Semantic Overflow

Question

2

"Adding semantics to the Web" is the title I have been given for my dissertation which I am finding it hard to formulate a project to develop on top of this. I have been doing a fair amount of reading both casually on the web and academic papers so I am not a total notive in this field.

Whenever I seem to go to see my dissertation advisor he doesn't really help and just makes suggestions that I make some kind of semantic search engine/crawler for a domain of my interest. However doing that seems like It's just reinventing what's already been done for sites like last.fm or imdb. The only other idea that I could think of would be crawling programming questions and storing these in a semantic way of sorts...

Does anyone have any ideas or suggestions for an area regarding adding semantics to the web?

edited Nov 5 at 22:03

asked Nov 5 at 21:25

Malachi
57●5

Andrew Matthews · Accepted Answer · 2009-11-05 22:25:00Z UTC

Perhaps you could explore those ideas from a specific angle - disaster recovery for example? I understand that when disaster strikes, one of the most pressing issues faced by recovery crews is finding local providers of goods and services. Perhaps if you can find ways to extract that kind of information from local (non-semantically annotated) sites, you would be 'adding semantics to the web' .

Aside from being very worthwhile in its own right, this would pose you non-trivial problems relating to NLP as well as semantic web development. And doing this stuff on the web scale (i.e. broad enough to be useful) would be worthy of kudos as well!

Egon Willighagen · Answer 2 · 2009-11-06 06:37:35Z UTC

Dear Malachi,

I tend to agree with your supervisor that the interesting challenges are now in the use of the semantic web, instead of building it. There are many initiatives working on making data available as RDF, e.g. Bio2RDF and the HCLS-IG in the life sciences domain (the area where I tend to contribute too), but the real 'problems' are in using this data in ways unique to the semantic web, and more efficient than old-world approaches. This is particularly visible in cheminformatics, where graph theory has been used for more than 50 years now, and doing queries purely in the RDF-world is not always optimal.

Except challenges in making efficient reasoning, which of course is greatly affected by the ontological modeling of the domain, there is also the linking of the two worlds, using facts expressed in RDF, and calculating triples on the fly; With Bioclipse-RDF I am trying to make this happen for life sciences (and focus of domain visualization at the same time)...

Please have a look at,for example, SADI, which hooks in computation of triples on the fly... for chemistry, with 10^60 molecules quite interesting for drug development, you cannot keep everything in a store (yet anyway).

So, two idea:

see how far reasoning goes in your domain of interest
see how to integrate on-demand triple calculation in reasoning

(Oh, and just make sure to keep an Open Notebook).

Pierre · Answer 3 · 2009-11-05 21:40:35Z UTC

1

Yes, have a look at RDFa and GRDDL

answered Nov 5 at 21:40

Pierre
155●2

Cheers - I've read a fair amount about RDF/RDFa but was unaware of GRDDL. Will look into this – Malachi Nov 5 at 21:57

Adding semantics to the Web - Needing ideas for a project

3 Answers