Skip to content

v2.0.0

Latest
Latest

Choose a tag to compare

View all tags
@schatzopoulos schatzopoulos released this 30 Jul 15:51
v2.0.0
254e486
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

BIP! Ranker v2.0.0

Local-first packaging of BIP! Ranker for reproducible runs and software-paper use: install via a Python venv, run the full pipeline on sample data with run_demo.py, and use a cleaner citation-graph input format. Cluster / spark-submit workflows remain supported.

Compared to v1.1.0, this release includes breaking changes. Update inputs and topic/FWCI invocation before upgrading production pipelines.

Breaking changes

  • Citation graph format — Each line is now
    paper_id <tab> referenced_papers|num_references <tab> initial_score <tab> publication_year.
    The initial score is no longer embedded in the citation payload. Old KanellosTM-style records with an embedded score will not parse correctly.
  • TopicClassesAndFWCI.py — Metrics are keyed on pid only. The --openaire-concepts-output flag and OpenAIRE-id–centric outputs are removed. Results are written under --output-dir as topics/, FWCI/, and 3-year_FWCI/.
  • Impact classes — PageRank.py, AttRank.py, CC.py, and TAR.py emit five-point classes C1–C5 only (aligned with topic-based classes). The three-point / legacy class column is dropped.

What's new

  • Local install — requirements.txt and README instructions for a Python venv (Python 3.8+, Java 8/11+, PySpark).
  • run_demo.py — Runs CC, 3-year CC, PageRank, AttRank, TAR (RAM/ECM), and topic/FWCI on the sample network with local Spark (no HDFS required). Builds sample-data/output/scores.tsv before topic normalisation. Supports --only and --skip.
  • sample-data/ — 50-paper citation graph, concept map, and checkpoint/output directories for a self-contained demo.
  • Documentation — README rewritten for local and cluster use, with per-script CLI argument tables.
  • TAR/ECM — Fixed year handling for the new graph schema.

Upgrade notes from v1.1.0

  1. Convert citation graphs to the four-column format above.
  2. Rebuild the scores table without openaire_id; join indicator outputs on pid.
  3. Drop --openaire-concepts-output from topic/FWCI jobs.
  4. Expect C1–C5 in ranking outputs and update any downstream consumers of class labels.
Assets 2
Loading