Artificial Intelligence Incident Database (AIID)

Information about the goals and organization of the AI Incident Database can be found on the production website. This page concentrates on onboarding for the following types of contributions to the database,

Contribute changes to the current AI Incident Database.
Contribute a new summary to the AI Incident Database. A "summary" is a programmatically generated summary of the database contents. Examples are available here.
Contribute a new taxonomy to the AI Incident Database. Details on taxonomies are available in the arXiv paper.
Contribute a new application facilitating a new use case for the database.

In most cases unless you are contributing quick fixes, we recommend opening an issue before contributing in any of these areas.

You can see the live site here

Contributing Changes

Anyone can contribute code to the project. The system is being built as a "do-ocracy", meaning those who "do" have influence over the development of the code.

The steps for contributing changes are the following,

Create a fork of the repository.
Clone the fork to your local environment.
Open a feature branch from whichever branch you would like to change. This is typically the master branch, so you can do git checkout master then git checkout -b feature-cool-new-thing.
Make your changes, commit them, then push them remote.
Open a pull request to the master branch.
Update the pull request based on the review.
See the pull request get pulled. :)

Please make sure your code is well organized and commented before opening the pull request.

Site Architecture

The site has three components that all be considered "serverless," meaning there is no dynamic backend templating the application or responding to API requests. The components include,

Web host. This is the web server hosting the Gatsby-based web application. The site is hosted in production on Netlify.
Index. The Algolia search index.
Database. The Atlas MongoDB service exposed via MongoDB Realm. Atlas provides the storage, and Realm supports the web API with user account provisioning. This database does not currently automatically populate the search index, but periodic dumps will be made from this database to Algolia. The full database can support documents and details that are either unsupported by Algolia, or would be too expensive to host there.

More details are available in the Production System information below. We recommend most people forego setting up a development environment with their own Index and Database. You should instead concentrate on setting up a Gatsby development site.

Style guide:

ESLint and Prettier has been configured to help enforcing code styles. Configuration details can be found on .eslintrc.json and .prettierrc.
Husky and lint-staged are installed and pre-commit hook added to check lint/prettier issues on staged files and fix them automatically before making commit.
format and lint scripts can be used manually to fix style issues.

Production System

Netlify

The site is hosted by Netlify and is integrated into GitHub to generate previews of all code states. This allows for seamless previewing of the application. However, the preview domains do not match the whitelisted domains known by the MongoDB service, so not all functionality is expected to work in the build previews without whitelisting the domain preview.

Builds: Builds are presently run at least every 12 hours automatically by a GitHub action. They are also run on merge requests from forks. The site deploys from the master branch automatically if the build succeeds.

MongoDB Database

See mongo.md

Algolia

Algolia is the instant search provider interfaced in the Discover application. It is presently manually when new incident reports are ingested into the database.

Cloudinary

Cloudinary is what we use to host and manage report images.

Setting up a development environment

Depending on what feature you are working on, there will be different systems you'll need to set up after you've forked and cloned this repository:

Basic setup

Get a Gatsby environment working. Most of the time, you'll only need to run:

npm i -g gatsby-cli

Create a .env file under site/gatsby-site with the following contents:

GATSBY_REALM_APP_ID=aiidstitch2-vsdrv
MONGODB_CONNECTION_STRING=mongodb+srv://readonlyuser:[email protected]
MONGODB_REPLICA_SET=aiiddev-shard-00-02.6zxh5.mongodb.net,aiiddev-shard-00-01.6zxh5.mongodb.net,aiiddev-shard-00-00.6zxh5.mongodb.net

GATSBY_ALGOLIA_APP_ID=JD5JCVZEVS
GATSBY_ALGOLIA_SEARCH_KEY=c5e99d93261645721a1765fe4414389c

In the same folder, install dependencies using npm (do not use yarn, it will ignore the package-lock.json file):

npm i

You are ready to start a local copy of the project:

gatsby develop

You should have a local copy of the project running on https://localhost:8000.

The values you placed into the env file are all associated with a staging environment that is periodically rebuilt from the production environment. While this helps you get setup more quickly, if you will be making changing to the backend you will need your own development backend that you can control, modify, and potentially break.

MongoDB setup

If the feature you are working on includes structural changes to the MongoDB database or Realm functions, you'll need to create your own project by going to https://cloud.mongodb.com and following these steps:

Create a new MongoDB project (the free tier will be enough)
Create a new Atlas cluster with the name: AIIDDev
Create a new Realm App. The name should be AIIDStitch2. Realm will give it an id like aiidstitch2-xxxxx
Create a new database user with admin access and another user with read-only permissions

Replicating the Database

Download the latest database backup from https://incidentdatabase.ai/research/snapshots

Then, run mongorestore (included in MongoDB tools) using the admin user created in the step above to upload the database backup:

mongorestore mongodb+srv://<USER>:<PASSWORD>@aiiddev.<CLUSTER>.mongodb.net/aiidprod aiidprod`

