Norman William Paton
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Norman William Paton has a total of 604 co-authors. Here are the top co-authors by number of shared publications:
Alvaro A. A. Fernandes
University of Manchester, Manchester, United Kingdom
41 shared publications- Rizos Sakellariou
University of Manchester
22 shared publications - Carole Goble
The Department of Computer Science, the University of Manchester, Manchester, United Kingdom
17 shared publications - Paul Watson
Newcastle University
17 shared publications - Suzanne M. Embury
The University of Manchester, Manchester, United Kingdom
17 shared publications Cornelia Hedeler
School of Computer Science, University of Manchester, Oxford Road, Manchester, United Kingdom
15 shared publications
Latest Publications
- Open Access
- 9 August 2026
- Published by Springer Nature in The VLDB Journal
- Vol. 35 (5) , 46
Conceptual schema inference for tabular datasets using large language models
Large collections of tabular data from data lakes, web tables and open data portals often originate from heterogeneous sources, leading to representational inconsistencies. Understanding and organizing such repositories therefore remains a major challenge. While prior work has primarily focused on dataset discovery and exploration, this paper addresses the complementary problem of conceptual schema inference: automatically deriving a conceptual schema that captures entity types, attributes and inter-type relationships directly from raw tables. We propose two large language model (LLM)-based approaches that use only column headers and cell values: $$\text {GeSI}$$ uses generative LLMs to infer hierarchical types and their attributes from table- and column-level semantics, and to integrate them into a global schema that also captures relationships across types; $$\text {EmSI}$$ employs LLM-based table embeddings to group tables by column-level semantics, infer attributes within each group, and construct hierarchical structures from shared attribute patterns. Finally, we report an experimental analysis demonstrating the effectiveness of our approaches in terms of the conciseness and structural quality of the inferred schema components, their scalability to large repositories, and a case study illustrating end-to-end schema inference.- 9 August 2026
- Published by Springer Nature in The VLDB Journal
- Vol. 35 (5) , 46
Open AccessConceptual schema inference for tabular datasets using large language models
Large collections of tabular data from data lakes, web tables and open data portals often originate from heterogeneous sources, leading to representational inconsistencies. Understanding and organizing such repositories therefore remains a major challenge. While prior work has primarily focused on dataset discovery and exploration, this paper addresses the complementary problem of conceptual schema inference: automatically deriving a conceptual schema that captures entity types, attributes and inter-type relationships directly from raw tables. We propose two large language model (LLM)-based approaches that use only column headers and cell values: $$\text {GeSI}$$ uses generative LLMs to infer hierarchical types and their attributes from table- and column-level semantics, and to integrate them into a global schema that also captures relationships across types; $$\text {EmSI}$$ employs LLM-based table embeddings to group tables by column-level semantics, infer attributes within each group, and construct hierarchical structures from shared attribute patterns. Finally, we report an experimental analysis demonstrating the effectiveness of our approaches in terms of the conciseness and structural quality of the inferred schema components, their scalability to large repositories, and a case study illustrating end-to-end schema inference. - Open Access
Schema Inference for Tabular Data Repositories Using Large Language Models
Minimally curated tabular data often contain representational inconsistencies across heterogeneous sources, and are accompanied by sparse metadata. Working with such data is intimidating. While prior work has advanced dataset discovery and exploration, schema inference remains difficult when metadata are limited. We present SI-LLM (Schema Inference using Large Language Models), which infers a concise conceptual schema for tabular data using only column headers and cell values. The inferred schema comprises hierarchical entity types, attributes, and inter-type relationships. In extensive evaluation on two datasets from web tables and open data, SI-LLM achieves promising end-to-end results, as well as better or comparable results to state-of-the-art methods at each step. All source code, full prompts, and datasets of SI-LLM are available at https://github.com/PierreWoL/SILLM.
Open AccessSchema Inference for Tabular Data Repositories Using Large Language Models
Minimally curated tabular data often contain representational inconsistencies across heterogeneous sources, and are accompanied by sparse metadata. Working with such data is intimidating. While prior work has advanced dataset discovery and exploration, schema inference remains difficult when metadata are limited. We present SI-LLM (Schema Inference using Large Language Models), which infers a concise conceptual schema for tabular data using only column headers and cell values. The inferred schema comprises hierarchical entity types, attributes, and inter-type relationships. In extensive evaluation on two datasets from web tables and open data, SI-LLM achieves promising end-to-end results, as well as better or comparable results to state-of-the-art methods at each step. All source code, full prompts, and datasets of SI-LLM are available at https://github.com/PierreWoL/SILLM.
