How hpi-information-systems/metanome works
Project Type and Purpose
• Metanome is a modular, full-stack data-profiling platform.
• Core is a Java/Maven multi-module system:
• A backend REST service (runs in Tomcat, exposes HTTP/JSON).
• An Angular/web frontend (separate repo, pulled in as submodule).
• A CLI tool for headless execution.
• Libraries for algorithm integration, helper data structures, testing and packaging.
• Purpose: load profiling algorithms as plugins, run them over input data (file or DB), collect and post-process results, persist them, and serve via an API/UI.
File-Structure Analysis
• Top-level modules:
• algorithm_helper: reusable data structures (PLIs, graphs) with unit tests.
• algorithm_integration: defines Algorithm interface, configuration/input/result APIs, plugin loading, execution and JSON serialization.
• algorithm_template_root: skeleton for building new profiling algorithms.
• backend: JAX-RS resources, algorithm loading/execution engine, input generators, result post-processing (analysis, comparison, ranking, visualization), Hibernate-based persistence.
• backendwar: packaged WAR with sample data.
• frontend: web UI (Angular) under WEB-INF, built via Maven.
• metanome-cli: command-line Java client that boots Algorithms or calls the REST API.
• file_upload: Python script to POST files to backend.
• deployment: assembly of Tomcat + frontend + backend for local run.
• testing_algorithms & test_helper: sample algorithms and testing utilities.
• Build orchestration via root pom.xml, sub-module poms, Travis CI (.travis.yml) and Git hooks for license checks.
Key Insights from README
• Backend communicates over HTTP REST with web frontend.
• Frontend is an Angular application in a separate repository.
• Build profiles: default (libraries + tests), “deployment-local” (packages Tomcat+frontend+sample data).
• Algorithms are supplied as JAR plugins; configuration and parameter types are pluggable.