Abstract

Deterministic entity resolution is a clear solution for differentiating between similar variants of products in retrieval-augmented commerce search. This study presents the schema-derived decision logic which can be used to alleviate the confusion between variants prior to retrieval and ranking based on the product title/brand field, the model terms, the pack signals, the attribute slots and the query intent markers. We have created a representative set of 171 instances for commerce-search for this study across six query patterns including exact product names, alias expressions, mixed attributes, ambiguous variants, bundle language and near duplicate product descriptions. The data was organized in a spreadsheet, processed with the Python scripting language and analyzed using rule tracing, text normalization, decision gates, summary tables, and plot visual outputs. This proposed framework requires no additional training labels; the decision of acceptance or rejection is made deterministically based on rules in the fields. The results suggest that the schema-aware logic helps to improve the separation of entities, reduces the number of irrelevant matches in high-intent commerce queries, and enables auditable query behavior. The study also demonstrates the value of compact decision traces, row-level evidence, and structured outputs in aiding product-search teams in diagnosis of retrieval errors, in addition to opaque scores. The framework is in general capable of disambiguating precisely, explainable and operationally for commerce search scenarios.

References

• S. S. Aanen, D. Vandic, and F. Frasincar, “Automated product taxonomy mapping in an e-commerce environment,” Expert Systems with Applications, vol. 42, no. 3, pp. 1298–1313, 2015. https://doi.org/10.1016/j.eswa.2014.09.032
• B. Gu, Z. Li, X. Zhang, A. Liu, G. Liu, K. Zheng, L. Zhao, and X. Zhou, “The interaction between schema matching and record matching in data integration,” IEEE Transactions on Knowledge and Data Engineering, vol. 29, no. 1, pp. 186–199, 2017. https://doi.org/10.1109/TKDE.2016.2611577
• P. Ristoski, P. Petrovski, P. Mika, and H. Paulheim, “A machine learning approach for product matching and categorization: Use case: Enriching product ads with semantic structured data,” Semantic Web, vol. 9, no. 5, pp. 707–728, 2018. https://doi.org/10.3233/SW-180300
• A. Allam, S. Skiadopoulos, and P. Kalnis, “Improved suffix blocking for record linkage and entity resolution,” Data & Knowledge Engineering, vol. 117, pp. 98–113, 2018. https://doi.org/10.1016/j.datak.2018.07.005
• G. Simonini, L. Gagliardelli, S. Bergamaschi, and H. V. Jagadish, “Scaling entity resolution: A loosely schema-aware approach,” Information Systems, vol. 83, pp. 145–165, 2019. https://doi.org/10.1016/j.is.2019.03.006
• J. Li, Z. Dou, Y. Zhu, X. Zuo, and J.-R. Wen, “Deep cross-platform product matching in e-commerce,” Information Retrieval Journal, vol. 23, no. 2, pp. 136–158, 2020. https://doi.org/10.1007/s10791-019-09360-1
• L. Akritidis, A. Fevgas, P. Bozanis, and C. Makris, “A self-verifying clustering approach to unsupervised matching of product titles,” Artificial Intelligence Review, vol. 53, pp. 4777–4820, 2020. https://doi.org/10.1007/s10462-020-09807-8
• D. C. Nascimento, C. E. S. Pires, and T. P. Nóbrega, “Configurable assembly of classification rules for enhancing entity resolution results,” Information Processing & Management, vol. 57, no. 3, art. 102224, 2020. https://doi.org/10.1016/j.ipm.2020.102224
• G. Papadakis, D. Skoutas, E. Thanos, and T. Palpanas, “Blocking and filtering techniques for entity resolution: A survey,” ACM Computing Surveys, vol. 53, no. 2, art. 31, pp. 1–42, 2020. https://doi.org/10.1145/3377455
• V. Christophides, V. Efthymiou, T. Palpanas, G. Papadakis, and K. Stefanidis, “End-to-end entity resolution for big data: A survey,” ACM Computing Surveys, vol. 53, no. 6, pp. 1–42, 2020. https://doi.org/10.1145/3418896
• O. S. Albayrak, T. Aytekin, and T. A. Kalaycı, “Duplicate product record detection engine for e-commerce platforms,” Expert Systems with Applications, vol. 193, art. 116420, 2022. https://doi.org/10.1016/j.eswa.2021.116420
• C. Ye, S. Jiang, H. Zhang, Y. Wu, J. Shi, H. Wang, and G. Dai, “JointMatcher: Numerically-aware entity matching using pre-trained language models with attention concentration,” Knowledge-Based Systems, vol. 251, art. 109033, 2022. https://doi.org/10.1016/j.knosys.2022.109033
• H. Ding, C. Dai, Y. Wu, W. Ma, and H. Zhou, “SETEM: Self-ensemble training with pre-trained language models for entity matching,” Knowledge-Based Systems, vol. 293, art. 111708, 2024. https://doi.org/10.1016/j.knosys.2024.111708.