TU Wien:Knowledge-based Systems VU (Egly, Eiter, Tompits)/Oral Exam 2019-12-10

Exam done with Eiter and Egly; Pr. Egly asked very few questions (≈15min), and then Pr. Eiter asked questions during 40 minutes about Description Logics [in-depth], Truth Maintenance Systems [overview], Probabilistic Reasoning [superficial].

Some questions that I remember:

• What is an interpretation in First Order Logic (formal definition with whiteboard) ?

• Is there another way to interpretate a FOL formula ? (expected: Herbrant basis / universe ...)

• What is a forward-chaining RBS ?

---

• What is the syntax of Description Logics ? What kind of formulas can we express in Description Logics ?

• How do we interpretate a DL formula ?

• Is there another way to interpretate DL ? (expected: transformation to FOL - formally)

• What kind of problems are we trying to solve in DL ? (KB consistency, satifiable concepts, etc)

• Some questions about the complexity of DL (complexity of ALC, complexity of Tableau for ALC compared to the Type elimination algorithm, complexity of different kinds of DL (ALC compared to other fragments), etc.

• Even though the type elimination algo has better theoretical complexity, why is the tableau better in practice ? (drawback of the type elim. algo.: one need to enumerate all types, this requires exponential time and space, just for the base case, while the tableau algo performs better on simple instances)

At this point questions were very in-depth so the professors were not expecting perfect answers but rather the ability to discuss a little bit about concepts.

• What kind of truth maintenance systems did we see in the lecture ?

• How does JTMS works ? (general explanation)

• How does ATMS works ? (superficial explanation)

• Bayes rules; what can you say about bayesian networks ? (expected: very superficial explanation)