I think the Semantic Web is a grand idea, but I’m not quite so sure it’ll work, for three very important reasons, all, unfortunately, societal:
First, it assumes that all the people providing information in these XML documents understand the information they are posting in the same way. In the previous writeup, the example was given of a student who posted information that he studied at Purdue, and in the example, an agent program could then take this information and deduce that the student studied, not worked, there. But what if the person who wrote the XML document was the student, and the tag he used implied worked? (After all, he does do a lot of academic work there.) And what if he is only a part time student, or is on the faculty (who sometimes study themselves), etc.? I think people will need help in making their personal Semantic Web documents conform to the assumptions made by the people who write the RDFs and the agents.
Second, I’m afraid that privacy concerns will prevent people from posting important personal information in Semantic Web documents. Or rather, I’m afraid that posting personal information in Semantic Web documents will be a threat to privacy. By posting information about oneself to the web in a machine readable form, it will make it easier for companies to harvest this data in massive quantities. Spambots already do this to ordinary web pages and USENET groups to gain lists of e-mail addresses to send to, which often requires the spoofing of e-mail addresses, thus defeating the original purposes of these e-mail reply fields.
And this brings me to my third point: that the Semantic Web depends on getting people who post these machine readable documents not to lie. Not only can it be perhaps abused by spambots, but also it might become victim to corporate interests publishing reams of purposely false data to the Internet, or maybe even more reams of true data in such quantity, or maybe in locations more accessed by agents, that it is given a false weight, in order to further their own interests. C’mon folks, you just know they’d do it.
I really hope the Semantic Web manages to overcome (or has already found ways around) all these concerns, especially the last two. I really hate it when petty commercial interests stomp all over such noble concepts.
(Written by a person whose concerns, he realizes, may be completely invalid, but decided to post them anyway. Please, please, correct me if I’m wrong.)