Deploy the Realm App

Install the realm-cli and follow the login process: https://docs.mongodb.com/realm/cli/

npm i -g mongodb-realm-cli

Once authenticated, you can deploy the realm app by going to site/realm of this repo and running:

 realm-cli push --remote=aiidstitch2-<YOUR-NEW-APP-ID>

Finally, update the previously created .env:

GATSBY_REALM_APP_ID =`aiidstitch2-xxxxx`
MONGODB_CONNECTION_STRING=mongodb+srv://<username>:<password>@aiiddev.<CLUSTER>.mongodb.net
MONGODB_REPLICA_SET=<CLUSTER>-shard-00-00.6zxh5.mongodb.net,aiiddev-shard-00-01.<CLUSTER>.mongodb.net,aiiddev-shard-00-02.<CLUSTER>.mongodb.net

Restart Gatsby, and your local app should fetch data from your MongoDB environment!

Algolia environment setup

Suppose the feature you are developing involves mutating the Algolia index (used in the Discover app). In that case, you'll have to create your own by signing up for Algolia here: https://www.algolia.com and following these steps:

Create a new App (free tier will be enough)
Create a new index called instant_search
Go to https://incidentdatabase.ai/downloadIndex and save the new_index.json file
Upload this file to your newly created index, by selecting Use File Algolia's UI
Now go to your Algolia App settings, and click API Keys, using the values from there to update your .env file accordingly.

GATSBY_ALGOLIA_APP_ID=<YOUR APP ID>
GATSBY_ALGOLIA_SEARCH_KEY=<YOUR SEARCH KEY>
ALGOLIA_ADMIN_KEY=<YOUR ADMIN KEY>

Algolia index settings are uploaded on build time, so you'll have to do that at least once from your local env:

gatsby build

Restart Gatsby, and you should have a complete working environment!

Translation Process

The translation process runs on Gatsby's postBuild event and consists of 3 steps:

-1 Get the list of languages, which is pulled from the /src/components/i18n/languages.js using the GATSBY_AVAILABLE_LANGUAGES environment variable as a filter:

GATSBY_AVAILABLE_LANGUAGES=en,es,it,af

-2 Translate each incident report to each language, and save the translated reports to a translations database under a collection for each language:

translations 
    |-- incident_reports_en
    |   |-- { title, text, report_number }
    |   |-- { title, text, report_number }
    |
    |--incident_report_es
    |   |-- { title, text, report_number }
        |-- { title, text, report_number }

To access this database, a user with read/write permissions needs to be provided through the following environment variable:

MONGODB_TRANSLATIONS_CONNECTION_STRING=mongodb+srv://<user>:<password>@aiiddev.<host>.mongodb.net

-3 Generate an Algolia index from each translated collection and upload them to Algolia. Each index has the following naming format:

instant_search-{language code}

After the first run, the following applies for subsequent runs: Translations of report fields load from the existing translations/incident_reports_{language}/{doc} document, and if not found, then the Translate API is hit. Algolia indexes are replaced every time the process runs

Cost

The translate API charges ~20USD per million characters and can translate to 111 languages.

At the time of writing, there are 1336 Incident Reports, each report consisting of ~4000 characters, with a total sum of ~5 million characters.

Considering the pricing above, translating all ingested reports to one language will cost (5 million / 1 million) * $20 = ~$100, and translating all incident reports to all languages $100 * 111= ~$11k.

The translation process defaults to a **dry run ** mode that prepends a string to every translated text instead of hitting Google's API.

Therefore, Translated texts in this mode will look like: translated-{language}-YouTube to crack down on inappropriate content masked as kids’ cartoons

The dry run is disabled through an environment variable as follows:

TRANSLATE_DRY_RUN=false

Geocoding

If the feature you are working on depends on Google's Geocoding API, please add the following environment variable with the appropriate value to your .env file.

GOOGLE_MAPS_API_KEY=XXXXXXXXXXXX

Deployment Setup

Deployment of the site consists of two parts: deployment of the backend related features that runs as a Github Action and deployment of the frontend related features that runs on Netlify:

Netlify

The Netlify build process runs every time a push is made to an open PR or master or develop. To correctly set up this process, the following environment variables need to be created using Netlify's build settings UI:

ALGOLIA_ADMIN_KEY=
AWS_LAMBDA_JS_RUNTIME=nodejs14.x # required to run the Gatsby v4
GATSBY_ALGOLIA_APP_ID=
GATSBY_ALGOLIA_SEARCH_KEY=
GATSBY_REALM_APP_ID=
MONGODB_CONNECTION_STRING=
MONGODB_REPLICA_SET=

Github Actions

Two workflows take care of deploying the Realm app to both production and staging environments, defined in realm-production.yml and realm-staging.yml. Each workflow looks for environment variables defined in a Github Environment named production and staging.

