Pregatire NLP

Finite state automata may be used for:

• Scripts for conversational bots

• NP-phrase identification

• None of the other options

• Semantic analysis

• Coreference resolution

Long distance dependencies in texts may be handled with:

• The bag of words approach

• Transformers

• Ontologies

• Formal grammars

• Recurrent Neural Networks

When implementing a conversational agent (for example, similar to "Siri") you might need:

• A corpus of dialogs

• Knowledge graphs

• An ontology

• Frames

• Machine learning

We can say about lexical chains that:

• We can use them for semantic disambiguation

• If a text has several lexical chains, it surely is coherent

• We can use them for the segmentation of texts

• If a text has several lexical chains it may be coherent

• We can use them for discourse analysis

We may say about the sentence "Jim comforted Kevin because he was so upset" that:

• Coreference resolution is ambiguous

• It is not cohesive

• None of the other options

• It is not coherent

• It is an example of a Winograd schema

What advantage has LDA as compared to LSA?

• It differentiates senses of words

• None of the other options

• LSA does not create semantic spaces

• LDA uses deep neural networks

• It does not have an advantage as compared to LSA

What are illocutionary acts?

• The consequence of a perlocutionary act

• A way to induce an action not explicitly asked

• A consequence of a locutionary act

• A kind of speech acts

• None of the other options

Multi-parties chat conversations, as compared to those with two participants, have these features:

• Zipf's law may be applied

• Turn taking strategies are changing

• There are no differences

• Multiple threads of discussion may appear in parallel

• Speech acts cannot be identified anymore

Coreference resolution may need:

• Inferences

• A history list

• Named Entity Recognition

• Clustering

• Part of Speech Tagging

We can say about the semantic distance of Rada that:

• It is the simplest to compute semantic distance

• It is the semantic distance with the best results

• It cannot be used for WordNet

• It is more suited to a semantic network than a taxonomy

• It does not depend on the position where is measured in an ontology