Problem: Showing top K ads to the user.

How does the system decide which ads to show:

• User context (location, recent searches, user preference etc..)
• Ads context (Ads popularity, ads budget/fees, ads age etc)
• Ads relevance = Match score between ads context & user context.

Assume you already have a Relevance system that provides ads relevance score (1(lowest match)-10(highest match)). You get following top K+Y (say 100) records from Relevance system and you need to decide which Top K (say 10) to display to the users.

Example: Relevance system output (5 records
Ad1: Relevance score=10/10 and Fee=1$Ad2:Relevancescore=10/10andFee=2$
Ad3: Relevance score= 9/10 and Fee=3$Ad4:Relevancescopre=8/10andFee=100$
Ad2: Relevence score=7/10 and Fee=100$

You job is to find out Top 2 ads to display to the user. What algorithm do you use here ?

Notes:

1. If you give weightage to Relevance score, then you loose your profit as a company.
2. If you give weightage to Fee, then you may loose your customers because you show lower relevant content.
3. Problem is to balance between this ? What potential algo can be used here ?