Unlocking the Power of Large Language Models for Multi-table Entity Matching

Taoyu Su; Tingwen Liu; Wenyuan Zhang; Xiaoyang Guo; Yingkai Tang

arxiv: 2604.21238 · v1 · submitted 2026-04-23 · 💻 cs.CL · cs.IR

Unlocking the Power of Large Language Models for Multi-table Entity Matching

Yingkai Tang , Taoyu Su , Wenyuan Zhang , Xiaoyang Guo , Tingwen Liu This is my paper

Pith reviewed 2026-05-09 22:13 UTC · model grok-4.3

classification 💻 cs.CL cs.IR

keywords multi-table entity matchinglarge language modelsentity resolutionprompt engineeringembedding matchingdata integrationdensity pruning

0 comments

The pith

LLM4MEM uses large language models with prompt coordination, consensus embeddings and density pruning to match entities across multiple tables.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper claims that pre-trained language models fall short on multi-table entity matching because numerical attributes create semantic inconsistencies and because the explosion of cross-table pairs creates efficiency and noise problems. It proposes LLM4MEM, an LLM-based framework that first aligns attributes via multi-style prompts, then reduces candidate pairs with transitive consensus embeddings, and finally prunes noisy entities with a density-aware step. Experiments on six MEM datasets show an average 5.1 percent F1 gain over the prior baseline, suggesting that the added LLM coordination and filtering steps deliver measurable gains in accuracy and speed.

Core claim

The central claim is that a single LLM-based pipeline can simultaneously resolve attribute-level semantic mismatches, reduce the quadratic cost of multi-source matching, and filter noisy candidates by combining multi-style prompt coordination, transitive consensus embedding pre-matching, and density-aware pruning, producing higher F1 scores than existing dual-table or PLM approaches on the evaluated collections.

What carries the argument

The LLM4MEM framework, which coordinates large language models through multi-style prompt attribute alignment, transitive consensus embeddings for pre-matching, and density-aware pruning to remove noisy entities.

If this is right

Multi-table entity matching no longer needs unique identifiers if prompts and embeddings can align attributes across sources.
The quadratic growth in candidate pairs can be tamed by first embedding and then transitively grouping entities before full LLM comparison.
Density-based pruning can be inserted as a final filter to improve precision without sacrificing recall in noisy multi-source settings.
The same three-module structure can be applied to other data-integration tasks that suffer from inconsistent attribute representations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The prompt-coordination idea may transfer to other LLM tasks that must reconcile heterogeneous tabular schemas.
If transitive consensus embeddings scale, they could become a general pre-filter for any large-scale entity resolution pipeline.
Density-aware pruning might be replaced or augmented by learned filters once more training data for multi-table noise patterns becomes available.

Load-bearing premise

That the three modules will reliably overcome semantic inconsistencies, efficiency bottlenecks, and noise on arbitrary multi-table collections beyond the six tested datasets.

What would settle it

A new multi-table dataset with substantial numerical value variation where LLM4MEM shows no F1 improvement or a drop relative to the strongest baseline.

Figures

Figures reproduced from arXiv: 2604.21238 by Taoyu Su, Tingwen Liu, Wenyuan Zhang, Xiaoyang Guo, Yingkai Tang.

**Figure 1.** Figure 1: An example of Multi-Table Entity Matching. Leveraging the advanced language understanding capabilities of large language models (LLMs), we propose LLM4MEM, a novel framework for multitable entity matching. Our approach proposes a multi-style prompt-enhanced LLM attribute coordination module to address semantic inconsistencies. Additionally, we develop a transitive consensus embedding matching module to … view at source ↗

**Figure 2.** Figure 2: The overview of LLM4MEM framework. 4 Method Our LLM4MEM framework, shown in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Figure 3: The process of generating DatasetAC from Dsample and Dataset, with (a) the prompt scheme for a simple dataset and (b) the prompt scheme for a difficult dataset. 4.2 Transitive Consensus Embedding Matching Module To address the challenge of comparing entities across multiple sources in entity matching, we introduce the Transitive Consensus Embedding Matching Module (TCEM). First, we compute embeddings for e… view at source ↗

