Given the numerous knowledge representation models (KR-schemes) that have been proposed, it would be desirable to have a formal, unifying model for the description of a KR-scheme, as well as a general method of comparing KR-schemes in terms of expressive power. This work attempts to fill this gap, by proposing an elegant, yet very general, model of describing KR-schemes. This formalization is used to describe any knowledge representation model, including databases, logic-based schemes, semantic networks etc. It is also applied to introduce a general comparison method for KR-schemes and a formal definition of the reduction of one scheme to another. Using this model, we can reason about KR-schemes in an abstract manner and to determine whether a certain reduction is possible or not.
Motivated by the possibilities of applying deductive database technology for efficient query answering in description logics, we present a translation operator μ that transforms non-recursive ALC ontologies into a disjunctive deductive database. Contrary to our previous work, in this paper we focus on handling negation, disjunction and existential quantifiers, which cannot be handled by deductive databases in a straightforward manner. We present a performance evaluation of our approach, confirming the intuition that techniques for optimizing query answering in disjunctive deductive databases may improve query answering in description logics.
Query answering in the presence of integrity constraints is a fundamental problem in several settings, such as information integration. Keys, foreign keys and inclusion dependencies are the most common forms of constraints used in databases. It has been established recently that, in the presence of such constraints, query answering is decidable for non-recursive queries. Obviously, in the absence of constraints, query answering is also decidable for recursive queries, which are a powerful querying mechanism that subsumes query languages for semistructured data and the semantic web. It was open whether answering recursive queries in the presence of the above classes of constraints is decidable. In this paper we show that this is indeed not the case. In particular, we show that answering recursive queries under keys and foreign keys or under inclusion dependencies is undecidable, both for unrestricted and for finite databases.
The Semantic Web initiative defines important challenges for knowledge representation and database systems. Recently, several standards for representation languages have been proposed (RDF, DAML+OIL, OWL). We briefly discuss the logical basis of these representation languages by referring to description logic (DL) inferences systems. By introducing the DL inference system Racer we demonstrate that practically usable query engines for Semantic Web representation languages are available. Then, we argue that current representation languages for the Semantic Web are not sufficient for simple and well-defined representation problems that naturally arise in the context of Semantic Web applications. In particular, we mention different kinds of algebraic constraints over various domains such as the reals or the natural numbers. We report on practical experiences with DL reasoning systems (e.g., Racer) already supporting these representation means and argue for extensions to DAML+OIL or OWL.
Keyword-based web query languages suffer from a lack of precision when searching for a precise kind of documents. Indeed, some documents cannot be simply characterized by a list of keywords. For example, searching for on-line pictures dealing with formula one using only simple keywords with general-purpose search-engines gives imprecise answers. This imprecision is due to the method that considers that a relevant document to a query is one that contains a lot of query keywords occurrences. This method is totally unefficient for poor textual-content documents like pictures, video streams . On the other hand, keyword based languages are often not powerful enough for expressing sophisticated document search like on-line java or c++ tutorials. We propose ” WeQueL ” a multi-criteria query langage for a better characterization of documents. The aim is to increase the precision of document retrieval on the web. In our experiments, we show the gain in accuracy for web document searching using our language.
Physicians and medical decision-support applications, such as for diagnosis, therapy, monitoring, quality assessment, and clinical research, reason about patients in terms of abstract, clinically meaningful concepts, typically over significant time periods. Clinical databases, however, store only raw, time-stamped data. Thus, there is a need to bridge this gap. We introduce the Temporal Abstraction Language (TAR) which enables specification of abstract relations involving raw data and abstract concepts, and use it for defining typical medical abstraction patterns. For each pattern we further analyze finiteness properties of the answer set.
: The Polylog framework is designed to provide a language for efficiently automating complex queries of information represented in multiple formats. A Polylog program contains a set of modules called specialists that store and make inferences about data in a particular representation. The FocusLoop algorithm answers queries by combining the knowledge and computation of all the specialists. Logic program duals for Polylog programs are introduced to prove that FocusLoop is sound and complete. A logic program dual makes the same inferences as the Polylog program it corresponds to. By using one program to formally characterize behavior and another to implement it, the traditional tradeoffs between provably correct automated question answering, representational flexibility and efficient execution are greatly reduced. Specialists using representations such as neural networks, ontologies, logical clauses and constraint graphs have already been implemented. They demonstrate that complex queries over multiple data sources can be automated without sacrificing efficiency for soundness and completeness. Finally, it is shown that FocusLoop generalizes logical deduction using operations such as resolution, forward inference and subgoaling and that these are common themes in many computational frameworks. In Polylog, each operation is implemented using multiple algorithms, enabling the weaknesses and impasses of one inference or representation technique to be compensated for by the strengths and resources the others.
Flat transactions with savepoints are a variation of the classical flat transactions that allows the user to go undo work done so far back to a certain point within the transaction. This is as opposed to pure classical flat transactions that either commit to whole work done so far or undo it. Recently, this mechanism is being offered by some major database products. Their semantics, however, seem not to be as well studied as the classical flat transactions. In this paper, we show how to use non-Markovian control in the situation calculus to capture flat transactions with savepoints. We also state some of their properties.
