09.12.2012 Aufrufe

Leibniztag - edoc-Server der BBAW - Berlin-Brandenburgische ...

Leibniztag - edoc-Server der BBAW - Berlin-Brandenburgische ...

Leibniztag - edoc-Server der BBAW - Berlin-Brandenburgische ...

MEHR ANZEIGEN
WENIGER ANZEIGEN

Sie wollen auch ein ePaper? Erhöhen Sie die Reichweite Ihrer Titel.

YUMPU macht aus Druck-PDFs automatisch weboptimierte ePaper, die Google liebt.

� e goal of KYOTO is to develop an information system that provides deep semantic<br />

search and access to large quantities of distributed multimedia data for both experts<br />

and the general public, covering a broad range of data from wide-spread sources<br />

in a number of culturally diverse languages. Speci� cally, KYOTO focuses on the environmental<br />

domain and involves users from two environmental groups, in addition<br />

to several international research groups. � e system is developed for English, Dutch,<br />

Italian, Spanish, Basque, Chinese and Japanese and relies on an ontology linked to<br />

wordnets – lexical semantic databases – in a variety of languages. Concept extraction<br />

and data mining are applied through a chain of semantic processors that re-use the<br />

knowledge for diff erent languages and for particular domains. � e shared ontology<br />

guarantees a uniform interpretation for diverse types of information from diff erent<br />

sources and languages. � e system can be maintained by � eld specialists using a Wikiplatform<br />

and used by experts and laymen alike. Ultimately, KYOTO is a generic system<br />

that will off er knowledge acquisition and transition for any domain and a wide<br />

range of user groups across linguistic, cultural and geographic bor<strong>der</strong>s.<br />

Research at the <strong>BBAW</strong>: Overview<br />

� e <strong>BBAW</strong> is one of several core project partners. � e <strong>Berlin</strong> group is led by Christiane<br />

Fellbaum and includes Axel Herold, Amanda Hicks and � omas Pfuhl. � e <strong>Berlin</strong><br />

team was assigned the main responsibility for Work Package Six, which can be described<br />

as follows.<br />

Domain-speci� c documents are appropriately marked up, tagged and parsed<br />

such that information about key concepts and their interrelations as well as stated<br />

and implied facts can be extracted. (� e mark-up follows the latest ISO standards.)<br />

Domain specialists without linguistic expertise can encode the information reliably<br />

and accurately in a pre-existing semantic network (wordnet) in their language. From<br />

these extensions to the individual wordnets appropriate mappings are made to a language-independent,<br />

formal ontology. (An ontology is a formal, logically structured<br />

representation of concepts.) � e ontology expresses not only the domain-speci� c<br />

concepts encoded by experts, but includes top-level as well as mid-level layers. � e<br />

shared ontology will be based on existing formal ontologies, in particular DOL-<br />

CE, and extended with semantic information that is currently stored in the English<br />

WordNet lexical database and various knowledge resources linked to it. Concepts in<br />

the enriched English WordNet are candidates for inclusion in the language-independent<br />

ontology.<br />

In or<strong>der</strong> to maximally exploit the created resources, new capabilities for reasoning<br />

and logical inference will be added to the knowledge base by means of advanced<br />

� eorem Provers. � ese capabilities are essential for verifying meta-properties like<br />

consistency, and for the deduction of implicit properties from the explicit informati-<br />

326 | Berichte <strong>der</strong> Projekte und Initiativen

Hurra! Ihre Datei wurde hochgeladen und ist bereit für die Veröffentlichung.

Erfolgreich gespeichert!

Leider ist etwas schief gelaufen!