**Figure 4.** Figure 4: The figure shows the sensitivity of key hyperparameters λ (a) and d (b) in the LLM4MEM method to experimental score F1. 6 Conclusion In this paper, we explore Multi-Table Entity Matching (MEM) to identify equivalent entities between multiple tables. To unlock the power of Large Language Models (LLMs) for MEM, we propose a novel LLM-based framework for MEM, termed LLM4MEM. First, the multi-style prompt-enh… view at source ↗

read the original abstract

Multi-table entity matching (MEM) addresses the limitations of dual-table approaches by enabling simultaneous identification of equivalent entities across multiple data sources without unique identifiers. However, existing methods relying on pre-trained language models struggle to handle semantic inconsistencies caused by numerical attribute variations. Inspired by the powerful language understanding capabilities of large language models (LLMs), we propose a novel LLM-based framework for multi-table entity matching, termed LLM4MEM. Specifically, we first propose a multi-style prompt-enhanced LLM attribute coordination module to address semantic inconsistencies. Then, to alleviate the matching efficiency problem caused by the surge in the number of entities brought by multiple data sources, we develop a transitive consensus embedding matching module to tackle entity embedding and pre-matching issues. Finally, to address the issue of noisy entities during the matching process, we introduce a density-aware pruning module to optimize the quality of multi-table entity matching. We conducted extensive experiments on 6 MEM datasets, and the results show that our model improves by an average of 5.1% in F1 compared with the baseline model. Our code is available at https://github.com/Ymeki/LLM4MEM.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

LLM4MEM gives a practical three-module LLM pipeline for multi-table entity matching with a reported 5.1% F1 gain, but the experiments are too lightly documented to judge whether the improvement holds up.

read the letter

The main thing to know is that the authors built LLM4MEM, a three-part system that uses large language models to match entities across several tables at once, and they report it beats baselines by 5.1% F1 on average across six datasets. The work focuses on practical fixes for semantic mismatches and scaling issues in multi-source record linkage. What they do is split the task into a multi-style prompt module to coordinate attributes despite inconsistencies, a transitive consensus embedding step to pre-match and reduce computation, and a density-aware pruning step to clean up noisy candidates. This combination looks tailored to the multi-table setting rather than just extending pair-wise methods. They draw on established ideas like prompt variation and embedding similarity but apply them together here, and releasing the code is a plus for reproducibility. The paper does a decent job laying out the motivation and the architecture in clear terms. For someone dealing with data from multiple databases without shared keys, this could be a straightforward way to experiment with LLMs instead of traditional similarity measures. The soft spots are in the evaluation. The headline result rests on those six datasets, but there's little information on what those datasets look like in terms of size, number of tables, or the severity of inconsistencies. No details on the baseline models or how they were set up for multi-table use, no error bars or significance tests, and no checks on how much the results depend on specific prompt wording. That makes it difficult to tell whether the modules are reliably solving the problems or if the gain is specific to the test cases chosen. The stress test note about limited dataset diversity seems fair based on what's shown. This paper is for people in data integration or knowledge base construction who need an off-the-shelf LLM approach to multi-table matching. A reader interested in engineering improvements for entity resolution would find the design ideas useful even if they want to adapt them. I would recommend sending it for peer review once the authors add more on the experimental setup and perhaps some ablation studies or cross-dataset tests. The core idea is sound enough to warrant referee input, though it will likely need revisions to strengthen the claims.

Referee Report

3 major / 2 minor

Summary. The paper proposes LLM4MEM, an LLM-based framework for multi-table entity matching (MEM) that introduces three modules: a multi-style prompt-enhanced LLM attribute coordination module to mitigate semantic inconsistencies from numerical attribute variations, a transitive consensus embedding matching module to address efficiency issues from increased entity counts across multiple sources, and a density-aware pruning module to filter noisy entities. It reports results from experiments on 6 MEM datasets showing an average 5.1% F1 improvement over baseline models, with code released at https://github.com/Ymeki/LLM4MEM.

