Popularity calculation Pipeline in SQL

[mathemancer]

Fixes

Fixes #431 by @kgodey

Description

This PR adds 3 Apache Airflow DAGs, defined at

• src/cc_catalog_airflow/dags/recreate_image_popularity_calculation.py
• src/cc_catalog_airflow/dags/refresh_all_image_popularity_data.py
• src/cc_catalog_airflow/dags/refresh_image_view_data.py These DAGs define a "pipeline" in the Upstream DB that culminate with the PostgreSQL materialized view image_view. This new view has a column standardized_popularity containing a float between 0 and 1 representing the popularity of a given image within a given provider (the domain from which the image and its metadata were collected).

Technical details

The standardized_popularity value is calculated from a raw 'popularity metric' such as views, comments, likes, etc. The current ones we use are

• views in the case of Flickr
• global_usage_count in the case of Wikimedia Commons (this is the number of 'wikiverse' pages that use an image). We map the raw values to the interval [0, 1) via a process that allows us to pin a given percentile value within the real distribution (over the values taken by the raw metric for a given provider) to a desired value in the interval. The current implementation pins the 85th percentile value to 0.85.

To calculate the standardized value in [0, 1) for a given metric, we do the following:

1. For some provider, calculate the 85th percentile value (denote it by V) within the distribution of the raw metric. In this implementation, the percentile value is stored for each provider, metric pair in the materialized view image_popularity_constants in the value column.
2. Calculate a constant C = (1 - 0.85) / 0.85 * V. This is also stored for each provider, metric pair in image_popularity_constants in the constant column.
3. For each image, with raw metric value of x, calculate the standardized popularity via P = x / (x + C). In this implementation, this is stored in the standardized_popularity column of the image_view materialized view.

Related to the PR: #426

Tests

There are a number of tests for the new functionality.

Also, the reviewer can follow the README to bring up the dev environment, turn on the tsv_to_postgres_loader DAG, then run the flicker_ingestion and wikimedia_commons_ingestion DAGs for awhile. Once data has been collected, run the recreate_image_popularity_calculation DAG, and then query the image_view in the local DB to see the results.

Screenshots

Checklist

- [X] My pull request has a descriptive title (not a vague title like `Update index.md`). - [X] My pull request targets the *default* branch of the repository (`main` or `master`). - [X] My commit messages follow [best practices][best_practices]. - [X] My code follows the established code style of the repository. - [X] I added tests for the changes I made (if applicable). - [ ] ~I added or updated documentation (if applicable).~ - [X] I tried running the project locally and verified that there are no visible errors. [best_practices]:https://gist.github.com/robertpainsi/b632364184e70900af4ab688decf6f53 ## Developer Certificate of Origin

Developer Certificate of Origin

``` Developer Certificate of Origin Version 1.1 Copyright (C) 2004, 2006 The Linux Foundation and its contributors. 1 Letterman Drive Suite D4700 San Francisco, CA, 94129 Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. Developer's Certificate of Origin 1.1 By making a contribution to this project, I certify that: (a) The contribution was created in whole or in part by me and I have the right to submit it under the open source license indicated in the file; or (b) The contribution is based upon previous work that, to the best of my knowledge, is covered under an appropriate open source license and I have the right under that license to submit that work with modifications, whether created in whole or in part by me, under the same open source license (unless I am permitted to submit under a different license), as indicated in the file; or (c) The contribution was provided directly to me by some other person who certified (a), (b) or (c) and I have not modified it. (d) I understand and agree that this project and the contribution are public and that a record of the contribution (including all personal information I submit with it, including my sign-off) is maintained indefinitely and may be redistributed consistent with this project or the open source license(s) involved. ```

[mathemancer]

It should be noted that the reason not to use standard quantile normalization is to preserve some of the idiosyncrasies of the original distributions, while still ending up with a somewhat comparable standardized popularity.