Understanding the internal mechanisms of deep neural networks remains a central challenge in the field of Explainable Artificial Intelligence (XAI). With the rapid advancement of neural architectures in natural language processing (NLP), analyzing the role of hidden neurons in capturing and processing linguistic features has become increasingly important. This study investigates Hidden Neuron Activation Analysis on labeled text data to reveal how individual neurons contribute to a model’s decision-making process. We propose a model-agnostic explainability framework for text classifiers that identifies concepts activating specific neurons involved in classification. An LSTM-based network is trained on the AG News topic classification dataset, comprising four distinct classes, and the final Dense layer with 64 neurons was analyzed. In addition, statistical analyses like the Mann-Whitney U Test is conducted to assess the robustness and reliability of the system. The statistical analysis shows that, concepts plays important role in the decision making process of neural network. Our findings enhance interpretability in NLP models and offer a foundation for optimizing neural architectures in text classification tasks.

Recent advances in Large Language Models (LLMs) have drawn interest in their capacity for logical reasoning, an area traditionally dominated by symbolic systems that rely on complete, manually curated knowledge bases represented in formal languages. This paper introduces a framework that leverages pretrained LLMs to generate Description Logic (DL) class expressions from instance-level examples and background knowledge, translated to natural language. The baseline is Concept Induction, a symbolic learning approach that is mostly based on formal logical reasoning over a DL theory. Drawing inspiration from the DL-Learner architecture, our approach replaces traditional symbolic methods with LLM-based models to generate DL class expressions from instance-level data. We evaluate our approach using three benchmark ontologies across two LLMs: gpt-4o and o3-mini. We use a symbolic reasoner, Pellet, to verify the LLM-generated results and incorporate the reasoner’s feedback into our pipeline to ensure logical consistency, thereby generating a hybrid neurosymbolic system. By introducing controlled variations to the background knowledge, we assess the models’ reliance on commonsense versus formal reasoning. Results show that o3-mini achieves near-perfect accuracy across settings, albeit with longer runtime. These findings demonstrate that LLMs have the potential to serve as scalable and flexible DL learners when coupled in a hybrid neurosymbolic setting, offering a promising alternative to symbolic approaches—particularly in contexts where high-quality ontologies are incomplete or unavailable.

Depression remains a pervasive mental health disorder that demands prompt diagnosis and intervention. Although social media data presents a promising avenue for early detection, traditional deep neural models are frequently critiqued for their lack of interpretability and susceptibility to bias. We introduce ProtoDep—a neurosymbolic framework that integrates clinically grounded categorizations (e.g., PHQ-9 symptoms) with large language model–assisted prototype learning. Unlike conventional black-box models, ProtoDep aligns individual tweets with symptom-level prototypes, offering interpretable explanations at three levels: (i) symptom-level insights that map user posts to recognized depressive patterns, (ii) case-based reasoning that compares users to representative prototype profiles, and (iii) transparent concept-level decisions, wherein classification at inference time is driven by the distances between the user profile and prototype user and symptom clusters, yielding clear, quantifiable explanations. By integrating symbolic mental health constructs with neural embeddings, ProtoDep achieves a mean F1-score of 94% across five benchmark datasets and establishes a foundation for interpretable depression screening pipelines with potential applicability in clinical settings.

Large Language Models bear the promise of significant acceleration of key Knowledge Graph and Ontology Engineering tasks, including ontology modeling, extension, modification, population, alignment, as well as entity disambiguation. We lay out LLM-based Knowledge Graph and Ontology Engineering as a new and coming area of research, and argue that modular approaches to ontologies will be of central importance.

KnowWhereGraph is one of the largest fully publicly available geospatial knowledge graphs. It includes data from 30 layers on natural hazards (e.g., hurricanes, wildfires), climate variables (e.g., air temperature, precipitation), soil properties, crop and land-cover types, demographics, and human health, various place and region identifiers, among other themes. These have been leveraged through the graph by a variety of applications to address challenges in food security and agricultural supply chains; sustainability related to soil conservation practices and farm labor; and delivery of emergency humanitarian aid following a disaster. In this paper, we introduce the ontology that acts as the schema for KnowWhereGraph. This broad overview provides insight into the requirements and design specifications for the graph and its schema, including the development methodology (modular ontology modeling) and the resources utilized to implement, materialize, and deploy KnowWhereGraph with its end-user interfaces and public query SPARQL endpoint.

