Skip to content

Releases: espositopasquale/KExtractor

Release list

Zero release

Zero release Latest
Latest

Choose a tag to compare

View all tags
@espositopasquale espositopasquale released this 28 Jul 11:24
0.1.0
69cff1f

KExtractor 0.1.0 — Zero Release

This is the initial release of KExtractor a Java-based CLI that transforms unstructured English text into structured, Semantic Web–compliant knowledge graphs.

Leveraging the Stanford CoreNLP toolkit, KExtractor orchestrates a full linguistic pipeline — coreference resolution, Named Entity Recognition (NER), Open Information Extraction (OpenIE) — using deterministic, rule-based components rather than a probabilistic, black-box approach. The result is a transparent pipeline where a given input consistently yields the same output, prioritizing reproducibility and analytical rigor.

Core Capabilities

  • Advanced NLP Pipeline — Uses a shift-reduce dependency parsing strategy for high-fidelity syntactic analysis and relationship extraction.
  • Deterministic Coreference Resolution — Resolves pronominal references before relationship extraction, reducing entity ambiguity in the final graph.
  • Semantic Web Integration — Outputs linked data as JSON-LD (default), Turtle, RDF/XML, or N-Quads via the --format flag. Graphs combine the schema.org vocabulary with a custom project namespace (kex), ready for downstream ingestion into graph databases and reasoning tools.
  • Corpus Processing — Accepts direct string input or batch processing of flat text documents from the terminal.
  • Dual Execution Modes — Run in-process for quick, one-off tasks (Local Mode), or route requests to a local background CoreNLP server on port 9000 (--server) for large-scale or repeated extraction without reloading models each time.
  • Cross-Platform Launch Scripts — run-mac-linux.sh and run-windows.bat wrappers detect whether the CoreNLP server is running, start it if needed, and wait for it to initialize before processing input.
  • Full Analysis Mode — --full-analysis flag triggers the complete analysis suite.

System Requirements

Requirement Details
Java Runtime JDK 21 or later
Build Tool Apache Maven (required for compilation)
Memory Minimum 4 GB RAM recommended
curl Required by launch scripts to check server status

Recommended heap sizes: -Xmx6g minimum for reliable execution, -Xmx8g as the general default, and -Xmx10g–-Xmx12g for large documents or batch processing.

Getting Started

git clone https://github.com/espositopasquale/KExtractor.git
cd KExtractor
mvn clean package
cp target/KExtractor-0.1.0.jar ./kextractor.jar

Then run a quick extraction via the launch script:

./run-mac-linux.sh "Linus Torvalds created the Linux kernel in 1991." output.json

Add --server for background-server mode, or --format=ttl / --format=rdf / --format=nquads to change the output serialization.

Licensing Notes

For full setup instructions, all CLI flags, output examples, and troubleshooting (memory tuning, port conflicts, terminating the background server), see the README.

Assets 3
Loading