Hacker News Reader

Comment by stolenmerch | Hacker News Reader

I was extremely excited by the Semantic Web back in 2003-2005 when working on a master's in Information Science. I was sure it was 5 years away, tops. I was eagerly building profile pages using FOAF and trying hard to implement other ontologies in my work. I got a job in an academic library and pushed hard for semantic technologies. Seemed like Linked Data was such a hot buzzword and topic that we all wanted to explore but could... just.. never.. get going. In 2015 I gave it all 5 more years and then would consider it essentially "vaporware". Maybe it'll catch on someday but it's 2020 and our data is siloed as much as ever. Feel like web scraping plus machine learning is the real winner in this space, but maybe that's an unfair comparison.

View on HN

I had the same feeling back then, although I was in high school. By the time I'd graduated high school it was clear that it was indeed mostly a pipe dream and how, as another comment says, the incentives are all misaligned. I grew fairly sceptical of the semantic web.

And what do you know... 15 years later and I'm working for TimBL himself, figuring out if we can make it work this time with personal data and changed (legislative) incentives. Who knows :)

The incentives aren’t there. It takes effort to get right and what do you get? Bots scraping your data?

I wish it was different that seems to be the thought process.

As I've said elsewhere I fear that it's actually worse then a lack of economic incentives, even in a Communist utopia it won't work because knowledge can not be encoded through semantic predicates.

Umm why not?

I mean, its clear that all knowledge probably can't (probably can't encode all knowledge in english either), but it seems intuitively obvious to me that there exists a useful subset of knowledge that can be, given sufficient effort. I think ability to encode useful knowledge in principle in semantic triples is about the only problem the semantic web doesn't have.

I've ranted about this before[1] and you can read it when searching for my nickname.

It is ultimately a philosophic question about the meaning of semantics (sorry), but suggest to a modern linguist that knowledge can be grounded with predicates logic and they'll tell you "Dude we haven't been trying to do this for more then a 100 years, ever since Wittgenstein had a nervous breakdown".

It is a shame the semantic web doesn't pay attention to other fields that have studied this questions, or maybe some did and left :)

[1]https://news.ycombinator.com/item?id=24583798

"Semantics" is a very overloaded term, but i'm not sure the goal of the semantic web matches the goals of "semantics" in other fields that you are thinking of.

> won't work because knowledge can not be encoded through semantic predicates.

“Can not” means “has the possibility of not happening”. Did you mean: “cannot”?

I have been incredibly interested in ontologies and information science recently. So I’m wondering, what are the hob prospects like? What are you doing now? Your classmates?

Reply on news.ycombinator.com