- Open Access
- 17 June 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 3 (3) , 1-28
TableDC: Deep Clustering for Tabular Data
Deep clustering (DC), a fusion of deep representation learning and clustering, has recently demonstrated positive results in data science, particularly text processing and computer vision. However, joint optimization of feature learning and data distribution in the multi-dimensional space is domain-specific, so existing DC methods struggle to generalize to other application domains (such as data integration). In data management tasks, where high-density embeddings and overlapping clusters dominate, a data management-specific DC algorithm should be able to interact better with the data properties to support data integration tasks. This paper presents a deep clustering algorithm for tabular data (TableDC) that reflects the properties of data management applications that cluster tables (schema inference), rows (entity resolution) and columns (domain discovery). To address overlapping clusters, TableDC integrates Mahalanobis distance, which considers variance and correlation within the data, offering a similarity method suitable for tabular data in high-dimensional latent spaces. TableDC also shows higher tolerance to outliers through its heavy-tailed Cauchy distribution as the similarity kernel. The proposed similarity measure is particularly beneficial where the embeddings of raw data are densely packed and exhibit high degrees of overlap. Data integration tasks may also involve large numbers of clusters, which challenges the scalability of existing DC methods. TableDC learns data embeddings with a large number of clusters more efficiently than baseline DC methods, which scale in quadratic time. We evaluated TableDC with several existing DC, Standard Clustering (SC), and state-of-the-art bespoke methods over benchmark datasets. TableDC consistently outperforms existing DC, SC and bespoke methods.
- 17 June 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 3 (3) , 1-28
Open AccessTableDC: Deep Clustering for Tabular Data
Deep clustering (DC), a fusion of deep representation learning and clustering, has recently demonstrated positive results in data science, particularly text processing and computer vision. However, joint optimization of feature learning and data distribution in the multi-dimensional space is domain-specific, so existing DC methods struggle to generalize to other application domains (such as data integration). In data management tasks, where high-density embeddings and overlapping clusters dominate, a data management-specific DC algorithm should be able to interact better with the data properties to support data integration tasks. This paper presents a deep clustering algorithm for tabular data (TableDC) that reflects the properties of data management applications that cluster tables (schema inference), rows (entity resolution) and columns (domain discovery). To address overlapping clusters, TableDC integrates Mahalanobis distance, which considers variance and correlation within the data, offering a similarity method suitable for tabular data in high-dimensional latent spaces. TableDC also shows higher tolerance to outliers through its heavy-tailed Cauchy distribution as the similarity kernel. The proposed similarity measure is particularly beneficial where the embeddings of raw data are densely packed and exhibit high degrees of overlap. Data integration tasks may also involve large numbers of clusters, which challenges the scalability of existing DC methods. TableDC learns data embeddings with a large number of clusters more efficiently than baseline DC methods, which scale in quadratic time. We evaluated TableDC with several existing DC, Standard Clustering (SC), and state-of-the-art bespoke methods over benchmark datasets. TableDC consistently outperforms existing DC, SC and bespoke methods.
- 1 June 2025
- Published by Springer Nature
- p. 403-422
Taxonomy Inference for Tabular Data Using Large Language Models
- 1 June 2025
- Published by Springer Nature
- p. 403-422
Taxonomy Inference for Tabular Data Using Large Language Models
- Open Access
Gem: Gaussian Mixture Model Embeddings for Numerical Feature Distributions
Embeddings are now used to underpin a wide variety of data management tasks, including entity resolution, dataset search and semantic type detection. Such applications often involve datasets with numerical columns, but there has been more emphasis placed on the semantics of categorical data in embeddings than on the distinctive features of numerical data. In this paper, we propose a method called Gem (Gaussian mixture model embeddings) that creates embeddings that build on numerical value distributions from columns. The proposed method specializes a Gaussian Mixture Model (GMM) to identify and cluster columns with similar value distributions. We introduce a signature mechanism that generates a probability matrix for each column, indicating its likelihood of belonging to specific Gaussian components, which can be used for different applications, such as to determine semantic types. Finally, we generate embeddings for three numerical data properties: distributional, statistical, and contextual. Our core method focuses solely on numerical columns without using table names or neighboring columns for context. However, the method can be combined with other types of evidence, and we later integrate attribute names with the Gaussian embeddings to evaluate the method's contribution to improving overall performance. We compare Gem with several baseline methods for numeric only and numeric + context tasks, showing that Gem consistently outperforms the baselines on four benchmark datasets.