Wikibase – which is the software underlying Wikidata – is a powerful platform for knowledge graph creation and management. However, it has been developed with a crowd-sourced knowledge graph creation scenario in mind, which in particular means that it has not been designed for use case scenarios in which a tightly controlled high-quality schema, in the form of an ontology, is to be imposed, and indeed, independently developed ontologies do not necessarily map seamlessly to the Wikibase approach. In this paper, we provide the key ingredients needed in order to combine traditional ontology modeling with use of the Wikibase platform, namely a set of axiom patterns that bridge the paradigm gap, together with usage instructions and a worked example for historical data.

Many methodologies and platforms for creating, deploying, and defining the manner of knowledge graphs are available. For this paper, we single out the platform, Wikibase. Using Wikibase comes with many advantages: out-of-the-box software for de-referencing, a convenient user interface, a consistent way to track and record provenance and lineage, and the ability to execute SPARQL queries against an RDF representation of the knowledge graph. However, the provenance mechanism and the exact nature of the structure of the Wikibase representation can complicate developing a principled schema for knowledge graphs, as well as the approach to the materialization of the data for upload to the platform. In this paper, we detail the methodology used to design, implement, and deploy the Enslaved.Org Hub, a nationally recognized knowledge graph for documenting the peoples of the historical slave trade.

Abstract Nominal schemas have been proposed as an extension to Description Logics (DL), the knowledge representation paradigm underlying the Web Ontology Language (OWL). They provide for a very tight integration of DL and rules. Nominal schemas can be understood as syntactic sugar on top of OWL. However, this naive perspective leads to inefficient reasoning procedures. In order to develop an efficient reasoning procedure for the language ${\mathcal {E}\mathcal {L}\mathcal {V}^{++}}$, which results from extending the OWL profile language OWL EL with nominal schemas, we propose a transformation from ${\mathcal {E}\mathcal {L}\mathcal {V}^{++}}$ ontologies into Datalog-like rule programs that can be used for satisfiability checking and assertion retrieval. The use of this transformation enables the use of powerful Datalog engines to solve reasoning tasks over ${\mathcal {E}\mathcal {L}\mathcal {V}^{++}}$ ontologies. We implement and then evaluate our approach on several real-world, data-intensive ontologies, and find that it can outperform state-of-the-art reasoners such as Konclude and ELK. As a lesser side result we also provide a self-contained description of a rule-based algorithm for ${\mathcal {E}\mathcal {L}^{++}}$, which does not require a normal form transformation.

Knowledge graphs are a rapidly growing paradigm and technology stack for integrating large-scale, heterogeneous data in an AI-ready form, i.e., combining data with the formal semantics required to understand it. However, toolchains that support data synthesis and knowledge discovery through information organization, search, filtering, and visualization have been developed at a pace lagging knowledge graph technology. In this paper, we present Knowledge Explorer, an open-source faceted search interface that provides environmentally intelligent services for interactively browsing and navigating KnowWhereGraph. Currently one of the largest open knowledge graphs, KnowWhereGraph contains over 12 billion statements with rich spatial and temporal information from more than 30 data layers. With an extensive collection of facets, Knowledge Explorer enables spatial, temporal, full-text, and expert search with dereferencing functionality to support "follow-your-nose"exploration, and it allows users to narrow their search by selecting facets. Given the size of the underlying graph and dependency on GeoSPARQL, we have improved query performance by implementing Elasticsearch indexing, spatial query generation, and caching. Knowledge Explorer is capable of retrieving information within seconds, answering a wide variety of competency questions posed by researchers, humanitarian relief organizations, and the broader public, thus helping better perform tasks such as cross-gazetteer place retrieval and disaster assessment from global to local geographic scales.

Powered by new advances in sensor development and artificial intelligence, the decreasing cost of computation, and the pervasiveness of handheld computation devices, biometric user authentication (and identification) is rapidly becoming ubiquitous. Modern approaches to biometric authentication, based on sophisticated machine learning techniques, cannot avoid storing either trained-classifier details or explicit user biometric data, thus exposing users’ credentials to falsification. In this paper, we introduce a secure way to handle user-specific information involved with the use of artificial neural networks for biometric authentication. Our proposed architecture, called a Neural Fuzzy Extractor (NFE), allows the coupling of pre-existing classifiers with fuzzy extractors, through an artificial-neuralnetwork-based buffer called an expander, with minimal or no performance degradation. The NFE thus offers all the performance advantages of modern deep-learningbased classifiers and all the security of standard fuzzy extractors. We demonstrate the NFE retrofit of a few classic artificial neural networks, for simple biometric authentication scenarios.

