Information Retrieval Questions
Normalized Discounted Cumulative Gain at k (NDCG@k) is an evaluation measure used in information retrieval to assess the quality of search engine results or recommendation systems. It takes into account both the relevance and ranking of the retrieved items.
NDCG@k calculates the cumulative gain of the top k items, where the gain of each item is discounted based on its position in the ranking. The relevance of each item is also considered, with higher relevance receiving a higher weight.
The formula for NDCG@k is as follows:
NDCG@k = DCG@k / IDCG@k
where DCG@k (Discounted Cumulative Gain at k) represents the cumulative gain of the top k items, and IDCG@k (Ideal Discounted Cumulative Gain at k) represents the ideal cumulative gain if the items were perfectly ranked.
NDCG@k ranges from 0 to 1, with 1 indicating perfect ranking and relevance, and 0 indicating no relevance or poor ranking. It provides a normalized measure of the quality of the retrieved items, allowing for comparison across different systems or experiments.