Research map / Graph learning / Knowledge and temporal graphs
ColNet: Embedding the Semantics of Web Tables for Column Type Prediction
Jiaoyan Chen, Ernesto Jiménez-Ruiz, Ian Horrocks, Charles Sutton
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ColNet prediction to Lookup-Vote using the ensemble rule (ablation) | ColNet · T2Dv2 all columns | F1: 7.9% | Improved |
| add ColNet prediction to Lookup-Vote using the ensemble rule (ablation) | ColNet · T2Dv2 all columns | F1: 4.9% | Improved |
| integrate prediction with entity lookup-voting | Lookup-Vote · Limaye | F1: 15.0% and 23.8% | Improved |
| use CNN prediction without lookup-voting | Lookup-Vote · Limaye | F1: 14.0% and 19.2% | Improved |
| form synthetic columns from multiple cells rather than classifying cells individually (ablation) | ColNet CNN · T2Dv2 synthetic-column tests | AUC (Person TM classes): around 0.93 to 0.97 | Improved |
| increase synthetic column size from 1 to 4 (ablation) | ColNet CNN · T2Dv2 synthetic-column tests | AUC (Person TM classes): around 0.93 to 0.97 | Improved |
| increase the knowledge gap by reducing particular entities from ratio 1.0 to 0.1 without transfer (ablation) | ColNet CNN · T2Dv2 simulated knowledge gap | AUC (TM classes): 7.2%, 4.4%, 4.8% and 8.1% | Hurt |
| pre-train with general samples before fine-tuning with particular samples (ablation) | ColNet CNN · T2Dv2 simulated knowledge gap | average AS (FM classes): 56.4%, 68.7%, 70.2%, 71.7% and 77.7% | Improved |
| include particular samples alongside general samples (ablation) | ColNet CNN for Organization FM classes · T2Dv2 | average AS: 25.8% lower | Improved |
| rely on general samples alone instead of using particular samples for fine-tuning (ablation) | ColNet CNN for Organization FM classes · T2Dv2 | average AS: 25.8% lower | Hurt |
Closest papers by meaning
- UniTabE: A Universal Pretraining Protocol for Tabular Foundation Model in Data Science (ICLR 2024 · similarity 0.85)
- WikiDBs: A Large-Scale Corpus Of Relational Databases From Wikidata (NeurIPS 2024 · similarity 0.85)
- CARTE: Pretraining and Transfer for Tabular Learning (ICML 2024 · similarity 0.84)
- ConTextTab: A Semantics-Aware Tabular In-Context Learner (NeurIPS 2025 · similarity 0.84)
- 4DBInfer: A 4D Benchmarking Toolbox for Graph-Centric Predictive Modeling on RDBs (NeurIPS 2024 · similarity 0.84)
- Making Pre-trained Language Models Great on Tabular Prediction (ICLR 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