We propose the combining of region connection calculi with nested hierarchical grids for representing spatial region data in the context of knowledge graphs, thereby avoiding reliance on vector representations. We present a resulting region calculus, and provide qualitative and formal evidence that this representation can be favorable with large data volumes in the context of knowledge graphs; in particular we study means of efficiently choosing which triples to store to minimize space requirements when data is represented this way, and we provide an algorithm for finding the smallest possible set of triples for this purpose including an asymptotic measure of the size of this set for a special case. We prove that a known constraint calculus is adequate for the reconstruction of all triples describing a region from such a pruned representation, but problematic for reasoning with hierarchical grids in general.

There has been a rapid growth in the use of symbolic representations along with their applications in many important tasks. Symbolic representations, in the form of Knowledge Graphs (KGs), constitute large networks of real-world entities and their relationships. On the other hand, sub-symbolic artificial intelligence has also become a mainstream area of research. This workshop brought together researchers to discuss and foster collaborations on the intersection of these two areas.

Since over a decade coreference resolution systems have been developed in order to find simple 1-to-1 equivalent mapping (sameAs relations) between instances of different linked datasets and knowledge graphs. Comparative evaluations of instance matching systems can inform us about the performance of such systems regarding artificial benchmarks or real-world data challenges. However, the lack of real data for evaluating these systems is currently a bottleneck. In this paper, we propose the use of the Cruise entities in the GeoLink data repository as a real-world instance matching benchmark for linked data and knowledge graphs. The GeoLink project has brought together seven datasets related to geoscience research. Both the ontology (T-box) and the instance data (A-box) of GeoLink are significantly larger than current benchmarks, and they have particularly interesting challenges, such as geospatial and temporal data. The benchmark we propose here consists of two real-world datasets in GeoLink called R2R data and BCO-DMO which includes manual curated owl:sameAs links between more than 900 Cruise entities of these two datasets. The reference alignment was discussed and generated by domain experts from different institutions and is expressed in the Alignment API format.

Ontology alignment has taken a critical place for helping heterogeneous resources to interoperate.It has been studied for over a decade, and over that time many alignment systems and methods have been developed by researchers to find simple 1:1 equivalence matches between two ontologies.However, very few alignment systems focus on finding complex correspondences.Even if the complex alignment systems are developed, the performance of finding complex relations still has a lot of room for improvement.One reason for this limitation may be that there are still few applicable alignment benchmarks that contain such complex relationships that can raise researchers' interests.In this paper, we propose a real-world dataset from the Enslaved project as a potential complex alignment benchmark.The benchmark consists of two resources, the Enslaved Ontology along with a Wikibase repository holding a large number of instance data from the Enslaved project, as well as a manually created reference alignment between them.The alignment was developed in consultation with domain experts in the digital humanities.The alignment not only includes simple 1:1 equivalence correspondences, but also more complex m:n equivalence and subsumption correspondences and are provided in both Expressive and Declarative Ontology Alignment Language (EDOAL) format and rule syntax.The Enslaved benchmark has been incorporated into the Ontology Alignment Evaluation Initiative (OAEI) 2020 and is completely free for public use to assist the researchers in developing and evaluating their complex alignment algorithms.

Concept Induction refers to the problem of creating complex Description Logic class descriptions (i.e., TBox axioms) from instance examples (i.e., ABox data). In this paper we look particularly at the case where both a set of positive and a set of negative instances are given, and complex class expressions are sought under which the positive but not the negative examples fall. Concept induction has found applications in ontology engineering, but existing algorithms have fundamental performance issues in some scenarios, mainly because a high number of invokations of an external Description Logic reasoner is usually required. In this paper we present a new algorithm for this problem which drastically reduces the number of reasoner invokations needed. While this comes at the expense of a more limited traversal of the search space, we show that our approach improves execution times by up to several orders of magnitude, while output correctness, measured in the amount of correct coverage of the input instances, remains reasonably high in many cases. Our approach thus should provide a strong alternative to existing systems, in particular in settings where other systems are prohibitively slow.

Today's scientific investigations are producing large numbers of scholarly products. These products continue to increase in diversity and complexity as researchers recognize that scholarly achievements are not only published articles but also datasets, software, and associated supporting materials. OceanLink is an online platform that addresses scholarly discovery and collaboration in the ocean sciences. The OceanLink project leverages Semantic Web technologies, web mining, and crowdsourcing to identify links between data centers, digital repositories, and professional societies to enhance discovery, enable collaboration, and begin to assess research contribution.

