The OWL 2 QL profile of the OWL 2 Web Ontology Language, based on the family of description logics called DL-Lite, is designed so that data stored in a standard relational database system (RDBMS) can be queried through an ontology via a rewriting mechanism, i.e., by rewriting the query into an SQL query that is then answered by the RDBMS system, without any changes to the data. In this paper we propose a language whose expressive power goes beyond that of DL-Lite while still allowing query answering via rewriting of queries into unions of conjunctive two-way regular path queries (UC2RPQs) instead of SQL queries. Our language is an extension of both OWL 2 QL and linear ELH: OWL 2 QL is extended by allowing qualified existential quantification on the left-hand side of concept inclusion axioms, and linear ELH by allowing inverses in role inclusion axioms. We identify a syntactic property of the extended language that guarantees UC2RPQ-rewritability. We propose a novel rewriting technique for conjunctive queries (CQs) under our ontology language that makes use of nondeterministic finite state automata. We show that CQ answering in our setting is NLOGSPACE-complete with respect to data complexity and NP-complete for combined complexity; we also show that answering instance queries is NLOGSPACE-complete for data complexity and in PTIME for combined complexity.

Decision dnnf (a.k.a. ∧_d-fbdd) is an important special case of Decomposable Negation Normal Form (dnnf), a landmark knowledge compilation model. Like other known dnnf restrictions, Decision dnnf admits fpt sized representation of cnfs of bounded primal treewidth. However, unlike other restrictions, the complexity of representation for cnfs of bounded incidence treewidth is wide open. In[arxiv:1708.07767], we resolved this question for two restricted classes of Decision dnnf that we name ∧_d-obdd and Structured Decision dnnf. In particular, we demonstrated that, while both these classes have fpt-sized representations for cnfs of bounded primal treewidth, they need xp-size for representation of cnfs of bounded incidence treewidth. In the main part of this paper we carry out an in-depth study of the ∧_d-obdd model. We formulate a generic methodology for proving lower bounds for the model. Using this methodology, we reestablish the xp lower bound provided in [arxiv:1708.07767]. We also provide exponential separations between fbdd and ∧_d-obdd and between ∧_d-obdd and an ordinary obdd. We study the complexity of Apply operation for ∧_d-obdd. While, in general, the Apply operation leads to exponential blow up of the resulting model, we identify a special restricted case where the Apply operation can be carried out efficiently. We introduce a relaxed version of Structured Decision dnnf that we name Structured ∧_d-fbdd and demonstrate that this model is quite powerful for cnfs of bounded incidence treewidth.

The Deep Web is constituted by data that are accessible through Web pages, but not readily indexable by search engines as they are returned in dynamic pages. In this paper we propose a conceptual framework for answering keyword queries on Deep Web sources represented as relational tables with so-called access limitations. We formalize the notion of optimal answer, characterize queries for which an answer can be found, and present a method for query processing based on the construction of a query plan that minimizes the accesses to the data sources.

Credit scoring models are critical for financial institutions to assess borrower risk and maintain profitability. Although machine learning models have improved credit scoring accuracy, imbalanced class distributions remain a major challenge. The widely used Synthetic Minority Oversampling TEchnique (SMOTE) struggles with high-dimensional, non-linear data and may introduce noise through class overlap. Generative Adversarial Networks (GANs) have emerged as an alternative, offering the ability to model complex data distributions. Conditional Wasserstein GANs (cWGANs) have shown promise in handling both numerical and categorical features in credit scoring datasets. However, research on extracting latent features from non-linear data and improving model explainability remains limited. To address these challenges, this paper introduces the Non-parametric Oversampling Technique for Explainable credit scoring (NOTE). The NOTE offers a unified approach that integrates a Non-parametric Stacked Autoencoder (NSA) for capturing non-linear latent features, cWGAN for oversampling the minority class, and a classification process designed to enhance explainability. The experimental results demonstrate that NOTE surpasses state-of-the-art oversampling techniques by improving classification accuracy and model stability, particularly in non-linear and imbalanced credit scoring datasets, while also enhancing the explainability of the results.

We demonstrate that Regular Resolution is FPT for two restricted families of CNFs of bounded incidence treewidth. The first includes CNFs having at most $p$ clauses whose removal results in a CNF of primal treewidth at most $k$. The parameters we use in this case are $p$ and $k$. The second class includes CNFs of bounded one-sided (incidence) treewdth, a new parameter generalizing both primal treewidth and incidence pathwidth. The parameter we use in this case is the one-sided treewidth.

The OWL 2 QL profile of the OWL 2 Web Ontology Language, based on the family of description logics called DL-Lite, allows for answering queries by rewriting, i.e. by reformulating a given query into another query that is then directly processed by a RDBMS system by pure querying, without materialising new data or updating existing data. In this paper we propose a new language whose expressive power goes beyond that of DL-Lite (in particular, our language extends both OWL 2 QL and linear ℰℒℋ , two well known DL ontology languages) while still allowing query answering via rewriting of queries into conjunctive two-way regular path queries (C2RPQs). Our language is identified by a syntactic property that can be efficiently checked. After defining our new language, we propose a novel rewriting technique for conjunctive queries (CQs) that makes use of nondeterministic finite state automata. CQ answering in our setting is NLogSpace-complete in data complexity and NP-complete in combined complexity; answering instance queries is NLogSpace-complete in data complexity and in PTime in combined complexity.

