Comment by izendejas | Hacker News Reader

Comment by izendejas | Hacker News Reader

This is huge for many reasons, but namely, this could finally lead to the "semantic web." Metaweb's video, which is linked to in the article, explains part of the "how".

The problem with the semantic web is that many need to embrace it. Many people need to tag text with these "bar codes" (uniquely identified entities). That can take a big effort and there has to be a ROI for this big undertaking. The other is that there is no standard. Well, Google just solved those. With a dominant market share, you don't need someone to agree on a standard, you just force them to--or else they lose out to competition. And as far as the ROI in tagging web pages? Well, what's the ROI on SEO? This will bring about a new form of SEO, except that Google can now undercut many of the search results and answer many of the queries directly--so that'll get interesting... and I'm sure Wolphram Alpha certainly agrees.

Google was also smart to buy Metaweb in order to give web app developers a good reason to use their entities and just FB's open graph entities.

Congrats to the Metaweb team! Freebase + Wikipedia are two of the best gifts to humanity.

The problem with the semantic web is you need a universal ontology. i.e. you need everyone to agree on the same thing. Cory Doctorow's Metacrap explains more http://www.well.com/~doctorow/metacrap.htm

That's a strawman. You in fact don't need a "universal ontology," you just need people to agree on first principles (e.g. URLs are unique, there are things called triples, etc.)

Yes. Absolutely correct. I worked at a semweb company for years, it's all about making a small functional ontology for a particular purpose.

Of course, that leads to the question : is that ontology actually relevant or is it just important that the data is structured?

I think you do need a universal ontology if you want to make the kind of progress the semantic web people talk about. If you just have a bunch of small, separately created ontologies, the situation can indeed seem great until each expands. Then the intersections and ambiguities become huge.

Sure, if you weren't concerned exactness and lack of ambiguity, you could expand the world of triples into a giant, poorly organized collection of information. It would be kind of like the web. The approach "works" but we, uh, already have the web.

Also, the Doctorow document excellent. Anyone expected naive metadata to be extensible should have a reply to it.

If you just have a bunch of small, separately created ontologies, the situation can indeed seem great until each expands. Then the intersections and ambiguities become huge.

Inferencing solves this problem.

That's a reason why the Semantic Web won't wake up and become conscious, like many bad sci-fi stories.

However, if you think of it "vocabularies for people to tag their own stuff in a structured way", which Google can then index and traverse, then it's more realistic.

That said, the Semantic Web people are indeed guilty of hyping this technology as being able to "reason" on its own.

I think even small agreed upon ontologies in different areas would be a big step up from the virtually none that we have. Current services like the experimental google squared currently have nothing to work with. Even a complicated mesh of partially interlinking ontologies would be better.

What about Facebook's new metadata? What are people in the semantic web area saying about that?

Many people need to tag text with these "bar codes" (uniquely identified entities).

This sort of thinking seems to be common in people who like the idea of the semantic Web but who are pessimistic about its implementation. I'm not sure it's going to be the case.

As we've seen happen with other technologies, I suspect we'll see a MetaWeb style approach of "deriving the barcodes" from existing and unformatted content. This will not be a 100% accurate process, but will be "good enough" to make the semantic Web a realistic and large scale underpinning to the next generation of search systems.

Deriving the ontology potentially has more value because it can help avoid spammers. If anyone can just assert that their content is of a specific type without any kind of verification, we end up with the meta tag keywords attribute all over again, where none of that data is trust-able.

Agreed. It won't be an entirely manual process. There are services out there like Zemanta who do a good job of automatically tagging text.

Good thing is that several different unique ids can be interlinked to the same entity. I was using Wikipedia page_ids and IMDB's titles when playing around with it previously. It's a little bit messy, but it works.

Exactly! And freebase has a lot of this data mapped. This data is so rich it holds huge potential.

For example (for those that are unfamiliar with the richness of this data), visit: http://www.freebase.com/view/en/y_combinator

And there you'll note that Y Combinator is mapped to the official site, it's Wikipedia page, it can tell you who the founders are, etc. If you link to Paul Grahm, you can then find out what his personal site is, and so on.

It's even more -- you can do really cool queries with the data. E.g., ask for all people linked less than three degrees of separation from PG and whose ages are between 20 and 40 (as a rather lame example).

I do hope Google will do something cool with this acquisition.

Reply on news.ycombinator.com