Natural Language Processing Named Entity Recognition and Information Extraction 2 — Questions and Answers
Question 1: What is dependency parsing in NLP?
- Analyzing the grammatical structure of a sentence by establishing relationships between 'head' words and words that modify them (Correct answer)
- Splitting a sentence into phrases using context-free grammar
- Identifying coreference chains in a document
- Classifying the topic of a paragraph
Correct answer: Analyzing the grammatical structure of a sentence by establishing relationships between 'head' words and words that modify them
Dependency parsing produces a tree where each word is connected to its syntactic head by a labeled directed arc, revealing grammatical roles like subject and object.
Question 2: What is constituency parsing?
- Analyzing a sentence into nested hierarchical phrases using a context-free grammar (Correct answer)
- Extracting named entities and their types from text
- Computing the semantic similarity between two sentences
- Tagging each word with its part of speech
Correct answer: Analyzing a sentence into nested hierarchical phrases using a context-free grammar
Constituency parsing builds a parse tree that groups words into nested phrases (NP, VP, etc.) according to a phrase-structure grammar.
Question 3: What is semantic role labeling (SRL)?
- Identifying the semantic roles of words or phrases with respect to a predicate, such as agent, patient, or instrument (Correct answer)
- Tagging words with their syntactic parts of speech
- Clustering sentences by topic
- Detecting the sentiment of a sentence
Correct answer: Identifying the semantic roles of words or phrases with respect to a predicate, such as agent, patient, or instrument
SRL answers 'who did what to whom, how, when, and where' by labeling each argument of a verb with its semantic role relative to that predicate.
Question 4: What is open information extraction (OpenIE)?
- Extracting relational triples (subject, relation, object) from text without a predefined ontology (Correct answer)
- Extracting entities only from Wikipedia
- Performing NER restricted to a closed set of entity types
- Parsing text using only rule-based patterns
Correct answer: Extracting relational triples (subject, relation, object) from text without a predefined ontology
OpenIE systems like AllenNLP's extract arbitrary (subject, relation, object) triples from sentences, enabling scalable knowledge acquisition without labeled training data.
Question 5: What is a knowledge graph in the context of NLP?
- A structured representation of entities and their relationships stored as a graph of nodes and edges (Correct answer)
- A lookup table of word embeddings
- A parse tree produced by a constituency parser
- A directed acyclic graph of dependency arcs
Correct answer: A structured representation of entities and their relationships stored as a graph of nodes and edges
Knowledge graphs (e.g., Wikidata, Freebase) store real-world facts as (entity, relation, entity) triples and power question answering, recommendation, and search systems.
Question 6: What is slot filling in task-oriented dialogue systems?
- Extracting values for predefined slots from user utterances, e.g., extracting destination and date from a travel request (Correct answer)
- Generating a response for each turn in a conversation
- Ranking candidate answers for an open-domain question
- Detecting whether a user is asking a question or making a statement
Correct answer: Extracting values for predefined slots from user utterances, e.g., extracting destination and date from a travel request
Slot filling identifies and extracts the required pieces of information (slots) from a user utterance so a dialogue system can fulfill a task like booking a flight.
Natural Language Processing Named Entity Recognition and Information Extraction 2
What is dependency parsing in NLP?