CIA Search Systems and Findability 3 — Questions and Answers
Question 1: An information architect recommends implementing 'best bets' (also called 'promoted results'). What is the primary rationale?
- To monetize high-traffic queries through advertising placements
- To surface curated, authoritative results for high-frequency or high-stakes queries regardless of algorithmic ranking (Correct answer)
- To replace the relevance ranking algorithm entirely
- To display results from external websites before internal content
Correct answer: To surface curated, authoritative results for high-frequency or high-stakes queries regardless of algorithmic ranking
Best bets allow editors to pin authoritative answers to predictable, important queries, compensating for algorithmic limitations on known high-value searches.
Question 2: What does 'recall' measure in search evaluation?
- How quickly users remember a previous search session
- The fraction of all relevant documents in the collection that were actually retrieved (Correct answer)
- The number of search queries processed per second
- The accuracy of spelling correction suggestions
Correct answer: The fraction of all relevant documents in the collection that were actually retrieved
Recall is the ratio of relevant documents retrieved to the total number of relevant documents in the collection, measuring completeness of retrieval.
Question 3: In enterprise search, which of the following is the most common cause of low findability despite a well-configured index?
- Overuse of Boolean operators by end users
- Poor metadata quality and inconsistent content tagging at the source (Correct answer)
- Excessive use of faceted navigation
- Search result caching that is too aggressive
Correct answer: Poor metadata quality and inconsistent content tagging at the source
Even a sophisticated search engine cannot surface content accurately if source metadata is missing, inconsistent, or inaccurate.
Question 4: A 'zero results' page in a search system should ideally:
- Display a generic error message and a link to the homepage
- Offer query suggestions, related terms, or browse paths to help users recover (Correct answer)
- Redirect users to a contact form
- Show a blank page to minimize user confusion
Correct answer: Offer query suggestions, related terms, or browse paths to help users recover
Zero results pages are critical failure points; best practice is to offer alternative pathways such as spelling suggestions, related queries, or top-level browse links.
Question 5: Which controlled vocabulary structure is specifically designed to express hierarchical, associative, and equivalence relationships among terms for use in search systems?
- Folksonomy
- Taxonomy
- Thesaurus (Correct answer)
- Ontology
Correct answer: Thesaurus
A thesaurus in information science formally captures broader/narrower term (BT/NT), related term (RT), and use/use-for (UF) relationships to support search.
Question 6: Stemming in a search index is the process of:
- Filtering out offensive content before indexing
- Reducing words to their root form so variants map to the same index entry (Correct answer)
- Highlighting query terms within the result snippet
- Grouping search results by document type
Correct answer: Reducing words to their root form so variants map to the same index entry
Stemming strips suffixes (e.g., 'running' → 'run') so that different morphological forms of a word match the same indexed token.
Question 7: In a federated search system, content is typically retrieved by:
- Consolidating all sources into a single unified index before query time
- Sending the query to multiple disparate indexes simultaneously and merging results (Correct answer)
- Limiting retrieval to a single authoritative repository
- Using machine learning to predict the best single source per query
Correct answer: Sending the query to multiple disparate indexes simultaneously and merging results
Federated search queries multiple independent indexes in parallel at runtime and merges their results, avoiding the need to maintain a single unified index.
An information architect recommends implementing 'best bets' (also called 'promoted results').
What is the primary rationale?