A user submits a query, q, to an information retrieval system which returns a ranked list to the user. Given the top k ranked documents.

i) Explain how, in terms of IR evaluation, one could measure the quality of this answer set. Discuss any limitations of this approach.

ii) Explain how the query q could be modified using evidence from the returned set to improve performance. Outlines any potential limitations of this approach.

iii) Given the original query q and the modified query q, discuss, in your words, how you might predict which query is likely to perform better.