National and Kapodistrian University of Athens
Database techniques for ontology-based data access
Abstract
dc:descriptionOntology-based Data Acess (OBDA) is a method for linking an ontology, which encodes knowledge about the classes and properties of entities for a given application domain,to underlying data sources. These data sources, managed by specialized systems, can be in various forms and usually reside in pre-existing repositories. The linking is accomplished through declarative mappings, which are used to generate ontology terms from information in the data sources. Instead of materializing all the ontology terms, the user of the relevant application can pose a query over the ontology, and then a process of query transformation is carried out, which has as a result a query in the native language of the underlying data sources. This resulted query is then executed, and the results are presented to the user, transformed as ontology terms. This approach, also known as virtual knowledge graph approach, has the advantage that provides the user with a familiar vocabulary over which he can pose a query, concealing details about the underlying data sources, such as complex schemas and storage particularities. On the other hand, the process of transforming the initial query over the ontology into a query over the under lyingsources, leads in many cases to complex and large queries. In this thesis, we study the problem of efficient query answering for OBDA systems from a database perspective, concentrating on the ontology language OWL 2 QL, which is a dialect of the OWL family specifically tailored for the case where massive data are stored in an external data source. We are also concentrating on the case where the initial query posed over the ontology is in the form of a union of conjunctive queries. As expected,this issue heavily depends on the exact kind of the underlying data source. For this reason we make a distinction between three different commonly encountered scenarios. In the first scenario we consider that the underlying system is a single relational database management system. In the second scenario we consider that we have a federation of different relational systems. Finally, in the third scenario we consider the case where data are stored in a specialized triple store in the form of RDF statements.For the first scenario, we identify redundant processing as a key problem in OBDA query execution over a relational system. Examples of such processing are duplicate answers obtained during query evaluation, which must finally be discarded, or common expressions evaluated multiple times from different parts of the same complex query. Many optimizations that aim to minimize this problem have been proposed and implemented,mostly based on semantic query optimization techniques, by exploiting ontological axioms and constraints defined in the database schema. However, operations that introduce re-dundant processing are still generated in many practical settings, and this is a factor that impacts query execution. To handle this issue, we propose a cost-based method for query translation, which starts from an initial default translation and uses information about redundant processing in order to come up with an equivalent, more efficient translation. The method operates in a number of steps, by relying on certain heuristics indicating that we obtain a more efficient query in each step. Through experimental evaluation using the Ontop system for ontology-based data access, we exhibit the benefits of our method. For the second scenario we have developed a system that acts as a mediator betweenthe OBDA system and the federated databases. This system, built using the Exareme engine, decomoses the produced query into different fragments, sends these fragments for evaluation in the external databases and imports the intermediate results that correspond to these fragments. Finally, these intermediate results are combined in order to produce the final query result. During this process, we have adapted techniques and methods from database literature for usage in the context of OBDA. These methods cover areas such as data integration, common subexpression identification, caching of intermediate query results and distributed processing. The developed mediator system has been integrated into the platform of the Optique research project and has been successfully deployed in a demanding real world use case, federating seven different databases which contain geological data. For the last scenario, we have developed PARJ, a specialized in-memory RDF store which takes into consideration ontological hierarchies during join processing with very low performance overhead, using on-the-fly computation of the inferences regarding class and property hierarchies. In the spirit of the OBDA approach, PARJ avoids expensive preprocessing and materialization of implications. PARJ is also amenable to straight forward parallelization. Specifically, we present a join implementation that allows to achieve any desired degree of parallelism on arbitrary join queries and RDF graphs stored in memory using compact vertical partitioning. We use an adaptive join processing approach,such that we take advantage of complete or even partial ordering of RDF data, which is compactly stored in order to increase spatial locality and keep memory consumption low,coupled with an ID-to-Position vector index used when ordering does not allow for efficient scanning of the input relation. Finally, we experimentally show the efficiency and scalability of our proposal.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2020
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Bilidas, Dimitris
- Μπηλίδας, Δημήτριος
Subjects
dc:subject × 15- Οντολογίες
- Ανάκτηση δεδομένων βάσει οντολογιών
- Βελτιστοποίηση επερωτήσεων βάσει κόστους
- Σημασιολογικός ιστός
- Γράφοι γνώσης
- Ενοποίηση δεδομένων
- OBDA
- Query optimization
- Semantic web
- Knowledge graphs
- Data integration
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/48680