Significance. If the reported gains prove robust, the work would meaningfully advance MEM research by showing how LLMs can be structured to handle multi-source semantic and scalability challenges that pre-trained language models struggle with. The public code release is a clear strength that supports reproducibility and community follow-up.

major comments (3)

[Abstract and §5] Abstract and §5 (Experiments): the central claim of an average 5.1% F1 lift is presented without any information on baseline implementations, statistical significance tests, standard deviation across runs, or prompt-sensitivity analysis, leaving the performance improvement only weakly supported.
[§5 and §3] §5 (Experiments) and §3 (Method): no dataset statistics (attribute-type distributions, scale, or inconsistency severity) or cross-dataset transfer experiments are provided, so it is unclear whether the three modules generalize beyond the particular 6 datasets or merely fit their specific characteristics.
[§5] §5 (Experiments): the evaluation does not include comparisons against stronger or more recent LLM-based entity-matching baselines, which is required to establish that the observed gains are attributable to the proposed modules rather than to the choice of weaker reference methods.

minor comments (2)

[§2] §2 (Related Work): ensure all cited MEM and LLM prompting papers are up to date and directly relevant to multi-table settings.
[Figures/Tables] Figure and table captions: clarify the exact definitions of 'baseline' and 'our model' variants so readers can interpret the reported F1 numbers without ambiguity.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed comments, which highlight important areas for strengthening the experimental validation and clarity of our work. We address each major comment below and commit to revisions that enhance the robustness of the reported results.

read point-by-point responses

Referee: [Abstract and §5] Abstract and §5 (Experiments): the central claim of an average 5.1% F1 lift is presented without any information on baseline implementations, statistical significance tests, standard deviation across runs, or prompt-sensitivity analysis, leaving the performance improvement only weakly supported.

Authors: We agree that additional details are needed to robustly support the performance claims. In the revised manuscript, we will expand the experimental section to include: detailed specifications of baseline implementations (including any adaptations for multi-table settings and hyperparameter choices), results from statistical significance tests (e.g., paired t-tests or McNemar's test with p-values), standard deviations and confidence intervals computed over multiple independent runs with varied random seeds, and a prompt-sensitivity analysis varying prompt styles and reporting performance ranges. These additions will directly address the concern and provide stronger evidence for the 5.1% average F1 improvement. revision: yes
Referee: [§5 and §3] §5 (Experiments) and §3 (Method): no dataset statistics (attribute-type distributions, scale, or inconsistency severity) or cross-dataset transfer experiments are provided, so it is unclear whether the three modules generalize beyond the particular 6 datasets or merely fit their specific characteristics.

Authors: We acknowledge that dataset statistics would improve interpretability. We will add comprehensive statistics in a dedicated table or subsection (likely in §3 or §5), covering attribute-type distributions, dataset scales (entities, tables, records), and inconsistency severity metrics (e.g., numerical variance across sources and semantic mismatch rates). For cross-dataset transfer experiments, our evaluation already spans six diverse MEM datasets to demonstrate applicability; however, we will add transfer experiments (training on subsets and evaluating on held-out datasets) where computationally feasible, or provide explicit discussion of the modules' design for generalization. This will clarify that the improvements are not dataset-specific. revision: partial
Referee: [§5] §5 (Experiments): the evaluation does not include comparisons against stronger or more recent LLM-based entity-matching baselines, which is required to establish that the observed gains are attributable to the proposed modules rather than to the choice of weaker reference methods.