These environments must contain the following variables:

GATSBY_REALM_APP_ID=
REALM_API_PRIVATE_KEY=
REALM_API_PUBLIC_KEY=

Testing

For integration testing, we use Cypress. You can run the desktop app continuously as part of your development environment or run it on demand in headless mode.

First, add two new environment variables:

E2E_ADMIN_USERNAME=
E2E_ADMIN_PASSWORD=

As their names imply, they should be an existing user's credentials with the admin role.

To use the desktop version, run:

npm run test:e2e

And to run it in continuous integration (headless) mode:

test:e2e:ci

Adding new Taxonomies

To add new taxonomies, follow these steps:

Let's say you want to add the CTECH Taxonomy.

Create a new MongoDB collection using the lowercased Taxonomy name: ctech.
Define the appropriate rules, relationships, and schema for this new collection. Specifically, a schema that specifies the fields of the taxonomy. An example for the fictional ctech schema is below. You can optionally populate the schema and use the MongoDB Realm API to generate the schema programmatically.

/site/realm/data_sources/mongodb-atlas/aiidprod/ctech/schema.json

{
    "properties": {
        "_id": {
            "bsonType": "objectId"
        },
        "classifications": {
            "bsonType": "object",
            "properties": {
                "taxonomy field 1": {
                    "bsonType": "bool"
                },
                "taxonomy field 2": {
                    "bsonType": "string"
                },
                "Publish": {
                    "bsonType": "bool"
                }
            }
        },
        "incident_id": {
            "bsonType": "int"
        },
        "namespace": {
            "bsonType": "string"
        },
        "notes": {
            "bsonType": "string"
        }
    },
    "title": "ctech"
}

Add the new document to the taxa collection, that lists the taxonomy fields. This annotates the classifications found in the new collection and determines properties of their display in the user interface.

{
    "_id": {
        "$oid": "61f158f4c19c105af2f3d6af"
    },
    "namespace": "ctech",
    "weight": {
        "$numberInt": "50"
    },
    "description": "# What are these resources?\n\nThe following resources have been associated with incidents in the database to provide tools and processes to persons and companies looking for best practices in the prevention or mitigation of similar incidents in the future.",
    "field_list": [
        {
            "short_name": "taxonomy field 1",
            "long_name": "taxonomy field 1",
            "short_description": "Lorem ipsum...",
            "long_description": "__Lorem ipsum__",
            "display_type": "bool",
            "mongo_type": "bool",
            "default": null,
            "placeholder": null,
            "permitted_values": null,
            "weight": {
                "$numberInt": "70"
            },
            "instant_facet": true,
            "required": false
        },
        {
            "short_name": "Publish",
            "display_type": "bool",
            "mongo_type": "bool"
        }
    ]
}

Update createCitiationPages.js to have it pull the new taxonomy definitions. The relevant lines are:
- /site/gatsby-site/page-creators/createCitiationPages.js#L125
- /site/gatsby-site/page-creators/createCitiationPages.js#L180
Restart Gatsby

Restarting Gatsby should make the new taxonomy available on the citation pages, so you can visit /cite/1 to see a form for editing the taxonomy. Please note that you will need to be logged in to a user account on the application to see the form.

License

The codebase currently carries an GNU Affero General Public License and all contributions to the project falls under the license. Contributors disclaim all rights to their contributions. Please reach out to discuss the license terms if they are causing you any issues. We selected the GNU Affero license to encourage contributing back to the project, but the license itself provides the community wide latitude in working with this codebase.

This program is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more details.

You should have received a copy of the GNU Affero General Public License along with this program. If not, see http://www.gnu.org/licenses/.

Contact

For inquiries, you are encouraged to open an issue on this repository or visit the contact page.

Name		Name	Last commit message	Last commit date
Latest commit History 437 Commits
.github/workflows		.github/workflows
scripts		scripts
site		site
.gitattributes		.gitattributes
.gitignore		.gitignore
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
LICENSE.txt		LICENSE.txt
README.md		README.md
mongo.md		mongo.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Artificial Intelligence Incident Database (AIID)

Contributing Changes

Site Architecture

Production System

Netlify

MongoDB Database

Algolia

Cloudinary

Setting up a development environment

Basic setup

MongoDB setup

Replicating the Database

Deploy the Realm App

Algolia environment setup

Translation Process

Cost

Geocoding

Deployment Setup

Netlify

Github Actions

Testing

Adding new Taxonomies

To add new taxonomies, follow these steps:

License

Contact

About

Releases

Packages

Languages

License

alexmcode/aiid

Folders and files

Latest commit

History

Repository files navigation

Artificial Intelligence Incident Database (AIID)

Contributing Changes

Site Architecture

Production System

Netlify

MongoDB Database

Algolia

Cloudinary

Setting up a development environment

Basic setup

MongoDB setup

Replicating the Database

Deploy the Realm App

Algolia environment setup

Translation Process

Cost

Geocoding

Deployment Setup

Netlify

Github Actions

Testing

Adding new Taxonomies

To add new taxonomies, follow these steps:

License

Contact

About

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages