v0.2.2

Full Changelog: v0.2.1...v0.2.2

v0.2.1

What's Changed

• Add Zenodo DOI badge; version 0.2.1 by @mohammadi-hadi in #7

Full Changelog: v0.2.0...v0.2.1

v0.2.0

Highlights

• Fixed a serving-time leak in the user-history feature: clicks on a results page no longer feed the features of later positions on the same page.
• Reproducibility: a given seed now yields the identical log on any machine (generator partition layout is pinned).
• Real library API: PipelineConfig, structured run() result, lazy typed exports with py.typed; the CLI gains --output metrics.json.
• Supported and CI-tested across PySpark 3.5–4.2 on Python 3.10–3.13 (dependency range widened).

What's Changed

• Update pyspark requirement from <3.6,>=3.5 to >=3.5,<4.3 by @dependabot[bot] in #3
• Bump actions/checkout from 4 to 7 by @dependabot[bot] in #1
• Leakage fix, reproducibility, library API, CI matrix by @mohammadi-hadi in #6

New Contributors

• @dependabot[bot] made their first contribution in #3

Full Changelog: v0.1.0...v0.2.0

v0.1.0

First release. End-to-end counterfactual learning-to-rank on synthetic marketplace search logs: randomized-slice propensity estimation, IPS-weighted GBT re-ranking, and NDCG/MRR evaluation against ground truth, with a full test suite (including propensity-recovery and leakage tests) and CI.

Install the wheel below, or straight from the tag:

pip install "spark-search-ranking @ git+https://github.com/mohammadi-hadi/spark-search-ranking.git@v0.1.0"

What's Changed

• Package for distribution: metadata, console script, release workflow by @mohammadi-hadi in #5

Full Changelog: https://github.com/mohammadi-hadi/spark-search-ranking/commits/v0.1.0