Joe

"Reason alone can never give rise to any original idea." D.Hume

"We don't uncover meaning, we create it." L.Wittgenstein

About

Based in Dublin. Focused on transforming unstructured text into structured knowledge by combining NER and SVO extraction for robust knowledge graph construction.

I work on methods to resolve ambiguity in text, extract accurate entity relations, and integrate SVO pipelines with NER systems. I prefer simple, reproducible approaches and clear evaluations.

Research interests

  • Named Entity Recognition (NER) — entity detection & classification
  • Subject-Verb-Object (SVO) extraction — relation templates from text
  • Knowledge Graph construction & linking — from extracted triples to KG
  • Disambiguation & entity linking — improving precision for downstream tasks
  • Evaluation & reproducibility — clear benchmarks and error analysis

Selected projects

Integration of NER + SVO pipeline

Goal: produce high-precision triples for knowledge graph ingestion.

Short description: Experimented with rule-based post-processing of SVO outputs filtered by entity labels from an NER model. Evaluated on domain-specific dataset.

Dataset work — SciIE & custom corpora

Using established datasets and curated domain text for training and evaluation.

Publications & writing

If you want to list publications here, add entries like:

  • Doe, J. (2025). Title of paper. Conference/Journal.
  • Doe, J., & Colleague. (2024). Another paper. Workshop proceedings.

I'll keep this page intentionally minimal — tell me if you'd like automated import from BibTeX or Google Scholar.

Contact

Prefer email or academic profile links.

Email: joe@example.com

GitHub / Papers / CV: add links here.