Users of Bing will probably not consciously be aware, but the Microsoft search engine tends to place fewer ads alongside their search results than its rivals. One of the reasons for this may be the fact that most advertisers still prefer to book their ads on Google. But Microsoft also has another explanation: The IT group uses advanced techniques to display most relevant ads.
The system, called "AdPredictor" will probably be tested soon to be much wider than before - the portal giant Yahoo is in the process to change its search engine like Bing. As part of this market campaign Microsoft researchers aired for first time some of this algorithms secrets, which make Bing users and advertising industry will happier than the competition.
Even if users find they often annoying - the ads that appear alongside search results are enormously important for the online economy. Last year in U.S more than 25 billion dollars in sales it generated. Most of the money flowed to Google - according to the latest ComScore statistics, almost 63 percent of queries answered from American households. Microsoft hopes to be able to cut through its partnership with Yahoo a bigger share of advertising pie.
It is particularly important to deliver the most relevant advertising algorithms. For only then more clicks on more ads per page will achieved.According to some experts that it's not necessarily mean more money with excessive advertising on per page?
To predict how many clicks should get an ad, Microsoft's algorithms analyze trends from the past few months to determine what users were doing earlier, when ads appeared to match a given search term. So far this has been determined by relatively simple rules in the search software: for example, when entering the search term "car" advertising for companies that offered low-cost oil change appeared.
Bing is now using software in the field of machine learning. AdPredictor was developed by Microsoft researchers Joachin Quinonero Candela, who conducts research in Cambridge. The technique is based on Bayesian models, an AI technique that helps the algorithms constantly fresh data to calculate the probability of an event (in this case to integrate the advertising click). In the area AdPredictor analyzed how many times an ad is clicked such as under varying conditions, Quinonero Candela said. Ads that fit cars will change then for example; depending on the time, the place from which a person seeks out or the concrete placement of an advertisement.
The result in any case that Bing could afford to get fewer ads, says Quinonero Candela.The likelihood that what was placed on a page where advertisement is clicked. If we try to increase the accuracy of our predictions, the fact that we are trying to annoy the users less and at the same time satisfy the advertisers.
Since the introduction of AdPredictor, according to market researchers at Bing click-through rate is so important when compared with the previous offer from Microsoft search engine (Live Search). According to a recent analysis of the advertising specialists Adgooroo shows that Bing generated now 3.85 ads per keyword, compared with 5.72 in Google - although it is not clear whether this is mainly due to the preference of advertisers for Google. When Bing comes to Yahoo the relatively high rate (currently 6.85 per search term) most likely goes down.
An important element of AdPredictor; the system learns with new data during online operation. According to experts from Emory University algorithms from the field of machine learning would initially need all the data before they could build their model. Bing is now pursuing this approach therefore is quite new on web.
Article Source: http://EzineArticles.com/?expert=Karl_Taylor
