Article citationsMore>>

Chu, X., Morcos, J., Ilyas, I.F., Ouzzani, M., Papotti, P., Tang, N., et al. (2015) KATARA: A Data Cleaning System Powered by Knowledge Bases and Crowdsourcing. Proceedings of the 2015 ACM SIGMOD International Conference on Management of Data, Melbourne, 31 May-4 June 2015, 1247-1261. https://doi.org/10.1145/2723372.2749431

has been cited by the following article:

• TITLE: A Study of Error Detection Methods for Tabular Data with a Small Number of Labelled Instances

  AUTHORS: Wenjia Tian, Jinwen Shao

  KEYWORDS: Data Governance, Data Quality, Error Detection, Tabular Data, Low Labelling Learning

  JOURNAL NAME: Journal of Computer and Communications, Vol.14 No.7, July 30, 2026

  ABSTRACT: The massive data of the information age raises stricter standards for data quality. Conventional data error detection approaches depend on handcrafted rules or abundant expert labels, limiting their performance under scarce annotation conditions. To tackle this issue, we present JCED (Joint Contextual Error Detector), a low-label error detection model for tabular data optimized end-to-end via four core components: 1) a distance correlation-based logical grouping module to capture nonlinear column correlations; 2) a joint representation module integrating context-agnostic word2vec and context-aware Sentence Transformer for better semantic context modeling; 3) a two-stage hard sample pair sampling module combining intra-cluster compactness and BCE-Rerank to select informative samples; 4) an inverse propensity score weighting classifier to alleviate sampling bias and boost generalization. Evaluations on five real-world datasets reveal that with merely 5 - 25 labeled samples, JCED outperforms state-of-the-art methods (Raha, HoloDetect, etc.) by 140% in average F1-score and 13.76% in detection efficiency. Ablation studies verify the complementary gains of all modules, and JCED exhibits superior robustness under dynamic error rates ranging from 5% to 30%. This research offers a novel technical solution for low-resource data governance.