Entity Linking and Disambiguation Quiz

Answers and explanations appear as you take the quiz.

1. What does entity linking connect a text mention to?
   • A random web page
   • A new training run
   • A spelling correction
   • A unique entry in a knowledge base
2. In the phrase 'the company Apple released a phone', what is the disambiguation task?
   • Deciding Apple means the fruit, not the company
   • Deciding Apple means the tech company, not the fruit
   • Translating Apple into another language
   • Counting the letters in Apple
3. What is the purpose of the candidate generation stage?
   • To write a new article
   • To delete ambiguous words
   • To retrieve a shortlist of possible matching entities
   • To translate the text
4. What role does a bi-encoder play in modern entity linking?
   • Fast retrieval via separate embeddings for mention and entities
   • Slow manual labeling of every mention
   • Generating images of entities
   • Encrypting the knowledge base
5. What does a NIL class represent in entity linking?
   • The most popular entity
   • A mention with no matching knowledge-base entry
   • A duplicate entity
   • A training error
6. How can document-level coherence help an entity-linking system?
   • By guaranteeing that every mention has the same entity identifier
   • By proving every document discusses only one topic
   • By removing the need to evaluate individual mention contexts
   • By considering whether candidate assignments fit the document context and one another
7. After fast candidate retrieval, what information does a cross-encoder jointly read for entity-linking reranking?
   • The mention's context and a candidate entity's description
   • Only the entity's numeric identifier with no text
   • Only the two input filenames
   • Only the number of images in the knowledge base
8. If an entity-linking pipeline never proposes the span 'Apple' for analysis, which initial stage has missed that mention?
   • The final candidate reranker only
   • The image-compression stage
   • Mention detection
   • The knowledge base's output filename formatter