Quantifiers For Search Engines

Yesterday, after a short talk with Sui, we’ve gained some idea about how existing search engines can increase their usability by an common method we’ve learned in university.

The motivating point is that Sui phoned me and asked for help in searching for the company of a wireless PCMCIA card, whose brand is not well know. The brand of the card is Alpha, which obviously not feasible to search it over since “alpha” is a common word widely used. Then he told me the model of the card is “AFWN411″ letter by letter, number by number. I searched it in Google and surpisingly, no result was returned.

I don’t believe that our beloved internet does not have any information of a network product, even it is not that well known. But Sui was very sure that the model code was correct. Then, I have tried several combinations, such as “A FWN411“, “AF WN411” and so on. Eventually, when I used the keyword “AFW N411“, the expected results returned, and I found that the company of this card is a company in China.

During the dinner with Sui, he praised me as an expert in searching the web. Well, I don’t really think so. Actually, everyone can be good at searching. It is just a matter of how much effort this person wants to put into it. What makes such easy task time consuming? I think it is the variability of the keywords one uses. Actually, search engines, such as Google, are already power enough in deriving “expresses“, “expression“, “expressing” from their origin word “express“, by principles in human languages. But is it enough?

In the study of computer science, we know that there is such a concept which is called “regular expression“, or even simpler which makes use of only “*” and “?” in order to quantify the string as required. Why doesn’t search engines have an option of using such quantifiers? It is not necessary for using all of them. In the example of search product model, if I could have a quantifier “?” and I search “A?F?W?N411″, the time consumed must be less for sure.

Then another issue raises now. How expensive for searching a keyword containing quantifiers? In our example, the question mark “?” is used to only present the possibility of a space. Searching time in this case is then 3 times longer than searching each of keywords without quantifiers. Imagine how many people will use such option during their searching? Not sure. But it is possible that the work load of search engine computers is increased, if more quantifiers are used, instead of just presenting spaces.

If this is true, maybe Google or Yahoo! can have a commercial version for each of their products, to support regular expression or simpler, to limit the computing work of computers. What is the price for each search? I don’t know. 5 cents is acceptable for me.

发表评论

电子邮件地址不会被公开。 必填项已用*标注