Google Speak and Semantic Search

Next-generation search engine tech aims to understand natural written language

By Gregory Mone Posted 02.22.2008 at 12:04 pm 4 Comments

Powerset Semantic Search:

A handful of start-ups are getting ready to challenge Google’s predominance in the Web sleuthing world by offering what’s known as “semantic search.”

The companies—Powerset, Hakia, Cognition Search, Lexxe—are trying to develop a search technology that would allow you to look for material on the Web while writing like a normal, educated human, instead of just entering keywords, and dropping all the in-between stuff that gives us those wonderful things called sentences.

It’s a bold aim. For example, Powerset’s stated goal is to build “a natural language search engine that reads and understands every sentence on the Web.” Forbes has an interesting analysis of the field, and points out one key competitive flaw: At this point, these companies don’t actually offer an advantage over Google. One analyst says they might be able to develop one, but doesn’t expect it anytime soon. And Google? Well, as you might have guessed, the Web giant, surely well on its way to develop semantic technologies of its own, isn’t too worried.

More on the Semantic Web: Wikipedia and W3.org.

Previous Article: The Physics of Nascar

Next Article: Researchers Create Self-Repairing Rubber

Want to learn more about breakthroughs in electronics, medicine, nanotech, and more?
Subscribe to Popular Science and enter to win $5,000!

4 Comments

link to this comment

subscriber

02/25/08 at 6:38 pm

Its quicker just to write in keywords anyway, I'm not seeing any point other then happy english teachers.

link to this comment

christopher.roush

03/04/08 at 1:29 pm

"subscriber," I imagine you mean "other than." No offense intended; simply illustrating my point. Read on.

I like a unified form of language because it allows us to understand each other. Your confusion of homophones isn't a total stumbling block for me, but it might prove difficult for a non-native speaker, given the fact that the two words in questions have totally unrelated meanings. I've also run into spelling and grammar errors so terrible that I literally couldn't decipher a meaning.

I'm also pretty sure that what's at stake here isn't just time and effort, but intuitiveness and viability. For example, my wife can never find what she's looking for because she's not familiar with exclusive boolean terms like "-" "not" "+" and quotes. The way you enter a search string can drastically change your results, especially with all the embedded keyword garbage that fills those sites nobody actually wants to see, but somehow everyone always ends up at.

link to this comment

dontbother

03/07/08 at 8:50 am

I assume language == English and that “a natural language search engine that reads and understands every sentence on the Web.” really means "a natural language search engine that reads and understands every English sentence on the Web.”.

link to this comment

kardelen133 (not verified)

07/02/08 at 4:14 am

very nice blogs.
thanks.
Kaza dün gece yarısı evden eve nakliyat Osmancık İlçesi D-100 karayolunda evden eve nakliyat meydana geldi. 28 yaşındaki Halil İbrahim Gezer evden eve yönetimindeki 34 JMK 02 evden eve plakalı yolcu
otobüsü, Şenyurt Mahallesi'ndeki nakliyat Karayolları 71'inci Şube
Müdürlüğü önünde Emrah Çevik'in kullandığı nakliyat 19 EE 375 plakalı motosiklete arkadan çarptı.
İddiaya göre arka lambaları evden eve nakliyatyanmadığı için farkedemediği motositlete çarpmasıyla Çevik olay yerinde hayatını kaybetti. Kazada otobüsteki yolculardan ön kısımda oturan Silver Dumancı (19) ve
Özcan Görmez (12) hafif şekilde yaralandı. Osmancık Devlet