Authors: We appreciate the call for stronger baselines to better attribute gains to our modules. While the current baselines encompass established PLM-based and traditional MEM methods adapted to multi-table scenarios, we will incorporate additional recent LLM-based entity matching approaches (e.g., zero-shot/few-shot GPT-based matchers and other contemporary LLM frameworks) in the revised experiments. These will be fairly adapted and evaluated under the multi-table setting to isolate the contributions of the attribute coordination, transitive matching, and pruning modules. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical evaluation stands on its own.

full rationale

The paper proposes an LLM-based framework (LLM4MEM) with three modules for multi-table entity matching and supports its central claim solely through experimental results on six datasets, reporting an average 5.1% F1 improvement over baselines. No equations, derivations, fitted parameters renamed as predictions, or self-citation chains appear in the provided text that would reduce any result to its inputs by construction. The work is self-contained as standard empirical ML research without load-bearing self-referential logic.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that LLMs possess sufficient language understanding to resolve numerical semantic inconsistencies via prompting, plus standard assumptions about embedding similarity and density-based outlier detection. No new physical or mathematical entities are introduced and no free parameters are explicitly fitted in the abstract.

axioms (1)

domain assumption Large language models possess powerful language understanding capabilities that can be leveraged via prompting to address semantic inconsistencies in attribute values.
Explicitly stated as the inspiration for the attribute coordination module.

pith-pipeline@v0.9.0 · 5508 in / 1323 out tokens · 41802 ms · 2026-05-09T22:13:19.315795+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

24 extracted references · 24 canonical work pages

[1]

In: 2024 IEEE 40th International Conference on Data Engineering (ICDE) (2024)

Fan, M., Han, X., Fan, J., Chai, C., Tang, N., Li, G., Du, X.: Cost-effective in- context learning for entity resolution: A design space exploration. In: 2024 IEEE 40th International Conference on Data Engineering (ICDE) (2024)

work page 2024
[2]

IEEE Trans

Ge, C., Wang, P., Chen, L., Liu, X., Zheng, B., Gao, Y.: Collaborem: A self- supervised entity matching framework using multi-features collaboration. IEEE Trans. Knowl. Data Eng. (2023)

work page 2023
[3]

In: International Con- ference on Management of Data, SIGMOD 2014, Snowbird, UT, USA, June 22-27,

Gokhale, C., Das, S., Doan, A., Naughton, J.F., Rampalli, N., Shavlik, J.W., Zhu, X.: Corleone: hands-off crowdsourcing for entity matching. In: International Con- ference on Management of Data, SIGMOD 2014, Snowbird, UT, USA, June 22-27,

work page 2014
[4]

Grattafiori, A., Dubey, A., Jauhri, A., Pandey, A., Abhishek Kadian, e.a.: The llama 3 herd of models (2024)

work page 2024
[5]

Howard, A., Liew, C., (Shopee), M.W., Dane, S.: Shopee - price match guarantee (2021), kaggle

work page 2021
[6]

In: Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles (2023)

Kwon, W., Li, Z., Zhuang, S., Sheng, Y., Zheng, L., Yu, C.H., Gonzalez, J.E., Zhang, H., Stoica, I.: Efficient memory management for large language model serv- ing with pagedattention. In: Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles (2023)

work page 2023
[7]

In: BTW 2021 (2021)

Lerm, S., Saeedi, A., Rahm, E.: Extended affinity propagation clustering for multi- source entity resolution. In: BTW 2021 (2021)

work page 2021
[8]

In: SIGMOD’21: International Conference on Management of Data

Li, P., Cheng, X., Chu, X., He, Y., Chaudhuri, S.: Auto-fuzzyjoin: Auto-program fuzzy similarity joins without labeled examples. In: SIGMOD’21: International Conference on Management of Data. ACM (2021)

work page 2021
[9]

Li,Y., Li, J., Suhara, Y.,Doan, A., Tan, W.: Deep entity matching with pre-trained language models. Proc. VLDB Endow. (2020)

work page 2020
[10]

IEEE Trans

Malkov, Y.A., Yashunin, D.A.: Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs. IEEE Trans. Pattern Anal. Mach. Intell. (2020)

work page 2020
[11]