Open AccessGem: Gaussian Mixture Model Embeddings for Numerical Feature Distributions
Embeddings are now used to underpin a wide variety of data management tasks, including entity resolution, dataset search and semantic type detection. Such applications often involve datasets with numerical columns, but there has been more emphasis placed on the semantics of categorical data in embeddings than on the distinctive features of numerical data. In this paper, we propose a method called Gem (Gaussian mixture model embeddings) that creates embeddings that build on numerical value distributions from columns. The proposed method specializes a Gaussian Mixture Model (GMM) to identify and cluster columns with similar value distributions. We introduce a signature mechanism that generates a probability matrix for each column, indicating its likelihood of belonging to specific Gaussian components, which can be used for different applications, such as to determine semantic types. Finally, we generate embeddings for three numerical data properties: distributional, statistical, and contextual. Our core method focuses solely on numerical columns without using table names or neighboring columns for context. However, the method can be combined with other types of evidence, and we later integrate attribute names with the Gaussian embeddings to evaluate the method's contribution to improving overall performance. We compare Gem with several baseline methods for numeric only and numeric + context tasks, showing that Gem consistently outperforms the baselines on four benchmark datasets.
- Open Access
- 9 November 2023
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 56 (4) , 1-37
Dataset Discovery and Exploration: A Survey
Data scientists are tasked with obtaining insights from data. However, suitable data is often not immediately at hand, and there may be many potentially relevant datasets in a data lake or in open data repositories. As a result, data discovery and exploration are necessary, but often time consuming, steps in a data analysis workflow. Data discovery is the process of identifying datasets that may meet an information need. Data exploration is the process of understanding the properties of candidate datasets and the relationships between them. Data discovery and data exploration often go hand in hand and benefit from tool support. This article surveys research areas that can contribute to data discovery and exploration, particularly considering dataset search, data navigation, data annotation and schema inference. For each of these areas, we identify key dimensions that can be used to characterize approaches and the values they can hold, and apply the dimensions to describe and compare prominent results. In addition, by surveying several adjacent areas that are often considered in isolation, we identify recurring techniques and alternative approaches to related challenges, thereby placing results within a wider context than is generally considered.
- 9 November 2023
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 56 (4) , 1-37
Open AccessDataset Discovery and Exploration: A Survey
Data scientists are tasked with obtaining insights from data. However, suitable data is often not immediately at hand, and there may be many potentially relevant datasets in a data lake or in open data repositories. As a result, data discovery and exploration are necessary, but often time consuming, steps in a data analysis workflow. Data discovery is the process of identifying datasets that may meet an information need. Data exploration is the process of understanding the properties of candidate datasets and the relationships between them. Data discovery and data exploration often go hand in hand and benefit from tool support. This article surveys research areas that can contribute to data discovery and exploration, particularly considering dataset search, data navigation, data annotation and schema inference. For each of these areas, we identify key dimensions that can be used to characterize approaches and the values they can hold, and apply the dimensions to describe and compare prominent results. In addition, by surveying several adjacent areas that are often considered in isolation, we identify recurring techniques and alternative approaches to related challenges, thereby placing results within a wider context than is generally considered.
- Open Access
Deep Clustering for Data Cleaning and Integration
Deep Learning (DL) techniques now constitute the state-of-the-art for important problems in areas such as text and image processing, and there have been impactful results that deploy DL in several data management tasks. Deep Clustering (DC) has recently emerged as a sub-discipline of DL, in which data representations are learned in tandem with clustering, with a view to automatically identifying the features of the data that lead to improved clustering results. While DC has been used to good effect in several domains, particularly in image processing, the impact of DC on mainstream data management tasks remains unexplored. In this paper, we address this gap by investigating the impact of DC in data cleaning and integration tasks, specifically schema inference, entity resolution, and domain discovery, tasks that represent clustering from the perspective of tables, rows, and columns, respectively. In this setting, we compare and contrast several DC and non-DC clustering algorithms using standard benchmarks. The results show, among other things, that the most effective DC algorithms consistently outperform non-DC clustering algorithms for data integration tasks. However, we observed a significant correlation between the DC method and embedding approaches for rows, columns, and tables, highlighting that the suitable combination can enhance the efficiency of DC methods.
Open AccessDeep Clustering for Data Cleaning and Integration
Deep Learning (DL) techniques now constitute the state-of-the-art for important problems in areas such as text and image processing, and there have been impactful results that deploy DL in several data management tasks. Deep Clustering (DC) has recently emerged as a sub-discipline of DL, in which data representations are learned in tandem with clustering, with a view to automatically identifying the features of the data that lead to improved clustering results. While DC has been used to good effect in several domains, particularly in image processing, the impact of DC on mainstream data management tasks remains unexplored. In this paper, we address this gap by investigating the impact of DC in data cleaning and integration tasks, specifically schema inference, entity resolution, and domain discovery, tasks that represent clustering from the perspective of tables, rows, and columns, respectively. In this setting, we compare and contrast several DC and non-DC clustering algorithms using standard benchmarks. The results show, among other things, that the most effective DC algorithms consistently outperform non-DC clustering algorithms for data integration tasks. However, we observed a significant correlation between the DC method and embedding approaches for rows, columns, and tables, highlighting that the suitable combination can enhance the efficiency of DC methods.
