Explore the vast range of titles available.

Multi-source heterogeneous knowledge graph fusion faces significant challenges due to schema heterogeneity, entity conflicts, and relationship inconsistencies across different knowledge sources. This paper proposes CausalFusion, a novel adaptive fusion algorithm that leverages causal discovery principles to guide the knowledge graph integration process. The algorithm incorporates a constraint-based causal discovery component specifically designed for relational data, an adaptive weight learning mechanism that dynamically adjusts source contributions based on causal strength, and a conflict resolution strategy that prioritizes causal consistency over statistical correlation. Experimental evaluation on benchmark datasets including DBpedia, Freebase, YAGO, and Wikidata demonstrates significant improvements in fusion quality, with the proposed method achieving 91.2% precision and 88.7% recall, outperforming state-of-the-art baselines by 1.9% and 1.5% respectively. The results validate the effectiveness of incorporating causal inference into knowledge graph fusion, particularly for preserving meaningful causal relationships while resolving heterogeneity conflicts.

Semantic schema alignment aims to match elements across a pair of schemas based on their semantic representation. It is a key primitive for data integration that facilitates the creation of a common data fabric across heterogeneous data sources. Deep learning approaches such as graph representation learning have shown promise for effective alignment of semantically rich schemas, often captured as ontologies. Most of these approaches are supervised and require large amounts of labeled training data, which is expensive in terms of cost and manual labor. Active learning (AL) techniques can alleviate this issue by intelligently choosing the data to be labeled utilizing a human-in-the-loop approach, while minimizing the amount of labeled training data required. However, existing active learning techniques are limited in their ability to utilize the rich semantic information from underlying schemas. Therefore, they cannot drive effective and efficient sample selection for human labeling that is necessary to scale to larger datasets. In this paper, we propose Alfa , an active learning framework to overcome these limitations. Alfa exploits the schema element properties as well as the relationships between schema elements (structure) to drive a novel ontology-aware sample selection and label propagation algorithm for training highly accurate alignment models. We propose semantic blocking to scale to larger datasets without compromising model quality. Our experimental results across three real-world datasets show that (1) Alfa leads to a substantial reduction (27–82%) in the cost of human labeling, (2) semantic blocking reduces label skew up to 40 × without adversely affecting model quality and scales AL to large datasets, and (3) sample selection achieves comparable schema matching quality (90% F1-score) to models trained on the entire set of available training data. We also show that Alfa outperforms the state-of-the-art ontology alignment system, BERTMap, in terms of (1) 10 × shorter time per AL iteration and (2) requiring half of the AL iterations to achieve the highest convergent F1-score.

Studying the surface water systems of the earth is important for many fields, from biology to agriculture to tourism. Much of the data relevant to surface water systems are stored in isolated repositories that interface with different ontologies, such as the US Geological Survey’s Surface Water Ontology or the Environment Ontology (ENVO). Effectively using these data requires integrating the ontologies so that the data can be seamlessly queried and analyzed. Automated alignment algorithms exist to facilitate this data integration challenge. In this paper we examine the utility of two leading automated alignment systems to integrate four pairs of ontologies from the surface water domain. We show that the performance of such systems in this domain lags behind their results on popular benchmarks, and therefore incorporate the alignment task described here into the set of benchmarks used by the alignment community. We also show that with minor modifications, existing alignment algorithms can be used effectively within a semi-automated system for the surface water domain. In addition, we analyze the unique challenges of this domain with respect to data integration and discuss possible solutions to pursue in order to address these challenges.

Purpose Ontology and Linked Data (LD) are the two prominent web technologies that have emerged in the recent past. Both of them are at the center of Semantic Web and its applications. Researchers and developers from both academia and business are actively working in these areas. The increasing interest in these technologies promoted the growth of LD sets and ontologies on the web. The purpose of this paper is to investigate the possible relationships between them. The effort is to investigate the possible roles that ontologies may play in further empowering the LD. In a similar fashion, the author also studies the possible roles that LD may play to empower ontologies. Design/methodology/approach The work is mainly carried out by exploring the ontology- and LD-based real-world systems, and by reviewing the existing literature. Findings The current work reveals, in general, that both the technologies are interdependent and have lots to offer to each other for their faster growth and meaningful development. Specifically, anything that we can do with LD, we can do more by adding an ontology to it. Practical implications The author envisions that the current work, in the one hand, will help in boosting the successful implementation and the delivery of semantic applications; on the other hand, it will also become a food for the future researchers in further investigating the relationships between the ontologies and LD. Originality/value So far, as per the author’s knowledge, there are very little works that have attempted in exploring the relationships between the ontologies and LD. In this work, the author illustrates the real-world systems that are based on ontology and LD, discusses the issues and challenges and finally illustrates their interdependency discussing some of the ongoing research works.

Ontologies are the backbone of intelligent computing on the World Wide Web but also crucial in many decision support situations. Many sophisticated tools have been developed to support working with ontologies, including prominently exploiting the vast array of existing ontologies. A system called ALIGN is developed that demonstrates how to use freely available tools to facilitate ontology alignment. First two ontologies are built with the ontology editor Protégé and represented in OWL. ALIGN then accesses these ontologies via Java’s JENA framework and SPARQL queries. The efficacy of the ALIGN prototype is demonstrated on a drug-drug interaction problem. The prototype could readily be applied to other domains or be incorporated into decision support tools.

