Natural Language Processing Named Entity Recognition and Information Extraction 1 — Questions and Answers
Question 1: What is Named Entity Recognition (NER) in NLP?
- The task of locating and classifying named entities in text into predefined categories such as person, organization, or location (Correct answer)
- The task of predicting the next word in a sentence
- The process of removing proper nouns from text
- A method for measuring sentence similarity
Correct answer: The task of locating and classifying named entities in text into predefined categories such as person, organization, or location
NER identifies spans of text that refer to real-world entities and categorizes them, e.g., tagging 'Apple' as an organization and 'Cupertino' as a location.
Question 2: Which tagging scheme is commonly used for NER to mark entity boundaries?
- BIO (Beginning, Inside, Outside) tagging (Correct answer)
- POS (Part-of-Speech) tagging
- Dependency labeling
- Chunking with IOB2 only
Correct answer: BIO (Beginning, Inside, Outside) tagging
BIO tagging marks the first token of an entity with B-, continuation tokens with I-, and non-entity tokens with O, clearly delineating entity boundaries.
Question 3: What is relation extraction in NLP?
- Identifying semantic relationships between entities mentioned in text, such as 'works at' or 'located in' (Correct answer)
- Extracting noun phrases from a parse tree
- Finding synonyms for words in a sentence
- Splitting compound sentences into simple ones
Correct answer: Identifying semantic relationships between entities mentioned in text, such as 'works at' or 'located in'
Relation extraction goes beyond NER to determine how entities are related, populating knowledge bases with structured (subject, relation, object) triples.
Question 4: What is coreference resolution?
- Determining when two or more expressions in a text refer to the same real-world entity (Correct answer)
- Resolving spelling errors in a document
- Linking a word to its definition in a dictionary
- Parsing the dependency structure of a sentence
Correct answer: Determining when two or more expressions in a text refer to the same real-world entity
Coreference resolution clusters mentions (e.g., 'Barack Obama', 'he', 'the president') that refer to the same entity, enabling coherent text understanding.
Question 5: What is event extraction in information extraction?
- Identifying mentions of specific events in text and extracting their participants, time, and location (Correct answer)
- Counting the frequency of verbs in a document
- Detecting grammatical errors in sentences
- Summarizing a document into a single sentence
Correct answer: Identifying mentions of specific events in text and extracting their participants, time, and location
Event extraction detects trigger words and fills argument slots (who, did what, to whom, when, where) to build a structured representation of events from unstructured text.
Question 6: Which evaluation metric is most commonly used for NER tasks?
- F1 score computed at the entity span level (Correct answer)
- BLEU score
- Perplexity
- Mean Reciprocal Rank (MRR)
Correct answer: F1 score computed at the entity span level
NER is evaluated using span-level F1, which requires both the entity type and exact boundary to match for a prediction to count as correct.
Natural Language Processing Named Entity Recognition and Information Extraction 1
What is Named Entity Recognition (NER) in NLP?