Fig. 4
From: Clinical language search algorithm from free-text: facilitating appropriate imaging
Search ranking relevance on generated indications dataset. This figure is a bar graph of the average NDCG on simple and complex queries in all 12 categories. Error bars indicate 1 standard deviation. The number of documents (and therefore queries) for each category is in parentheses. There is a significant difference between the NDCG values among categories (p = 0.00054, Kruskal–Wallis H-test). Note: ‘Major Trauma’ category was excluded from statistical analysis due to sample size of 1. NDCG (normalized discounted cumulative gain)