21.11.2013 Views

YEARS OF EUROPEAN ONLINE ANNÉES DE EN LIGNE ...

YEARS OF EUROPEAN ONLINE ANNÉES DE EN LIGNE ...

YEARS OF EUROPEAN ONLINE ANNÉES DE EN LIGNE ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

WORKSHOP<br />

Thesauri<br />

thesauri are controlled vocabularies with a hierarchical structure. In this<br />

respect, they reuse the same approach as the classiication methodologies, but<br />

on the level of the components of natural speech. the main items which are<br />

registered in a thesaurus are generally called descriptors, as they are used to<br />

describe the contents of a textual object. the entries in the thesaurus are reined<br />

by a sophisticated system of links to broader terms, narrower terms,<br />

synonyms and related terms.<br />

the advantage of the use of thesauri is that research can automatically be:<br />

• redirected from registered synonyms to the corresponding descriptors;<br />

• limited to narrower terms if the number of results exceeds a certain threshold;<br />

• extended to broader and/or related terms if the results are too poor.<br />

Although a big step forwards, thesauri also have their shortcomings.<br />

the quality of the research results depends on the quality of the use of the<br />

descriptors. furthermore, only the main concepts of a text will be taken<br />

into consideration. So there is always the risk that texts are only retrieved<br />

in the context they are prepared for, while other present concepts are not<br />

found.<br />

Semantic networks<br />

the idea of thesauri is driven forward by the technologies in the context<br />

of a semantic network. they are mainly based on the use of ontologies, which<br />

are comparable to thesauri, but instead of linking terms, they describe the relations<br />

of concepts. the example overleaf illustrates this approach in comparison<br />

with thesauri.<br />

the example is simpliied, but it is obvious that the description of the nature<br />

of the relation is much more useful than a pure link.<br />

the evaluation of these descriptions allows for some logical conclusions<br />

which can automatically be made on such basis. In semantic networks, heritage<br />

is an important aspect. So, on a deeper level, the properties of the hyperonymic<br />

level are inherited and are also available for association with other<br />

concepts.<br />

As the work for the creation and maintenance of thesauri is already very<br />

time-consuming and complex, the elaboration of ontologies is even more<br />

complicated. this is why this technology is not yet used for big projects or<br />

large amounts of data, but for small excerpts. the spanning of the Internet by<br />

162 | 163<br />

01_2007_5222_txt_ML.indd 163 6-12-2007 15:14:06

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!