In: Proceedings of the 2018 International Conference on Management of Data, SIGMOD Confer- ence 2018, Houston, TX, USA, June 10-15, 2018

Mudgal, S., Li, H., Rekatsinas, T., Doan, A., Park, Y., Krishnan, G., Rohit Deep, e.a.: Deep learning for entity matching: A design space exploration. In: Proceedings of the 2018 International Conference on Management of Data, SIGMOD Confer- ence 2018, Houston, TX, USA, June 10-15, 2018. ACM (2018)

work page 2018
[12]

In: The Semantic Web – ISWC 2021: 20th International Semantic Web Con- ference, ISWC 2021, Virtual Event, October 24–28, 2021, Proceedings

Primpeli, A., Bizer, C.: Graph-boosted active learning for multi-source entity reso- lution. In: The Semantic Web – ISWC 2021: 20th International Semantic Web Con- ference, ISWC 2021, Virtual Event, October 24–28, 2021, Proceedings. Springer- Verlag (2021)

work page 2021
[13]

In: Proceedings of EMNLP (2019)

Reimers, N., Gurevych, I.: Sentence-bert: Sentence embeddings using siamese bert- networks. In: Proceedings of EMNLP (2019)

work page 2019
[14]

Reimers, N., Gurevych, I.: Sentence-bert: Sentence embeddings using siamese bert- networks. In: Proceedings of the 2019 Conference on Empirical Methods in Nat- ural Language Processing and the 9th International Joint Conference on Natural Language Processing, EMNLP-IJCNLP 2019. Association for Computational Lin- guistics (2019)

work page 2019
[15]

In: International Conference on Knowledge Engineering and Ontology Development (2021)

Saeedi, A., David, L., Rahm, E.: Matching entities from multiple sources with hierarchical agglomerative clustering. In: International Conference on Knowledge Engineering and Ontology Development (2021)

work page 2021
[16]

In: Proceedings of the 13th International LLM4MEM 13 Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management, IC3K 2021, Volume 2: KEOD

Saeedi, A., David, L., Rahm, E.: Matching entities from multiple sources with hierarchical agglomerative clustering. In: Proceedings of the 13th International LLM4MEM 13 Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management, IC3K 2021, Volume 2: KEOD. SCITEPRESS (2021)

work page 2021
[17]

In: KEOD (2021)

Saeedi, A., David, L., Rahm, E.: Matching entities from multiple sources with hierarchical agglomerative clustering. In: KEOD (2021)

work page 2021
[18]

Singh, R., Meduri, V.V., Elmagarmid, A.K., Madden, S., Papotti, P., Quiané-Ruiz, J., Solar-Lezama, A., Tang, N.: Synthesizing entity matching rules by examples. Proc. VLDB Endow. (2017)

work page 2017
[19]

Team, Q.: Qwen2.5: A party of foundation models (September 2024)

work page 2024
[20]

Team, T.: The falcon 3 family of open models (December 2024)

work page 2024
[21]

In: SIGMOD ’22: International Conference on Management of Data, Philadelphia, PA, USA, June 12 - 17, 2022

Tu, J., Fan, J., Tang, N., Wang, P., Chai, C., Li, G., Fan, R., Du, X.: Domain adaptation for deep entity resolution. In: SIGMOD ’22: International Conference on Management of Data, Philadelphia, PA, USA, June 12 - 17, 2022. ACM (2022)

work page 2022
[22]

Wang, P., Zeng, X., Chen, L., Ye, F., Yuren Mao, e.a.: Promptem: Prompt-tuning for low-resource generalized entity matching. Proc. VLDB Endow. (2022)

work page 2022
[23]

In: Proceedings of the 31st International Conference on Computational Linguistics, COLING 2025, Abu Dhabi, UAE, January 19-24, 2025

Wang, T., Chen, X., Lin, H., Chen, X., Han, X., Sun, L., Wang, H., Zeng, Z.: Match, compare, or select? an investigation of large language models for entity matching. In: Proceedings of the 31st International Conference on Computational Linguistics, COLING 2025, Abu Dhabi, UAE, January 19-24, 2025. pp. 96–109. Association for Computational Linguistics (2025)

work page 2025
[24]

