3 environment as described in Step 3.
• Execution: From directory src/python/git_cloner run:

$ sh run.sh

Projects will be cloned into the subfolder apache_repos.

<!--- ## 5. Mine pull requests (for git projects only) *Setup* Use Python 3 environment as described in Step 3. Also, add a new file `gh/github-api-tokens.txt` and enter a GitHub API access token per line -- the more the better. *Execution* From directory `src/python/pr_downloader` run: ```bash $ sh run.sh ``` -->

6. Get developers' location from GitHub

• Setup:
  1. In MySql command line enter following instruction:

     set character set utf8mb4; 
     

  2. Use Python 3 environment as described in Step 3.
  3. Add a new file github-api-tokens.txt and enter a GitHub API access token
• Execution: From directory src/python/github_users_location run:

$ sh run.sh [reset]

where:

• reset: to empty db table containing github users location

7. Unmask aliases (identify unique developer IDs)

• Setup:
  1. Use Python 3 environment as described in Step 2.
  2. At first run, execute from dicrectory src/python/unmasking:

     $ python nltk_download.py
     

• Execution: From directory src/python/unmasking run:

$ sh run.sh

8. Build developer commit history (for Git projects only)

• Setup: Use Python 3 environment as described in Step 3.
• Execution: From directory src/python/commit_analyzer run:

$ sh run.sh 

9. Compute developers' Big Five traits scores per month from emails (for Git projects only)

• Setup:
  1. Install NLoN package as described here
  2. Use Python 3 environment as described in Step 3.
• Execution: From directory src/python/big5_personality run:

$ sh run.sh <tool> [reset]

where:

• tool: tool name, either liwc15 or p_insights
• reset: to empty the db tables containing personality data before new computing

10. Export results

• Setup: Use Python 3 environment as described in Step 3.
• Execution: From directory src/python/export_results run:

$ sh run.sh <tool>

where:

• tool: tool name, with values in {liwc07, liwc15, p_insights}

Results are stored in files personality_liwc.csv and personality_p_insights.csv.