NDCG measures: ranking quality with graded relevance scores

Ever wonder why some search results feel more helpful than others?

Imagine you're searching for a recipe online. You want the best results at the top, but sometimes irrelevant results sneak in and make you click through a lot.

Think of each search result like a gift. The closer to the top, the more valuable it is. NDCG helps us figure out how good the search results are by giving more points to the top gifts.

Example

You find a top-rated recipe at the first result, a good one at the second, and a less relevant one at the third. NDCG would reward the top-rated recipe more because it's more helpful.

Remember this

NDCG rewards search engines for giving you the most valuable results at the top, making your search experience better.

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

score matching does: learns the gradient of the log-density without normalizing

Ever wonder how we can compare apples and oranges fairly in studies?

GQA reduces KV-cache memory by the group factor

Ever wondered how websites stay fresh in search results?

MAP (mean average precision) measures: area under the precision-recall curve averaged across queries

Ever wondered how search engines rank your queries?

Boosting (machine learning)

Boosting reduces bias in ML models

Greedy vs beam search decoding: greedy picks best token, beam maintains k candidates

Ever wondered why Google Search sometimes shows you the top results first?

mean pooling often outperforms [CLS] for sentence similarity tasks

Mean pooling captures overall sentence meaning better than [CLS] token embedding

Swipe through 100 ML concepts daily

Open Pocket Polymath