Given an arbitrary set A of two-dimensional points over a totally-ordered domain, a two-sided planar range query consists on finding all points of A within an arbitrary quadrant. In this paper we present a novel data structure that uses linear space in |A| while allowing for two-dimensional orthogonal range queries with logarithmic pre-processing and constant-delay enumeration.

The Deep Web is the large corpus of data accessible on the Web through forms and presented in dynamically-generated pages, but not indexable as static pages, and therefore invisible to search engines. Deep Web data are usually modelled as relations with so-called access limitations, that is, they can be queried only by selecting certain attributes. In this paper we give some fundamental complexity results on the problem of processing conjunctive (select-project-join) queries on relational data with access limitations.

The Deep Web is constituted by data accessible through dynamic Web pages requested through a Web interface. While Deep Web data sources have been usually modelled as relational in the literature, in many cases it is useful to export Deep Web data as Linked Data sets. In this context, processing queries poses some algorithmic challenges due to the inherent limitations of Deep Web sources which, requiring some inputs to be queried, function as services. In this paper we present a framework and a system to export Deep Web data as Linked Data sets. Further, we characterise a class of SPARQL queries that are executable directly, in spite of the limitations.

We give a non-FPT lower bound on the size of structured decision DNNF and OBDD with decomposable AND-nodes representing CNF-formulas of bounded incidence treewidth. Both models are known to be of FPT size for CNFs of bounded primal treewidth. To the best of our knowledge this is the first parameterized separation of primal treewidth and incidence treewidth for knowledge compilation models.

The Deep Web is constituted by dynamically generated pages, usually requested through HTML forms; it is notoriously difficult to query and to search, as its pages are obviously non-indexable. Recently, Deep Web data have been made accessible through RESTful services that return information usually structured in JSON or XML format. We propose techniques to make the Deep Web available in the Linked Data Cloud, and we study algorithms for processing queries posed in a transparent way on the Linked Data, providing answers based on the underlying Deep Web sources. We present a software prototype that exposes RESTful services as Linked Data datasets thus allowing a smoother semantic integration of different structured information sources in a global data and knowledge space.

We define a general framework for ontology-based information retrieval (IR). In our approach, document and query expansion rely on a base taxonomy that is extracted from a lexical database or a Linked Data set (e.g. WordNet, Wiktionary etc.). Each term from a document or query is modelled as a vector of base concepts from the base taxonomy. We define a set of mapping functions which map multiple ontological layers (dimensions) onto the base taxonomy. This way, each concept from the included ontologies can also be represented as a vector of base concepts from the base taxonomy. We propose a general weighting schema which is used for the vector space model. Our framework can therefore take into account various lexical and semantic relations between terms and concepts (e.g. synonymy, hierarchy, meronymy, antonymy, geo-proximity, etc.). This allows us to avoid certain vocabulary problems (e.g. synonymy, polysemy) as well as to reduce the vector size in the IR tasks.

The Deep Web is constituted by data that are accessible through Web pages, but not indexable by search engines as they are returned in dynamic pages. In this paper we propose a conceptual framework for answering keyword queries on Deep Web sources represented as relational tables with so-called access limitations. We formalize the notion of optimal answer and characterize queries for which an answer can be found.

The term Deep Web (sometimes also called Hidden Web) [2, 5, 8] refers to the data content that is accessible through Web pages, typically via HTML forms, but is not available on static pages for indexing by search engines. Deep Web data reside in databases and are made available dynamically, as Web pages, upon a specific search or query. An example is when we query a Yellow Pages website: the generated output is the result of a query posed on an underlying database, and is not stored as static pages; such output can normally be represented in relational form.

The Deep Web consists of those structured data that are available as dynamically generated pages, typically requested through HTML forms. Deep Web pages cannot be indexed by search engines, and are notoriously difficult to query and integrate due to the limited access that they offer. We propose a novel framework for integrating Deep Web sources by means of a mediated schema that represent the underlying, distributed sources. Our goal is to compute answers to queries posed on the mediated schema. To this aim, we propose the use of techniques from the area of Distributed Information Retrieval. We discuss a novel approach to automated sampling, size estimation and selection of Deep Web sources, as well as a technique for merging result lists.

In this tutorial we address the problem of ontology querying, that is, the problem of answering queries against a theory constituted by facts (the data) and inference rules (the ontology). A varied landscape of ontology languages exists in the scientific literature, with several degrees of complexity of query processing. We argue that Datalog(+/-), a family of languages derived from Datalog, is a powerful tool for ontology querying. To illustrate the impact of this comeback of Datalog, we present the basic paradigms behind the main Datalog(+/-) as well as some recent extensions. We also present some efficient query processing techniques for some cases.