10 07 12 1

Page 1

Schoolof Information Engineering Knowledge Engineering 2011–12 Test 3 - 10th July 2012 Part II Family name Given names ID 3 4 points Describe all the relevant differences between DTD and XSD schema languages. If a language L we are using to describe incoming messages is not representable exactly in our schema language, which is the best policy to approximate it? Why ? 1. see slides about DTD and XSD for comparison 2. if L is not representable exactly, we can approximate it by RELAXING some constraints. This can be done, for example, by: • removing cardinalities in iterations • removing orders in sequences • allowing some extra or different elements in contents • etc.. This approach is necessary to avoid to exclude any correct document from L. This means that the language we will describe exactly, lets call it L', will be a superlanguage of L, that is it will contain all the documents in L, but will validate some extra “incorrect” documents due to approximations. Relaxed constraints will be recovered by adding some custom validation in application layer. This requires an extrea effort by the application developer, but he will consider the document valid againsta L' schema, so its job can be simplified. If, at the opposite, L' removes any document from L, we will have no way to recover it at the application level. continue on page 4 4 7 points For each axiom in the following list say which expressivity (RDFS, OWL(1)Lite, OWL(1)Full or none) is required to express it (or an equivalent set of axioms) both if (1) it is taken alone and (2) if it is considered together with the other axioms. Write a GrOWL representation for the second case 1. A B⊔ ⊑ CD⊓ 2. R∃ ⊑ B 3. B ⊑ R.C∀ 4. D ⊑ ≤1R 5. A ≡ ¬(=5R-.B) 6. A ⊑ ¬(=5S.C) (given that S R)⊑ Say if (and why) the query SELECT ?x…