The Semantic Web journal by IOS Press follows a unique open and transparent process during which each submitted manuscript is available online together with the full history of its successive decision statuses, assigned editors, solicited and voluntary reviewers, their full text reviews, and in many cases also the authors’ response letters. Combined with a highly-customized, Drupal-based journal management system, this provides the journal with semantically rich manuscript time lines and networked data about authors, reviewers, and editors. These data are now exposed using a SPARQL endpoint, an extended Bibo ontology, and a modular Linked Data portal that provides interactive scientometrics based on established and new analysis methods. The portal can be customized for other journals as well. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Linked Open Data (LOD) has emerged as one of the largest collections of interlinked structured datasets on the Web. Although the adoption of such datasets for applications is increasing, identifying relevant datasets for a specific task or topic is still challenging. As an initial step to make such identification easier, we provide an approach to automatically identify the topic domains of given datasets. Our method utilizes existing knowledge sources, more specifically Freebase, and we present an evaluation which validates the topic domains we can identify with our system. Furthermore, we evaluate the effectiveness of identified topic domains for the purpose of finding relevant datasets, thus showing that our approach improves reusability of LOD datasets.

Description logics (DLs) have become a prominent paradigm for representing knowledge in a variety of application areas, partly due to their ability to achieve a favourable balance between expressivity of the logic and performance of reasoning. Horn description logics are obtained, roughly speaking, by disallowing all forms of disjunctions. They have attracted attention since their (worst-case) data complexities are in general lower than those of their non-Horn counterparts, which makes them attractive for reasoning with large sets of instance data (ABoxes). It is therefore natural to ask whether Horn DLs also provide advantages for schema (TBox) reasoning, that is, whether they also feature lower combined complexities. This article settles this question for a variety of Horn DLs. An example of a tractable Horn logic is the DL underlying the ontology language OWL RL, which we characterize as the Horn fragment of the description logic SROIQ without existential quantifiers. If existential quantifiers are allowed, however, many Horn DLs become intractable. We find that Horn- ALC already has the same worst-case complexity as ALC , that is, ExpTime , but we also identify various DLs for which reasoning is PSpace -complete. As a side effect, we derive simplified syntactic definitions of Horn DLs for which we exploit suitable normal form transformations.

Ontology design patterns ease the engineering of ontologies, improve their quality, foster reusability, and support the alignment of ontologies by acting as common building blocks or strategies for reoccurring modeling problems. This makes ontology design patterns key enablers of semantic interoperability and, hence, a crucial technology for representing the body of knowledge of such heterogeneous domains as the geosciences. While different types of patterns can be distinguished, existing work on geo-ontology design patterns has solely focused on content patterns, i.e., design solutions for domain classes and relationships. In this work, we propose a logical pattern that addresses a frequent modeling problem that has hampered the development of sophisticated geo-ontologies in the past, namely how to model the quantification over types. We argue for the need for such a pattern, explain why it is difficult to model, demonstrate how to implement it using the Web Ontology Language OWL, and finally show how it can be applied to modeling concepts such as biodiversity.

We propose a description-logic style extension of OWL 2 with nominal schemas which can be used like variable nominal classes within axioms. This feature allows ontology languages to express arbitrary DL-safe rules (as expressible in SWRL or RIF) in their native syntax. We show that adding nominal schemas to OWL 2 does not increase the worst-case reasoning complexity, and we identify a novel tractable language SROELV3(∩, x) that is versatile enough to capture the lightweight languages OWL EL and OWL RL.

Measuring inconsistency in knowledge bases has been recognized as an important problem in several research areas. Many methods have been proposed to solve this problem and a main class of them is based on some kind of paraconsistent semantics. However, existing methods suffer from two limitations: (i) they are mostly restricted to propositional knowledge bases; (ii) very few of them discuss computational aspects of computing inconsistency measures. In this article, we try to solve these two limitations by exploring algorithms for computing an inconsistency measure of first-order knowledge bases. After introducing a four-valued semantics for first-order logic, we define an inconsistency measure of a first-order knowledge base, which is a sequence of inconsistency degrees. We then propose a precise algorithm to compute our inconsistency measure. We show that this algorithm reduces the computation of the inconsistency measure to classical satisfiability checking. This is done by introducing a new semantics, named S[n]-4 semantics, which can be calculated by invoking a classical SAT solver. Moreover, we show that this auxiliary semantics also gives a direct way to compute upper and lower bounds of inconsistency degrees. That is, it can be easily revised to compute approximating inconsistency measures. The approximating inconsistency measures converge to the precise values if enough resources are available. Finally, by some nice properties of the S[n]-4 semantics, we show that some upper and lower bounds can be computed in P-time, which says that the problem of computing these approximating inconsistency measures is tractable.

