Skip to content

README.md

Latest commit

 

History

History
71 lines (52 loc) · 4.78 KB

README.md

File metadata and controls

71 lines (52 loc) · 4.78 KB
Raw

Graph Algorithms Domain

This directory contains the implementation and resources for analysing code graphs using graph algorithms within the Code Graph Analysis Pipeline. It follows the vertical-slice domain pattern: all Cypher queries, shell scripts, and report templates needed for this analysis live here.

This domain covers three related analysis areas:

  • Centrality: Which nodes (packages, types, artifacts, TypeScript modules) are most influential in the dependency graph — PageRank, ArticleRank, Betweenness, Harmonic, Closeness, HITS, Bridges, CELF.
  • Community Detection: How code units cluster into communities — Louvain, Leiden, Strongly Connected Components, Weakly Connected Components, Label Propagation, K-Core, Approximate Maximum k-cut, HDBSCAN, Local Clustering Coefficient.
  • Similarity: Which code units share the most common dependencies — Jaccard Node Similarity.

The goal is to run the majority of graph algorithms provided by the Neo4j Graph Data Science (GDS) library and compare their results to get a basis for exploration and experimentation.

These algorithms are not included:

Entry Points

The following scripts are discovered and invoked automatically by the central compilation scripts in scripts/reports/compilations/. They are found by filename pattern.

  • centralityCsv.sh: Entry point for centrality CSV reports. Discovered by CsvReports.sh (*Csv.sh pattern).
  • communityCsv.sh: Entry point for community detection CSV reports. Discovered by CsvReports.sh (*Csv.sh pattern).
  • similarityCsv.sh: Entry point for similarity CSV reports. Discovered by CsvReports.sh (*Csv.sh pattern).
  • graphAlgorithmsMarkdown.sh: Entry point for the Markdown summary report. Discovered by MarkdownReports.sh (*Markdown.sh pattern).

Folder Structure

domains/graph-algorithms/
├── README.md                              # This file
├── PREREQUISITES.md                       # Detailed prerequisite documentation
├── COPIED_FILES.md                        # Original → copy mapping for deprecation follow-up
├── centralityCsv.sh                       # Entry point: centrality CSV reports
├── communityCsv.sh                        # Entry point: community detection CSV reports
├── similarityCsv.sh                       # Entry point: similarity CSV reports
├── graphAlgorithmsMarkdown.sh             # Entry point: Markdown summary
├── queries/
│   ├── centrality/                        # 43 Cypher queries (copied from cypher/Centrality/)
│   ├── community-detection/               # 64 Cypher queries (copied from cypher/Community_Detection/)
│   ├── similarity/                        # 10 Cypher queries (copied from cypher/Similarity/)
│   └── statistics/                        # 8 summary Cypher queries for the Markdown report
└── summary/
    ├── graphAlgorithmsSummary.sh          # Markdown assembly logic
    ├── report.template.md                 # Main report template
    └── report_no_graph_data.template.md   # Fallback when no data is present

Prerequisites

See PREREQUISITES.md for full details. Key requirements:

Execution Order

  1. centralityCsv.sh — runs centrality algorithms, writes node properties and CSV files
  2. communityCsv.sh — runs community detection algorithms, writes node properties and CSV files
  3. similarityCsv.sh — runs similarity algorithm, writes SIMILAR relationships and CSV files
  4. graphAlgorithmsMarkdown.sh — assembles the final Markdown report from statistics queries

What This Domain Produces

All output goes into reports/graph-algorithms/:

  • reports/graph-algorithms/<NodeType>/centrality/ — centrality scores per node type and algorithm (e.g., Java_Package, Java_Type, Typescript_Module)
  • reports/graph-algorithms/<NodeType>/communities/ — community assignments per node type and algorithm
  • reports/graph-algorithms/<NodeType>/similarity/ — top similar node pairs per node type
  • reports/graph-algorithms/ — assembled Markdown report (graph_algorithms_report.md)