Established evaluation methodologies for information retrieval (IR) are not well suited to the task of comparing systems in many real settings. We have implemented a comparison tool which support this task. By offering a working search interface the tool permits study of real information needs as they occur, uses the documents actually available at the time of the search, and records judgements taking into account the instantaneous needs of the searcher. We report experiments, to validate the tool and explore potential biases, and ongoing and future applications.What occurred to me was that this technique could be applied to web advertisements. As Paul commented, the leading commercial public search engines give comparable results. If you do a web search with Google or Yahoo, you usually get some useful, plausible answers. Improving on these already good results is a very hard problem. But the technology of placing ads on web pages is not so well developed and has a lot of room for improvement.
From: Evaluating Information Retrieval in Context, Paul Thomas, Department of Computer Science, Australian National University and CSIRO, 2006-10-04
Google AdWords allows you to specify the keywords for an advertisement you want matched with web pages the ad will appear on. When I look at the results of this process with ads placed on my own web pages with Google AdSense, I frequently find the results jarring. Out of four ads Google puts on the web page, usually two or three will be applicable. But often one or two of the ads will be on something not related to the web page content. This is distracting for the reader and not useful for the advertiser.
Perhaps Paul's techniques could be used to help advertisers better match ads to web pages. Better matched ads might make advertising more acceptable, helping pay for better web content. Also ads more attuned with the reader's interests might be more profitable for the advertiser. This might be the new frontier in search engines and provide the next web startup company.
No comments:
Post a Comment