My guess is that you need to know how’s the repercussion of a given subject among the news of the day, considering a bunch of trusted news sources. So if a user searches for ‘iraq war’, the system will check how often such terms appeared among the news of the day (number of sources that published news with the query terms Vs. total number of sources) and their relevance (considering position, idf, frequency, if it’s a perfect match or not, etc, etc.).

Ex.: if 10 among 100 sources published news where ‘iraq war’ appeared on title, it’s obviously a relevant subject of the day, then you need to determine which scores deserve a top or a fourth place. To determine that you probably need to run experiments with real people, making them vote for first or fourth among several news related queries to determine the threshold score.

Maybe that’s why ‘great war’ or even ‘the great war’ also shows a DD on screen, and I don’t think it should, since The Great War is far from being a news topic of the day.