- Open Access
- 2 June 2023
- Published by Springer Nature in SN Computer Science
- Vol. 4 (4) , 1-20
Data Preparation: A Technological Perspective and Review
Data analysis often uses data sets that were collected for different purposes. Indeed, new insights are often obtained by combining data sets that were produced independently of each other, for example by combining data from outside an organization with internal data resources. As a result, there is a need to discover, clean, integrate and restructure data into a form that is suitable for an intended analysis. Data preparation, also known as data wrangling, is the process by which data are transformed from its existing representation into a form that is suitable for analysis. In this paper, we review the state-of-the-art in data preparation, by: (i) describing functionalities that are central to data preparation pipelines, specifically profiling, matching, mapping, format transformation and data repair; and (ii) presenting how these capabilities surface in different approaches to data preparation, that involve programming, writing workflows, interacting with individual data sets as tables, and automating aspects of the process. These functionalities and approaches are illustrated with reference to a running example that combines open government data with web extracted real estate data.
- 2 June 2023
- Published by Springer Nature in SN Computer Science
- Vol. 4 (4) , 1-20
Open AccessData Preparation: A Technological Perspective and Review
Data analysis often uses data sets that were collected for different purposes. Indeed, new insights are often obtained by combining data sets that were produced independently of each other, for example by combining data from outside an organization with internal data resources. As a result, there is a need to discover, clean, integrate and restructure data into a form that is suitable for an intended analysis. Data preparation, also known as data wrangling, is the process by which data are transformed from its existing representation into a form that is suitable for analysis. In this paper, we review the state-of-the-art in data preparation, by: (i) describing functionalities that are central to data preparation pipelines, specifically profiling, matching, mapping, format transformation and data repair; and (ii) presenting how these capabilities surface in different approaches to data preparation, that involve programming, writing workflows, interacting with individual data sets as tables, and automating aspects of the process. These functionalities and approaches are illustrated with reference to a running example that combines open government data with web extracted real estate data.
- Open Access
- 23 November 2022
- Published by Association for Computing Machinery (ACM) in Journal of Data and Information Quality
- Vol. 14 (4) , 1-26
Fairness-aware Data Integration
Machine learning can be applied in applications that take decisions that impact people’s lives. Such techniques have the potential to make decision making more objective, but there also is a risk that the decisions can discriminate against certain groups as a result of bias in the underlying data. Reducing bias, or promoting fairness, has been a focus of significant investigation in machine learning, for example, based on pre-processing the training data, changing the learning algorithm, or post-processing the results of the learning. However, prior to these activities, data integration discovers and integrates the data that is used for training, and data integration processes have the potential to produce data that leads to biased conclusions. In this article, we propose an approach that generates schema mappings in ways that take into account: (i) properties that are intrinsic to mapping results that may give rise to bias in analyses; and (ii) bias observed in classifiers trained on the results of different sets of mappings. The approach explores a space of different ways of integrating the data, using a Tabu search algorithm, guided by bias-aware objective functions that represent different types of bias.The resulting approach is evaluated using Adult Census and German Credit datasets to explore the extent to which and the circumstances in which the approach can increase the fairness of the results of the data integration process.
- 23 November 2022
- Published by Association for Computing Machinery (ACM) in Journal of Data and Information Quality
- Vol. 14 (4) , 1-26
Open AccessFairness-aware Data Integration
Machine learning can be applied in applications that take decisions that impact people’s lives. Such techniques have the potential to make decision making more objective, but there also is a risk that the decisions can discriminate against certain groups as a result of bias in the underlying data. Reducing bias, or promoting fairness, has been a focus of significant investigation in machine learning, for example, based on pre-processing the training data, changing the learning algorithm, or post-processing the results of the learning. However, prior to these activities, data integration discovers and integrates the data that is used for training, and data integration processes have the potential to produce data that leads to biased conclusions. In this article, we propose an approach that generates schema mappings in ways that take into account: (i) properties that are intrinsic to mapping results that may give rise to bias in analyses; and (ii) bias observed in classifiers trained on the results of different sets of mappings. The approach explores a space of different ways of integrating the data, using a Tabu search algorithm, guided by bias-aware objective functions that represent different types of bias.The resulting approach is evaluated using Adult Census and German Credit datasets to explore the extent to which and the circumstances in which the approach can increase the fairness of the results of the data integration process.