The field of Virtual Knowledge Graphs (VKGs) continues to grow in both academic and applied contexts. Yet, the issue of updates in VKG systems has not yet received adequate attention, although it is crucial to manage data modifications at the data source level through the lens of an ontology. In this paper, we focus on VKGs whose ontology is specified in the lightweight ontology language DL-Lite 𝐴 , and we propose diverse settings and research directions we intend to explore to address the challenge of translating ontology-based updates into updates at the level of data sources. We also pay attention to the important problem of automated analysis of mappings, which plays a major role when it comes to reformulating ontology-based update requests into update requests over the data sources.
Clinical Decision Support (CDS) is designed to provide medical guidance and decision support based on patient information. More and more patient data is being collected from medical devices and fitness trackers, and some decisions need to be made quickly to allow for timely medical treatment. In our work, we combine a Semantic Web rule-based approach for representing medical guidelines with medical data represented in the “Fast Healthcare Interoperability Resources” (FHIR) standard. We implement a complex event processing strategy to deliver near-real-time decisions for a physician. This is relevant in intensive care and emergency medicine in particular, as medical treatment is time-critical and can decide on life or death of the patients. Different approaches exist for real-time clinical decision support. However, the integration of FHIR and Semantic Web technologies is missing. By integrating FHIR, medical ontologies, and Prova [1] as a semantic rule engine, we can define the medical guideline and include hooks for data extraction and decision points. We describe the entire process, starting with the medical device submitting the FHIR observations, through the data handling and decision points, to the notification of the attending physician. We extend the functionality of the existing rule engine for data handling and decision-making. In this work, we outline a solution that is capable of clinical decision support, including FHIR, and makes use of a semantic rule engine that allows medical guidelines to be expressed in RuleML. With these features, we also include the option of real-time complex event processing.
Ontology-based data access (OBDA) facilitates access to heterogeneous data sources through the mediation of an ontology (e.g. OWL), which captures the domain of interest and is connected to data sources through a declarative mapping. In our study, large, heterogeneous earth observational (EO) data, known as raster data , and geometrical data, known as vector data , are considered as (heterogeneous) data sources. Raster data represent, e.g., Earth’s natural phenomena, such as surface temperature, elevation, or air pollution, as multidimensional arrays. In contrast, vector data depict, e.g., locations, networks, or regions on Earth, using geometries. Domain experts, such as earth scientists and GIS practitioners, still struggle to undertake advanced studies by querying large raster and vector data in an integrated way because, unlike relational data, they come in diverse formats and different data structures. In our approach to integration, we use a geospatial extension of an RDBMS to represent vector data as relational data, and a domain-agnostic array DBMS to handle raster data. Our aim is to extend the OBDA paradigm to effectively deal with relational, vector, and raster data in a combined way, while leveraging the built-in capabilities of data management tools relevant to each type of data. We also plan to develop techniques to calculate on the fly for each user query posed over the ontology an optimal query plan that exploits, at best, the query processing capabilities of each tool, while limiting costly data transfer operations between tools.
Operational decisions are an important part of knowledge-intensive organizations, as these are taken in a high volume on a daily basis. For this purpose, the Decision Model and Notation (DMN) standard describes an intuitive and user-friendly notation to model, communicate and execute business decisions. However, manually modeling the decisions in DMN remains a time-consuming task, as various textual sources need to be analyzed. As such, automation of decision modeling is beneficial for the decision process. This work investigates an automated approach to generating decision tables from natural language based on the GPT-3 large language model. Through a total of 72 experiments over six problem descriptions, this work evaluates GPT-3’s decision logic modeling and reasoning capabilities. While GPT-3 demonstrates promising abilities in extracting decision context and identifying relevant variables from natural language, further enhancements are needed to improve its decision table capabilities for efficient automation of DMN modeling.
The Beers Criteria, widely used by healthcare professionals, list so-called Potentially Inappropriate Medications (PIMs) which older adults in certain circumstances should avoid. Manually identifying medications that belong to the Beers Criteria can be time-consuming and error-prone, as the criteria are complex and subject to frequent updates. Moreover, it is not available in a (formal) representation that health systems can interpret and reason with automatically. This paper proposes an ontology as a formal representation of the Beers Criteria, and describes the elements and the taxonomy underlying the ontology. We include inference rules to enable automated detection and categorisation of drugs classified as PIMs. By automatically detecting drugs that belong to the Beers Criteria, the ontology, once linked with decision support systems, can be used to support healthcare providers in ensuring that older adults receive safe and effective medical care.
Large Language Models (LLMs) have gained prominence in the AI landscape due to their exceptional performance. Thus, it is essential to gain a better understanding of their capabilities and limitations, among others in terms of nonmonotonic reasoning. This paper proposes a benchmark that corresponds to various defeasible rule-based reasoning patterns. We modified an existing benchmark for defeasible logic reasoners by translating defeasible rules into text suitable for LLMs. We conducted preliminary experiments on nonmonotonic rule-based reasoning using ChatGPT and compared it with reasoning patterns defined by defeasible logic.
