HeadersIQ develops an explainable, header-centric approach to Semantic Table
Interpretation and Data Quality Assessment. The latest QKG@ESWC 2026 paper
presents the large-scale framework, while the earlier extended preprint and
IEEE/ACM BDCAT 2024 paper document the initial development and evaluation of
the attribute-based approach.
Publications
Peer-reviewed papers, preprints, supporting artefacts, and related research outputs.
An Explainable Header-Centric Framework for Large-Scale Semantic Table Interpretation and Data Quality Assessment
🏆 Best Paper Award Finalist
This paper presents an explainable, metadata-only, header-centric framework
for Semantic Table Interpretation, Column Type Annotation, and Data Quality
Assessment. It maps column headers to interpretable semantic formats,
preserves traceability through SourceKeywords, activates rule-grounded data
quality checks, and summarises detected issues through the HeadersIQ metric.
Authors: Marcelo Valentim Silva, Hannes Herrmann, Valerie Maxville
• Workshop on Quality of Knowledge Graphs (QKG 2026)
• Co-located with the European Semantic Web Conference - ESWC 2026
• CEUR Workshop Proceedings, Vol. 4205, pp. 66–83
Recognition:
Selected as one of four finalists for the Best Paper Award at QKG@ESWC 2026.
@inproceedings{silva2026explainable,
author = {Silva, Marcelo Valentim and Herrmann, Hannes and Maxville, Valerie},
title = {An Explainable Header-Centric Framework for Large-Scale Semantic Table Interpretation and Data Quality Assessment},
booktitle = {Proceedings of the Workshop on Quality of Knowledge Graphs (QKG 2026), co-located with the 23rd European Semantic Web Conference (ESWC 2026)},
series = {CEUR Workshop Proceedings},
volume = {4205},
pages = {66--83},
year = {2026},
publisher = {CEUR-WS.org},
url = {https://ceur-ws.org/Vol-4205/paper12.pdf}
}
Attribute-Based Semantic Type Detection and Data Quality Assessment (extended)
Extended version presenting the full method and results:
semantic-type classification (~23 types + bounded numerics),
evaluation on 50 datasets / 922 columns, and a
comparison where the method detects 81 missing-value cases
vs 1 by YData Profiling.
Authors: Marcelo V. Silva, Hannes Herrmann, Valerie Maxville • Type: Preprint
@misc{silva2024attribute,
title={Attribute-Based Semantic Type Detection and Data Quality Assessment},
author={Silva, Marcelo V. and Herrmann, Hannes and Maxville, Valerie},
year={2024},
eprint={2410.14692},
archivePrefix={arXiv},
primaryClass={cs.DB},
url={https://arxiv.org/abs/2410.14692}
}
Attribute-Based Semantic Type Detection and Data Quality Assessment (short)
Honourable Mention
Short, six-page version presented at the 2024 IEEE/ACM International Conference on Big Data Computing,
Applications and Technologies (BDCAT). Focuses on the attribute-based idea and initial evaluation.
Authors: Marcelo V. Silva, Hannes Herrmann, Valerie Maxville • Venue: IEEE BDCAT 2024 • Type: Short paper