Linking data by finding matching instances in different datasets requires considering many characteristics, such as structural heterogeneity, implicit knowledge, and URI (Uniform Resource Identifier)-oriented identification. The authors propose a context-independent approach to align Linked data through an alignment process based on the ontological model’s components and considering data’s multidimensionality. The researchers experimented with the proposed approach against two methods for aligning linked data in two datasets and evaluated precision, recall, and f-measure metrics. The authors also conducted a case study in a real scenario considering a Brazilian publication dataset on computers and education. This study’s results indicate that the proposed approach overcomes the other methods (regarding the precision, recall, and f-measure metrics), requiring less work when changing the dataset domain. This work’s main contributions include enabling real datasets to be semi-automatically linked, presenting an approach capable of calculating resource similarity.

Data integration is a classical problem in databases, typically decomposed into schema matching, entity matching and data fusion. To solve the latter, it is mostly assumed that ground truth can be determined. However, in general, the data gathering processes in the different sources are imperfect and cannot provide an accurate merging of values. Thus, in the absence of ways to determine ground truth, it is important to at least quantify how far from being internally consistent a dataset is. Hence, we propose definitions of concordant data and define a discordance metric as a way of measuring disagreement to improve decision-making based on trustworthiness. We define the discord measurement problem of numerical attributes in which given a set of uncertain raw observations or aggregate results (such as case/hospitalization/death data relevant to COVID-19) and information on the alignment of different conceptualizations of the same reality (e.g., granularities or units), we wish to assess whether the different sources are concordant, or if not, use the discordance metric to quantify how discordant they are. We also define a set of algebraic operators to describe the alignments of different data sources with correctness guarantees, together with two alternative relational database implementations that reduce the problem to linear or quadratic programming. These are evaluated against both COVID-19 and synthetic data, and our experimental results show that discordance measurement can be performed efficiently in realistic situations.

In this paper, we consider the problem of full schema induction from either multiple list pages or singleton pages with the same template. Existing approaches do not work well for this problem because they use fixed abstraction schemes that are suitable for data-rich detection, but they are not appropriate for small records and complex data found in other sections. We propose an unsupervised full schema web data extraction via Divide-and-Conquer Alignment with Dynamic Encoding (DCADE for short). We define the Content Equivalence Class (CEC) and Typeset Equivalence Class (TEC) based on leaf node content. We then combine HTML attributes (i.e., id and class) in the paths for various levels of encoding, so that the proposed algorithm can align leaf nodes by exploring patterns at various levels from specific to general. We conducted experiments on 49 real-world websites used in TEX and ExAlg. The proposed DCADE achieved a 0.962 F1 measure for non-recordset data extraction (denoted by FD), and a 0.936 F1 measure for recordset data extraction (denoted by FS), which outperformed other page-level web data extraction methods, i.e., DCA (FD= 0.660), TEX (FD= 0.454 and FS= 0.549), RoadRunner (FD= 0.396 and FS= 0.330), and UWIDE (FD= 0.260 and FS= 0.081).

Automatic data extraction from template pages is an essential task for data integration and data analysis. Most researches focus on data extraction from list pages. The problem of data alignment for singleton item pages (singleton pages for short), which contain detail information of a single item is less addressed and is more challenging because the number of data attributes to be aligned is much larger than list pages. In this paper, we propose a novel alignment algorithm working on leaf nodes from the DOM trees of input pages for singleton pages data extraction. The idea is to detect mandatory templates via the longest increasing sequence from the landmark equivalence class leaf nodes and recursively apply the same procedure to each segment divided by mandatory templates. By this divide-and-conquer approach, we are able to efficiently conduct local alignment for each segment, while effectively handle multi-order attribute-value pairs with a two-pass procedure. The results show that the proposed approach (called Divide-and-Conquer Alignment, DCA) outperforms TEX (Sleiman and Corchuelo 2013) and WEIR (Bronzi et al. VLDB 6(10):805–816 2013) 2% and 12% on selected items of TEX and WEIR dataset respectively. The improvement is more obvious in terms of full schema evaluation, with 0.95 (DCA) versus 0.63 (TEX) F-measure, on 26 websites from TEX and EXALG (Arasu and Molina 2003).

Integrating flood-response datasets across municipalities is often hindered by heterogeneous and non-standard variable names, a challenge amplified in Korean by local naming conventions and linguistic variation. This study addresses scalable schema alignment to standardize municipal flood datasets with reduced manual effort while maintaining semantic consistency for downstream modeling. We propose a BERT-based schema matching framework that augments standardized attribute names with paraphrases generated by a generative language model and filtered to reduce semantic drift. Both standardized and target variable names are encoded using a flood-domain-adapted Korean BERT model, and candidate correspondences are retrieved via cosine-similarity ranking to produce top-k match suggestions for automated or human-in-the-loop alignment. Experiments on real flood-related tables from Busan and Incheon, evaluated jointly to diversify variable expressions, show that augmentation substantially improves top-k retrieval accuracy. In the combined evaluation, Hit@5 improves from 0.71 to 0.95, supporting more reliable schema harmonization for simulation-ready inputs.