Arguments against the Semantic Web
From GetSemantic
Contents |
[edit] Misconceptions and Strawmen
[edit] Blind Trust
[ RDF ] is all about trusting assertions on webpages
RDF is a language that lets people make assertions in a machine-readable way. It in no way requires you or your machine to read or believe assertions from sources you don't trust - just as your RSS reader doesn't require you to read feeds that you don't like. It is possible to build trust mechanisms on top of RDF, but this is an extra layer, not what RDF is 'all about'.
That said, as human beings, we do trust things written on web pages a lot of the time. Some of us have flown thousands of miles on the basis of exchanged e-mails and blog posts. Making that data available in a machine readable format doesn't mean that, when it is eventually presented to us, we have to trust it. How often does a business lie about it's telephone number?
[edit] Global co-operation
The Semantic Web will never work because it depends on businesses working together, on them cooperating.
The Semantic Web doesn't depend on anybody wanting to co-operate any more than they already do - it just gives them better tools for co-operation where it is in their interests to co-operate. There are business interests in co-operation - Google Maps has become the standard map service for a lot of people over competing services because it provides APIs, JavaScript they can include etc. There is a business interest in a company putting up their address on the website because it helps people buy things from them (or whatever other business objectives they have).
We have seen a number of technologies which have come to the fore on the Internet thanks to business co-operation - URI, HTTP, HTML, XML, Unicode and e-mail. There may be reasons why Semantic Web or 'web of data' technology will be harder to get businesses to adopt than technologies lower down the stack, but this is not a good objection.
For every stick-in-the-mud business who refuses to join the web of data, there'll be a close competitor who will try and exploit the opportunity. We have seen this with the way that independent artists and record labels have jumped in and offered "pod safe" licences for podcasters when big labels ignore or deny what is going on.
[edit] Top-down Ontologies and consistent Worldviews
The general idea is that nobody acts as director; semantics are created ad-hoc and evolve over time... What we're talking about is the "lowercase semantic web", as opposed to the Semantic Web
The semantic web is akin to evolution
The Semantic Web is akin to Intelligent Design
Eric Meyer, Emerging Semantics, 2005, SXSW
[The Semantic Web will fail because people will not] "adopt a common vocabulary"
Not true - the "lower-case" semantics are not "opposed to" the upper-case semantics, except in the mind of some of the proponents of "lower-case" semantics! There is no requirement for a common vocabulary, either. Common vocabularies can happen, and they are useful when they do happen, but OWL has a number of built-in mechanisms for declaring equivalence of both classes and properties - namely owl:equivalentClass (and rdfs:subClassOf for derivative classes) and owl:equivalentProperty. For instances, owl:sameAs can define equivalance, and rdfs:seeAlso allows equivalence relationships for parsing (it is used, for instance, with FOAF).
The 'commonness' of vocabularies will - to some extent - be defined by the commonness in perception. The vast majority of people in the world will be able to see that people exist, or that, say, carbon or cheddar exists. It will become fairly easy to define terms which we can all agree - "foaf:Person" maps reasonably well with most people's conception of what a person is. Whether "foaf:Person" can, say, be married to someone of their own gender may present a few more difficulties - because those difficulties exist in reality. Nobody has ever sold the Semantic Web as a technological solution to deep social problems.
Tim Berners-Lee responds to this myth in these slides: "The semantic web is about a fractal mess of interconnected ontologies..."
See also: Tim Berners-Lee The Fractal nature of the Web.
The Semantic Web, with its neat ontologies and its syllogistic logic, is a nice vision. However, like many visions that project future benefits but ignore present costs, it requires too much coordination and too much energy to effect in the real world, where deductive logic is less effective and shared worldview is harder to create than we often want to admit."
Clay Shirky, The Semantic Web, Syllogism, and Worldview, November 7, 2003
The Semantic Web isn't all about shared worldviews, global top-down ontologies, grand deductive reasoning, or trying to package messy, inconsistent bodies of knowledge into one giant, consistent understanding of the world. RDF (the language of the semantic web) is a syntax for letting anyone say anything about anything. Anyone can write an ontology, and you can pick and choose terms from any ontology you like. If an existing ontology contains a term definition that applies to your data, then great, you can use it; if not, fine, you can create your own term to describe it. Just because RDF enables data to be linked and merged, it doesn't mean that the Semantic Web is about linking and merging all the data in the world for some super computer to assimilate.
The 'Grand Vision' of the Semantic Web isn't everybody sharing the same beliefs and understandings, working together to create "The Semantic Web". It's people using the medium of the web, and the technologies of the semantic web to solve their own problems with sharing, linking, and understanding data.
The small semantic advantages - which projects like Microformats share - can solve immediate problems now, like helping you locate a decent restaurant near your hotel (or some other real world task where meaning is important). The deductive logic comes for free, but if it doesn't happen or isn't used isn't a problem.
The problem which Shirky and others ignore is that if you get rid of namespaces (the 'neat ontologies'), you are still stuck with a default namespace, which does nothing to help us distinguish 'Ford' (the car company) from 'ford' (shallow crossing place in a river). URI's solve that problem, and it would be silly for us to ignore that power.
[edit] (Don't) Repeat Yourself
Some people have suggested that the Semantic Web vision requires that you publish your data twice, breaking the Don't Repeat Yourself design pattern of the Web. If you think of the Semantic Web simply as, say, RDF/XML, then, yes, you would have to publish your data twice. The approaches shown on the GetSemantic site allow you to put data out there without repeating yourself at all:
- RDFa and eRDF use semantic XHTML (and tag soup, in the case of eRDF) to store data. There is no repetition here.
- GRDDL allows you to specify a transformation, and sits a level above the document
- A large amount of the data on the web sits in dynamic contexts - databases, as XML data and within services like Flickr and Upcoming. It is possible for people to provide this data through APIs in formats like XML and JSON. RDF allows that data to become linked with other data and form the Semantic Web, and OWL specifies rules by which data linking can happen.
This claim has some validity depending on how you define 'publish twice'. If providing a structure or schema over the top of the data counts as a second publication, then the Semantic Web vision is guilty as charged - but, this argument seems facile for the reasons listed above.
[edit] It's not happening, so let's just give up
Another common lament is that the Semantic Web just isn't progressing, and that nothing is happening. On the basis of this, we should just give up all hope. The problem with this is that often technology takes far longer than first expected to catch on, but does have a lot of value when it happens.
- Stylesheets for the web go all the way back to the early days of the Web[1], but it took a long time for them to become standard practice.
- Although portable MP3 devices were available in 1997-8[2], it took a long time until the market matured - arguably with the arrival of the Apple iPod in 2001.
The other problem is one of the scope of the claim of "it's not happening". There may not be large numbers of venture-backed companies producing RDF and Semantic Web-based products (although Twine, Freebase and similar services are beginning to change that), nor are there yet "killer apps" based on Semantic Web technology, there are a number of niche uses of the technology, such as in the field of health and biomedical science - hence the existence of the W3C Semantic Health Care and Life Sciences Interest Group.
If you do not arbitrarily limit the Semantic Web to exclude things like tagging and Microformats, then the Semantic Web is growing. With technology like GRDDL, there is a very live possibility of bridging the web of documents and the "upper-case" Semantic Web.
[edit] Other arguments against the SW
- Clay Shirky: The Semantic Web, Syllogism, and Worldview
- Why RDF and the Semantic Web are Ludicrous Ideas
- The 7 (f)laws of the Semantic Web - Dan Zambonini, XML.com
- Rick Jelliffe's 1998 post on xmldev about relationship between HTML, RDF and XML Schema