In: 40th IEEE International Conference on Data Engineering, ICDE 2024

Zeng, X., Wang, P., Mao, Y., Chen, L., Liu, X., Gao, Y.: Multiem: Efficient and effective unsupervised multi-table entity matching. In: 40th IEEE International Conference on Data Engineering, ICDE 2024. IEEE (2024)

work page 2024

[1] [1]

In: 2024 IEEE 40th International Conference on Data Engineering (ICDE) (2024)

Fan, M., Han, X., Fan, J., Chai, C., Tang, N., Li, G., Du, X.: Cost-effective in- context learning for entity resolution: A design space exploration. In: 2024 IEEE 40th International Conference on Data Engineering (ICDE) (2024)

work page 2024

[2] [2]

IEEE Trans

Ge, C., Wang, P., Chen, L., Liu, X., Zheng, B., Gao, Y.: Collaborem: A self- supervised entity matching framework using multi-features collaboration. IEEE Trans. Knowl. Data Eng. (2023)

work page 2023

[3] [3]

In: International Con- ference on Management of Data, SIGMOD 2014, Snowbird, UT, USA, June 22-27,

Gokhale, C., Das, S., Doan, A., Naughton, J.F., Rampalli, N., Shavlik, J.W., Zhu, X.: Corleone: hands-off crowdsourcing for entity matching. In: International Con- ference on Management of Data, SIGMOD 2014, Snowbird, UT, USA, June 22-27,

work page 2014

[4] [4]

Grattafiori, A., Dubey, A., Jauhri, A., Pandey, A., Abhishek Kadian, e.a.: The llama 3 herd of models (2024)

work page 2024

[5] [5]

Howard, A., Liew, C., (Shopee), M.W., Dane, S.: Shopee - price match guarantee (2021), kaggle

work page 2021

[6] [6]

In: Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles (2023)

Kwon, W., Li, Z., Zhuang, S., Sheng, Y., Zheng, L., Yu, C.H., Gonzalez, J.E., Zhang, H., Stoica, I.: Efficient memory management for large language model serv- ing with pagedattention. In: Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles (2023)

work page 2023

[7] [7]

In: BTW 2021 (2021)

Lerm, S., Saeedi, A., Rahm, E.: Extended affinity propagation clustering for multi- source entity resolution. In: BTW 2021 (2021)

work page 2021

[8] [8]

In: SIGMOD’21: International Conference on Management of Data

Li, P., Cheng, X., Chu, X., He, Y., Chaudhuri, S.: Auto-fuzzyjoin: Auto-program fuzzy similarity joins without labeled examples. In: SIGMOD’21: International Conference on Management of Data. ACM (2021)

work page 2021

[9] [9]

Li,Y., Li, J., Suhara, Y.,Doan, A., Tan, W.: Deep entity matching with pre-trained language models. Proc. VLDB Endow. (2020)

work page 2020

[10] [10]

IEEE Trans

Malkov, Y.A., Yashunin, D.A.: Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs. IEEE Trans. Pattern Anal. Mach. Intell. (2020)

work page 2020

[11] [11]

In: Proceedings of the 2018 International Conference on Management of Data, SIGMOD Confer- ence 2018, Houston, TX, USA, June 10-15, 2018

Mudgal, S., Li, H., Rekatsinas, T., Doan, A., Park, Y., Krishnan, G., Rohit Deep, e.a.: Deep learning for entity matching: A design space exploration. In: Proceedings of the 2018 International Conference on Management of Data, SIGMOD Confer- ence 2018, Houston, TX, USA, June 10-15, 2018. ACM (2018)

work page 2018

[12] [12]

In: The Semantic Web – ISWC 2021: 20th International Semantic Web Con- ference, ISWC 2021, Virtual Event, October 24–28, 2021, Proceedings

