About the TAC AA competition
The Trading Agent Competition is an annual competition with the goal of encouraging research into the design of autonomous trading agents.
In the Ad Auctions scenario, agents compete as advertisers
to see who can make the most profit from selling a limited range of home entertainment products through sponsored search advertising,
that is, paying a search engine to display query-specific ads beside query results.
In each game, eight agents compete for 60 simulated game days, each lasting 10 seconds.
Products are classified by manufacturer (flat, pg, and lioneer) and by component (tv, dvd, and audio)
for a total of nine products.
Search engine users, the potential customers, submit queries consisting of
a manufacturer and a component, although either or both may be null, i.e. missing.
There are thus 16 total query types.
Each day, for each of the 16 query types, a keyword auction is run.
For each auction, an advertiser submits i) a (real, non-negative) bid indicating the amount it is willing to pay per click, ii) an ad, and iii) a spending limit
(optional).
Each time a user submits a query, the top five bidders (who have not yet hit their spending limit) have their ads shown in order.
The user then chooses which ad(s), if any, to click on, and possibly makes a purchase.
Ads in higher positions are more likely to be clicked.
Each advertiser specializes in a certain manufacturer and component, and is limited in how many products it can sell in a given period.
Advertisers receive limited information about the actions and outcomes of other advertisers (and thus the cost and benefit from a given ad position), resulting in a challenging bidding problem.
Additional background on the TAC AA scenario is available
here.
The full specification is available as a PDF
here.
Illustration of a query, taken from the specification document
 |