We discuss a number of distance functions encountered in the theory of computation, including metrics, ultra-metrics, quasi-metrics, generalized ultra-metrics, partial metrics, d-ultra-metrics and generalized metrics. We consider their properties, associated fixed-point theorems and some general applications they have within the theory of computation. We consider in detail the applications of generalized distance functions in giving a uniform treatment of several important semantics for logic programs, including acceptable programs and natural generalizations of them, and also the supported model and the stable model in the context of locally stratified extended disjunctive logic programs and databases.

With the advent of the Semantic Web, the question becomes important how to best combine open-world based ontology languages, like OWL, with closed-world rules paradigms. One of the most mature proposals for this combination is known as Hybrid MKNF knowledge bases [11], which is based on an adaptation of the stable model semantics to knowledge bases consisting of ontology axioms and rules. In this paper, we propose a well-founded semantics for such knowledge bases which promises to provide better efficiency of reasoning, which is compatible both with the OWL-based semantics and the traditional well-founded semantics for logic programs, and which surpasses previous proposals for such a well-founded semantics by avoiding some issues related to inconsistency handling.

We introduce description logic (DL) rules as a new rule-based formalism for knowledge representation in DLs. As a fragment of the Semantic Web Rule Language SWRL, DL rules allow for a tight integration with DL knowledge bases. In contrast to SWRL, however, the combination of DL rules with expressive description logics remains decidable, and we show that the DL 𝒮ℛ𝒪ℐ𝒬 – the basis for the ongoing standardisation of OWL 2 – can completely internalise DL rules. On the other hand, DL rules capture many expressive features of 𝒮ℛ𝒪ℐ𝒬 that are not available in simpler DLs yet. While reasoning in 𝒮ℛ𝒪ℐ𝒬 is highly intractable, it turns out that DL rules can be introduced to various lightweight DLs without increasing their worst-case complexity. In particular, DL rules enable us to significantly extend the tractable DLs ℰℒ++and DLP.

Increased availability of mobile computing, such as personal digital assistants (PDAs), creates the potential for constant and intelligent access to up-to-date, integrated and detailed information from the Web, regardless of one's actual geographical position.Intelligent question-answering requires the representation of knowledge from various domains, such as the navigational and discourse context of the user, potential user questions, the information provided by Web services and so on, for example in the form of ontologies.Within the context of the SmartWeb project, we have developed a number of domain-specific ontologies that are relevant for mobile and intelligent user interfaces to open-domain question-answering and information services on the Web.To integrate the various domain-specific ontologies, we have developed a foundational ontology, the SmartSUMO ontology, on the basis of the DOLCE and SUMO ontologies.This allows us to combine all the developed ontologies into a single SmartWeb Integrated Ontology 1 (SWIntO) having a common modeling basis with conceptual clarity and the provision of ontology design patterns for modeling consistency.In this paper, we present SWIntO, describe the design choices we made in its construction, illustrate the use of the ontology through a number of applications, and discuss some of the lessons learned from our experiences.1 http://www.smartweb-project.org 2 SmartWeb is a large international

A new methodology has been proposed which allows one to derive uniform characterizations of different declarative semantics for logic programs with negation. One result from this work is that the well-founded semantics can formally be understood as a stratified version of the Fitting (or Kripke–Kleene) semantics. The constructions leading to this result, however, show a certain asymmetry which is not readily understood. We study this situation here with the result that we obtain a coherent picture of relations between different semantics for normal logic programs.

Part of the theory of logic programming and nonmonotonic reasoning concerns the study of fixed-point semantics for these paradigms. Several different semantics have been proposed during the last two decades, and some have been more successful and acknowledged than others. The rationales behind those various semantics have been manifold, depending on one's point of view, which may be that of a programmer or inspired by commonsense reasoning, and consequently the constructions which lead to these semantics are technically very diverse, and the exact relationships between them have not yet been fully understood. In this paper, we present a conceptually new method, based on level mappings, which allows to provide uniform characterizations of different semantics for logic programs. We will display our approach by giving new and uniform characterizations of some of the major semantics, more particular of the least model semantics for definite programs, of the Fitting semantics, and of the well-founded semantics. A novel characterization of the weakly perfect model semantics will also be provided.