- : May be added later. Αν είστε ανυπόμονοι, ίσως αντέξετε προσωρινά τα… σπαστά Ελληνικά του Google Translation, ΕΔΩ)
Today I started writing this blog-post after testing Powerset, using human questions about Macedonia (a well-known SEMANTIC dispute between Greece and another Balkan state). After all, the infamous «naming dispute» between Greece and the Former Yugoslav Republic of Macedonia is -to a large extent- a SEMANTIC DISPUTE.
I am a registered beta-tester of the Powerset Semantic Search Engine (www.powerset.com) , where one can ask human questions and get (more-or-less) human answers, based on Natural Language Processing (NLP) from a Knowledge-Base that uses Wikipedia.
- Semantic search-engines are multiplying faster than rabbits! Two months ago, a Semantic Web Search Engine Roundup was posted by James Simmons. Among these, Natural language Search Engines are perhaps the best.
This morning, Andreas Blumauer, an Austrian blogger working for a Semantic Web Company in Vienna, published a post titled «Natural language search – a new breakthrough?». After reading Andreas’ post I also tested www.ask.com, another natural language search engine, open to anyone, not just beta-testers. Here are the questions and some of the results:
Question 1: What is Macedonia?
- (answers were collected here: http://omadeon.com/nlp/macedonia.html)
Question 2: Who are the Macedonians?
- (answers were collected here: http://omadeon.com/nlp/macedonians.html)
Here are some more sophisticated human answers:
Question 1: «Macedonia» can mean a «geographic region», OR a sovereign state (F.Y.R.O.M), OR a large province of another sovereign State (Greece), OR an an ethnic identity (of Macedonian Slavs) OR another (very different) ethnic identity (of Greeks from Greek Macedonia), OR an Ancient Empire (of Alexander the Great), etc.
Question 2: «Macedonians» can mean anyone among a variety of identities: Greek Macedonians, Slav Macedonians, Albanian Macedonians, Bulgarian Macedonians, even Jewish Macedonians (and other ethnic groups that live -or have lived- in the region of Macedonia).
You can experiment with more natural language questions, e.g. «Who was Alexander the Great», or «Where did Macedonians come from», or «Who are the real Macedonians». (After I collect more results, this post will probably be updated).
- WARNING: This is a new search engine technology, and (although Natural Language Processing has existed for more than two decades) it can sometimes go wrong. So, expect impressive results, but also a lot of mistakes!
- Unfortunately, the second search-engine (www.ask.com) appears to be considerably biased, in the naming dispute, in favour of Slav Macedonians. However, before condemning it, consider the possibility (almost a certainty) that the cause of this is not the fault of the search-engine itself, but the overwhelming propaganda presenting inhabitants of FYROM as the ONLY «true Macedonians»!
- So, why is Powerset more unbiased? -Probably because it is based exclusively on ONE source (Wikipedia) where the balance between the 2 sides is better than in the whole World Wide Web (sadly for Greece).
Gooogle has serious competitors, now…
Only a few dozen results have been uploaded from Powerset. More results will be uploaded in a short while.
- Unfortunately, the Powerset Natural Language Search Engine is only available to registered Beta-testers, at the moment. The results can’t be accessed directly in public. They are instead uploaded by hand, via ftp to my site.
- I am also a registered tester of the «TrueKnowledge» Semantic Search Engine (based in Cambridge, UK). New posts about this will probably appear in this blog, in the next few days.