The field of biological and biomedical science quickly generate large quantities of data and knowledge; often, domain knowledge is formalised using ontologies expressed in the Web Ontology Language (OWL). Ontology repositories such as Bioportal and Ontobee have been an important infrastructural component for managing ontologies, specifically to search, browse and download ontologies over the Web. We present the AberOWL system, a novel ontology repository that allows access to multiple ontologies through automated reasoning, utilizing parts of the OWL of the ontologies alongside a web interface and web services. AberOWL contains over 300 ontologies and integrates reasoning over ontologies with access to literature and SPARQL endpoints.
Major ontology repositories such as BioPortal
One feature that the existing ontology repositories lack is to utilize automated reasoning over ontologies to enrich the set of services they provide. Automated reasoning over the axioms in the ontologies enables the use of deductive inference when processing an ontology, which can improve the utility of many services provided by an ontology repository:
Verify Consistency and coherence – The releases and updates of ontologies can be automatically classified by the repository, checking them for logical consistency and existence of unsatisfiable classes, and informing the ontology maintainers and the users of the result.
Semantic Query – Searching can make use of inferred knowledge
‘created’ during the reasoning process. Furthermore, the results
can be integrated in retrieval of text documents or data
characterized with ontologies accessible through through public
SPARQL endpoints
Versioning – Ontology versions can be compared semantically based on the inferred knowledge.
Reasoning API – Applications building on the platform have access to classified data, and may build applications requiring ontology semantics and inferred knowledge without having to classify the ontology on the client-side.
However, enabling automated reasoning over multiple ontologies
is a challenging task since automated reasoning can be highly
complex and costly in terms of time and memory consumption
Previous approaches to reasoning over a large set of ontologies
have often involved working with existing collections of ontologies,
usually from one of the large repositories such as Bioportal
The AberOWL
In the demonstration, we will show AberOWL and its functionality, highlighting the differences between our system over traditional ontology repositories that do not utilize automated reasoning. 2
AberOWL consists of an ontology repository, web services which facilitate semantic queries over specific ontologies or the entire set of ontologies contained in the repository, and a user interface. It aims to provide a full framework for interacting with ontologies through an automated reasoner. 2.1
The core of AberOWL is a server which handles the loading, classification of and interaction with ontologies, exposing its Slater et al functionality through a JSON REST API. Specifically, the AberOWL server provides the possibility to query one or all ontologies using the Manchester OWL Syntax. It also supports classifying new ontologies and new versions of existing ontologies, immediately making these available through the API.
Additionally, the AberOWL server provides a searchable index
of all classes available in its set of ontologies, including class
labels, synonyms, description and other associated information. It
utilises an Apache Lucene
Another component of AberOWL scans existing ontology repositories, such as BioPortal, and a set of URLs, for new ontologies and new versions of existing ontologies. When a new ontology or a new version of an ontology is found, a request is sent to the AberOWL server to load the new ontology or to retrieve and classify a new version of an existing ontology. 2.3
The AberOWL repository is a web frontend to AberOWL which interacts with the AberOWL server to provide a user interface for querying ontologies, alongside features found in other ontology repositories such as ontology upload, browsing of ontologies, and downloading ontologies and their versions. New ontologies and new versions of ontologies can be uploaded, after which they are immediately classified by the AberOWL server and made available for querying.
The prime goal of the AberOWL repository is to have a userinterface in which all structural information about ontologies is derived in real time from the AberOWL reasoning server. Currently, the interface can be used to upload and classify new ontologies, browse and query ontologies, visualize ontology structure and retrieve information on ontology classes. 3
In the future, we plan to improve AberOWL’s potential to retrieve a single class based on search of the class’ metadata and axioms. Furthermore, we intend to provide the possibility for the visual comparison of ontology versions, utilizing automated reasoning to show differences in the inferences that can be drawn from both ontologies.
A second area of research is to increase AberOWL’s integration with biomedical data. In particular, we aim to more closely integrate AberOWL with publicly available SPARQL endpoints to enable access to these SPARQL endpoints through automated reasoning over ontologies.