Primpeli, A., Bizer, C.: Graph-boosted active learning for multi-source entity reso- lution. In: The Semantic Web – ISWC 2021: 20th International Semantic Web Con- ference, ISWC 2021, Virtual Event, October 24–28, 2021, Proceedings. Springer- Verlag (2021)

work page 2021

[13] [13]

In: Proceedings of EMNLP (2019)

Reimers, N., Gurevych, I.: Sentence-bert: Sentence embeddings using siamese bert- networks. In: Proceedings of EMNLP (2019)

work page 2019

[14] [14]

Reimers, N., Gurevych, I.: Sentence-bert: Sentence embeddings using siamese bert- networks. In: Proceedings of the 2019 Conference on Empirical Methods in Nat- ural Language Processing and the 9th International Joint Conference on Natural Language Processing, EMNLP-IJCNLP 2019. Association for Computational Lin- guistics (2019)

work page 2019

[15] [15]

In: International Conference on Knowledge Engineering and Ontology Development (2021)

Saeedi, A., David, L., Rahm, E.: Matching entities from multiple sources with hierarchical agglomerative clustering. In: International Conference on Knowledge Engineering and Ontology Development (2021)

work page 2021

[16] [16]

In: Proceedings of the 13th International LLM4MEM 13 Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management, IC3K 2021, Volume 2: KEOD

Saeedi, A., David, L., Rahm, E.: Matching entities from multiple sources with hierarchical agglomerative clustering. In: Proceedings of the 13th International LLM4MEM 13 Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management, IC3K 2021, Volume 2: KEOD. SCITEPRESS (2021)

work page 2021

[17] [17]

In: KEOD (2021)

Saeedi, A., David, L., Rahm, E.: Matching entities from multiple sources with hierarchical agglomerative clustering. In: KEOD (2021)

work page 2021

[18] [18]

Singh, R., Meduri, V.V., Elmagarmid, A.K., Madden, S., Papotti, P., Quiané-Ruiz, J., Solar-Lezama, A., Tang, N.: Synthesizing entity matching rules by examples. Proc. VLDB Endow. (2017)

work page 2017

[19] [19]

Team, Q.: Qwen2.5: A party of foundation models (September 2024)

work page 2024

[20] [20]

Team, T.: The falcon 3 family of open models (December 2024)

work page 2024

[21] [21]

In: SIGMOD ’22: International Conference on Management of Data, Philadelphia, PA, USA, June 12 - 17, 2022

Tu, J., Fan, J., Tang, N., Wang, P., Chai, C., Li, G., Fan, R., Du, X.: Domain adaptation for deep entity resolution. In: SIGMOD ’22: International Conference on Management of Data, Philadelphia, PA, USA, June 12 - 17, 2022. ACM (2022)

work page 2022

[22] [22]

Wang, P., Zeng, X., Chen, L., Ye, F., Yuren Mao, e.a.: Promptem: Prompt-tuning for low-resource generalized entity matching. Proc. VLDB Endow. (2022)

work page 2022

[23] [23]

In: Proceedings of the 31st International Conference on Computational Linguistics, COLING 2025, Abu Dhabi, UAE, January 19-24, 2025

Wang, T., Chen, X., Lin, H., Chen, X., Han, X., Sun, L., Wang, H., Zeng, Z.: Match, compare, or select? an investigation of large language models for entity matching. In: Proceedings of the 31st International Conference on Computational Linguistics, COLING 2025, Abu Dhabi, UAE, January 19-24, 2025. pp. 96–109. Association for Computational Linguistics (2025)

work page 2025

[24] [24]

In: 40th IEEE International Conference on Data Engineering, ICDE 2024

Zeng, X., Wang, P., Mao, Y., Chen, L., Liu, X., Gao, Y.: Multiem: Efficient and effective unsupervised multi-table entity matching. In: 40th IEEE International Conference on Data Engineering, ICDE 2024. IEEE (2024)

work page 2024