View-based query processing is the problem of processing queries over a database based on a set of materialized views, rather than on the raw data in the database. It is a fundamental problem in several settings, such as data warehousing, mobile computing, query optimization, and data integration, and comes about in different forms. The basic task is view-based query answering, which amounts to directly computing the answer to a query by exploiting, besides the query and the view definitions, also the data in the materialized views. A second form of viewbased query processing is query rewriting, where we are given a query and a set of view definitions and the task is to reformulate the query in terms of the views. The rewritten query can then be directly evaluated over the materialized views. Finally, query containment and view-based query containment are the essential tasks for evaluating the quality of the obtained rewritings, by allowing one to compare queries and rewritings. In the talk we discuss and analyze basic techniques for the different forms of view-based query processing in the setting of semistructured data, which are an abstraction for data on the web and XML. In such a setting flexible mechanisms for navigating the data are required, and regular path queries are considered the basic building blocks of more powerful query languages. The techniques we present apply to variants of regular path queries, and are based on automata, exploiting also a strong connection between view-based query answering and constraint satisfaction.
We report about some preliminary issues from the DFG project “Description Logics and Spatial Reasoning” (“DLS”, DFG Grant NE 279/8-1), one of whose goals is to develop a prototypical deductive hybrid Geographic Information System (GIS) with a DL-component. In this paper we discuss the multi-dimensionality of the space of design decisions from a software engineering perspective. In order to support appropriate representation of spatial and thematic aspects and, considering the different aspects of the geographic data, querying the GIS in a uniform way, we are developing a hybrid representation and reasoning framework, offering support for different description languages (not necessarily being description logics). In order to be applicable to a wide range of representation and reasoning tasks, the exploited description languages are not fixed, but exchangeable. The paper sketches our vision of a deductive GIS and we evaluate how standard description logic systems can be of value in this setting. We also introduce a class of spatio-thematic conjunctive queries which is useful in our setting here and argue that query satisfiability and containment are decidable.
This paper introduces schema tuple queries and argues for their suitability in representing knowledge over standard relational databases. Schema tuple queries are queries that return only whole tuples of schema relations. In particular a subclass of the schema tuple queries is identified that is decidable for satisfiability and is closed over syntactic query difference. These properties enable the determination of query containment, equivalence and disjointness. Given this, the identified query class possesses many of the desirable properties of description logics. Additionally such schema tuple queries may be directly translated to SQL and applied over standard n-ary database relations.
In this paper we suggest the novel URI scheme db for identifying not only databases, but also their schema and data components like tables or columns. One of the features of this scheme is that it may not only be used for relational database systems, but for virtually any type of database or data source. We therefore have combined the advantages of both global uniqueness of URIs and the high flexibility of knowledge representation with RDF as part of the Semantic Web. With this novel identifier we are now able to enhance every data record exchanged between databases with metadata: an exact and identifying location of that data in the data source. As a result not only the system administrator is able to backtrack the data to its exact position in the data source but also the database system itself.
This paper considers the problem of containment of conjunctive queries CQ with disjunctive integrity constraints Query containment problem in the pres ence of integrity constraints has been studied broadly especially with functional and inclusion dependencies However to handle incomplete information in the database disjunctions are needed to be expressed as integrity constraints In this paper we introduce disjunctive referential integrity constraints and give a sound and complete algorithm for checking the containment of conjunctive queries under disjunctive referential and implication constraints
We refer to a framework for information integration where a global ontology, expressed in Description Logic, is used as a means for integrating diierent, autonomous information sources. Each source is in turn modeled in terms of a local ontology, and suitable mapping assertions specify the relationship between the local and the global ontologies. Users formulate their queries over the global on-tology, and the system provides the answers to such queries on the basis of the information stored at the sources. To answer queries, the integration system must take into account that the virtual integrated knowledge base should satisfy the integrity constraints that are expressed, either implicitly or explicitly, in the global ontology. This requirement poses interesting challenges to the design of the integration system, especially related to the fact that, in real world settings, sources are usually incomplete and inconsistent, i.e. provide incomplete information to the integrated knowledge base, and may contain contradictory information. In the invited talk, these problems will be discussed in detail, and several techniques for their solution will be presented.
Active databases are usually centered around the notion of Event-Condition-Action (ECA) rules. An ECA rule’s action is executed upon detection of an event whenever the rule’s condition is true. Events are traditionally described using an event algebra. Virtually every proposed active database management system (ADBMS) brings about a different event algebra. This makes it very difficult to analyze these proposals in a uniform way by spelling out what they may have in common, or how they may differ. Typically, logic might act as a framework for dealing with these issues. This paper treats events as (somewhat constrained) formulas of the situation calculus, a (second order) logic for reasoning about actions in general, and database updates and transactions in particular. We provide a framework for devising the semantics of complex events in this logic. Such semantics, formulated as theories of a special kind, are used for reasoning about the occurrence and consumption modes, which are an important dimension of events.
This paper presents an approach to answering queries over an ontology modelled using a description logic. The ontology acts as a global schema, providing a declarative description of the concepts of the domain, the instances of which are stored in (potentially many) object-wrapped sources. Queries are expressed using terms from the rich vocabulary of the ontology, and are translated into an equivalent calculus expression, which references only the objects available in the source databases. The query is then optimised on the basis of information from the ontology and the source databases.
