NLP - Text Similarity and Search
You want to combine cosine similarity with another metric to improve document clustering. Which combination is most reasonable?
You want to combine cosine similarity with another metric to improve document clustering. Which combination is most reasonable?
hard📝 Application Q9 of 15
Step-by-Step Solution
Solution:
Step 1: Understand complementary metrics
Jaccard similarity measures overlap of token sets, complementing cosine similarity's vector angle.Step 2: Evaluate options
Random guessing and sorting lengths do not improve clustering; Euclidean distance on raw counts is less meaningful without normalization.Final Answer:
Cosine similarity with Jaccard similarity on token sets -> Option AQuick Check:
Combine cosine with Jaccard for better clustering [OK]
Quick Trick: Combine cosine with set-based similarity like Jaccard [OK]
Common Mistakes:
MISTAKES- Using random or irrelevant metrics
- Ignoring normalization for Euclidean distance
- Using document length sorting as similarity
Master "Text Similarity and Search" in NLP
Want More Practice?
15+ quiz questions · All difficulty levels · Free
Free Signup - Practice All QuestionsMore NLP Quizzes