Using entities from NER on GOV.UK content to power personalisation.
- Data science
- Deep Graph Infomax investigations
- `docs` folder
- govuk-entity-personalisation documentation
Using entities on GOV.UK content to power personalisation
To run the code in this GitHub repository, please make sure your system meets the following requirements:
- Unix-like operating system (macOS, Linux, …);
direnvinstalled, including shell hooks;
direnvto use the environment variables - see below;
- Python 3.5 or above; and
- Python packages installed from the
Note there may be some Python IDE-specific requirements around loading environment variables, which are not considered here.
To allow/trust the
.envrc run the
allow command using
direnv at the top level of this repository.
Note: If you’re using PyCharm, then you will need to apply a few more steps before running
direnv allowin your shell:
- In your shell, run
pip install python-dotenv.
- On PyCharm, click
Pluginsand download the
- On PyCharm, edit your configuration to
Enable EnvFileby ticking the checkbox.
- On PyCharm, click
Build, Execution, Deployment->
Python Consoleand in the
Starting scriptsection, add the following Python code:
from dotenv import load_dotenv
Installing Python packages
To install required Python packages via
pip, first set up a Python virtual
environment; this ensures you do not install the packages globally.
Then run the following
make command at the top level of this repository:
Once you have installed the packages, remember to set up pre-commit hooks.
Creating a Python virtual environment
Base Python interpreter
If you are using base Python, there are multiple ways to create virtual environments in Python using
(but not limited to):
Follow the documentation of your chosen method to create a Python virtual environment.
An overview of the folder structure, and the top-level files can be found here.
Installing pre-commit hooks
This repo uses the Python package
pre-commit to manage pre-commit hooks. Pre-commit hooks
are actions which are run automatically, typically on each commit, to perform some common set of tasks. For example, a
pre-commit hook might be used to run any code linting automatically, providing any warnings before code is committed,
ensuring that all of our code adheres to a certain quality standard.
For this repo, we are using
pre-commit for a number of purposes:
- Checking for any secrets being committed accidentally;
- Checking for any large files (over 5MB) being committed; and
- Cleaning Jupyter notebooks, which means removing all outputs and execution counts.
We have configured
pre-commit to run automatically on every commit. By running on each commit, we ensure
pre-commit will be able to detect all contraventions and keep our repo in a healthy state.
In order for
pre-commit to run, action is needed to configure it on your system.
pre-commitpackage into your Python environment; and
pre-commit installto set-up
pre-committo run when code is committed.
Setting up a baseline for the
detect-secrets hook (if one doesn’t already exist)
detect-secrets hook requires that you generate a baseline file if one is not already present within the root
directory. This is done via running the following at the root of the repo:
detect-secrets scan > .secrets.baseline
Next, audit the baseline that has been generated by running:
detect-secrets audit .secrets.baseline
When you run this command, you’ll enter an interactive console and be presented with a list of high-entropy string / anything which could be a secret, and asked to verify whether or not this is the case. By doing this, the hook will be in a position to know if you’re later committing any new secrets to the repo and it will be able to alert you accordingly.
pre-commit detects secrets during commit:
pre-commit detects any secrets when you try to create a commit, it will detail what it found and where to go to
check the secret.
If the detected secret is a false-positive, you should update the secrets baseline through the following steps:
detect-secrets scan --update .secrets.baselineto index the false-positive(s);
- Next, audit all indexed secrets via
detect-secrets audit .secrets.baseline(the same as during initial set-up, if a secrets baseline doesn’t exist); and
- Finally, ensure that you commit the updated secrets baseline in the same commit as the false-positive(s) it has been updated for.
If the detected secret is actually a secret (or other sensitive information), remove the secret and re-commit. There is no need to update the secrets baseline in this case.
If your commit contains a mixture of false-positives and actual secrets, remove the actual secrets first before updating and auditing the secrets baseline.
Note on Jupyter notebook cleaning
It may be necessary or useful to keep certain output cells of a Jupyter notebook, for example charts or graphs visualising some set of data. To do this, add the following comment at the top of the input block:
This will tell
pre-commit not to strip the resulting output of this cell, allowing it to be committed.