In ontology-mediated query answering, access to incomplete data sources is mediated by a conceptual layer constituted by an ontology, which can be formulated in a description logic (DL) or using existential rules. In the literature, there exists a multitude of complex techniques for incorporating ontological knowledge into queries. However, few of these approaches were designed for explainability of the query answers. We tackle this challenge by adapting an existing proof framework toward conjunctive query answering, based on the notion of universal models. We investigate the data and combined complexity of determining the existence of a proof below a given quality threshold, which can be measured in different ways. By distinguishing various parameters such as the shape of the query, we obtain an overview of the complexity of this problem for several Horn DLs.
In this paper, we extend the KLM approach to defeasible reasoning beyond the propositional setting. We do so by making it applicable to a restricted version of first-order logic. We describe defeasibility for this logic using a set of rationality postulates, provide a suitable and intuitive semantics for it, and present a representation result characterising the semantic description of defeasibility in terms of our postulates. An advantage of our semantics is that it is sufficiently general to be applicable to other restricted versions of first-order logic as well. Based on this theoretical core, we then propose a version of defeasible entailment that is inspired by the well-known notion of Rational Closure as it is defined for defeasible propositional logic and defeasible description logics. We show that this form of defeasible entailment is rational in the sense that it adheres to the full set of rationality postulates.
Autonomous vehicles require highly sophisticated decision-making to determine their motion. This paper describes how such functionality can be achieved with a practical rule engine learned from expert driving decisions. We propose an algorithm to create and maintain a rule-based behaviour planner, using a two-layer rule-based theory. The first layer determines a set of feasible parametrized behaviours, given the perceived state of the environment. From these, a resolution function chooses the most conservative high-level maneuver. The second layer then reconciles the parameters into a single behaviour. To demonstrate the practicality of our approach, we report results of its implementation in a level-3 autonomous vehicle and its field test in an urban environment.
LegalRuleML is a comprehensive XML-based representation framework for modeling and exchanging normative rules. The TPTP input and output formats, on the other hand, are general-purpose standards for the interaction with automated reasoning systems. In this paper we provide a bridge between the two communities by (i) defining a logic-pluralistic normative reasoning language based on the TPTP format, (ii) providing a translation scheme between relevant fragments of LegalRuleML and this language, and (iii) proposing a flexible architecture for automated normative reasoning based on this translation. We exemplarily instantiate and demonstrate the approach with three different normative logics.
Counterfactual reasoning has been the subject of extensive study in philosophy, logics, and AI. The connection between counterfactual reasoning and theory revision is well-known since Ramsey’s intuition, according to which “to find out whether the counterfactual ‘if A were true, then B would be true’ is satisfied in a state S, change the state S minimally to include A, and test whether B is satisfied in the resulting state". In this paper we study how to model this idea in Defeasible Logic for devising logics for counterfactual reasoning and suitable selection function models.
DatalogMTL is an extension of Datalog with metric temporal operators that has found applications in temporal ontology-based data access and query answering, as well as in stream reasoning. Practical algorithms for DatalogMTL are reliant on materialisation-based reasoning, where temporal facts are derived in a forward chaining manner in successive rounds of rule applications. Current materialisation-based procedures are, however, based on a naïve evaluation strategy, where the main source of inefficiency stems from redundant computations. In this paper, we propose a materialisation-based procedure which, analogously to the classical seminaïve algorithm in Datalog, aims at minimising redundant computation by ensuring that each temporal rule instance is considered at most once during the execution of the algorithm. Our experiments show that our optimised seminaïve strategy for DatalogMTL is able to significantly reduce materialisation times.
The representation of a dynamic problem in ASP usually boils down to using copies of variables and constraints, one for each time stamp, no matter whether it is directly encoded or via an action or temporal language. The multiplication of variables and constraints is commonly done during grounding and the solver is completely ignorant about the temporal relationship among the different instances. On the other hand, a key factor in the performance of today's ASP solvers is conflict-driven constraint learning. Our question is now whether a constraint learned for particular time steps can be generalized and reused at other time stamps, and ultimately whether this enhances the overall solver performance on dynamic problems. Knowing full well the domain of time, we study conditions under which learned dynamic constraints can be generalized and propose a simple translation of the original logic program such that, for the translated programs, all learned constraints can be generalized to other time points. Last but not least, we empirically evaluate the impact of adding the generalized constraints to an ASP solver.
The Logic of “Inferable” L-DINF has been recently proposed as a declarative framework to formally model via epistemic logic (aspects of) the group dynamics of cooperative agents. The framework permits to model groups of cooperative agents that can jointly perform actions. Various aspects of Multi-Agent Systems can be formalized in L-DINF, such as costs of actions, agents’ preferences, and roles of agents within groups. In this paper we extend the framework by introducing the possibility of cooperation among different groups: if a group has not an agent with the right role for performing an action, the group can ask another group in order to have the action performed.
