Nitin Karandikar of the Software Abstractions Blog had this to say about Cognition:

May 7, 2008

Cognition Technologies, which focuses on Semantic natural language processing technology, was named by KMWorld as one of the top 100 Companies That Matter in Knowledge Management for 2008.

Says Cognition CEO Scott Jarus:

One of the biggest barriers to building a natural language understanding system is to build the semantic map and the dictionary with details of the syntactic behavior of words (i.e. how words behave within context). Cognition’s team has spent more than 20 years building this capability into Cognition’s Semantic NLP for the English language … and our technology is commercially available today!

Semantic search and NLP technologies seem to have arrived - they are generating a lot of buzz lately. In addition to mainstays Hakia and Powerset, there is a spate of new entries, including Cognition, BooRah and eeggi. We will be reviewing some of these new alternate search engines on this blog in the near future.

Congratulations, Scott and the Cognition team!

We recently received a question from an organization that was interested in putting Cognition’s Medline search on their site.

Here is the question, followed by a great answer from Dr. Kathy Dahlgren.

Question:

“Can you tell me what additional benefits would come from linking to your site that would not be present if we simply linked to PubMed?”

Answer:

CognitionSearch has far greater recall than PubMed, and eliminates irrelevant documents.

The documents eliminated are:

1) When the query contains a series of words that form a phrase, such as “obsessive compulsive disorder” or “polycystic kidney disease”, only those documents in which the phrase or its synonym (such as OCD, or PKD) are retrieved, not documents in which parts of the phrase a found in various sections of the abstract.

2) When the query contains ambiguous words, only the words (or their synonyms) in the intended meanings are retrieved. For example, in the query “eating eggs on a low-fat diet”, only the meaning of egg as a food is retrieved (and synonyms like omelet), but not the meaning of ovum. On the other hand the query “fertilized eggs” retrieves to the ovum and not the food.

3) Documents are eliminated when query terms are not all found in one sentence or the title. If the user desires, the user can choose to override this property of the search, and permit retrievals from various sentences in the advanced search feature called “Find search concepts anywhere in the document”.

CognitionSearch’s results are far more comprehensive than PubMed’s because of synonymy and reasoning in an ontology (or taxononmy). Here are examples of queries where CognitionSearch gets good results, and PubMed retrieves nothing (even though they match anywhere in the document).

• breast milk, toxic exposure and cancer in Canada (1 for Cognition, 0 for Pubmed)
• tests for breast cancer PAH genotoxicity (1 for Cognition, 0 for Pubmed)
• fatal incidents of DKA (33 for Cognition, 0 for Pubmed)
• tri-cyclic antidepressants suicide and mourning (1 for Cognition, 0 for Pubmed)
• 1l3rb1 (142 for Cognition, 0 for Pubmed)

Many users report that with Cognition they have been able to find research results or other information that they had been looking for a very long time and had not been able to find with the standard PubMed search.

Dr. Daniel Albro’s background

What was your first computer?

I saved up my money in elementary school, mostly by skipping lunch, and went out and bought an Atari 2600. This inspired my Dad, who went out and bought an Apple //e not too much later.

How old were you when you first wrote a computer program of any kind and what was it?

10 PRINT “Dan Albro is the ruler of the universe!”
20 GOTO 10

Something like that… Before long, though, I had learned to program the thing in assembly language, modify the operating system to change all the prompts to silly things, all that kind of stuff. A lot of my early programs were silly games in which you tried to kill turkeys (I was a misguided, turkey-hating youth…).

Who are your computer science heroes?

Oh, probably Donald Knuth would be the big one, then McCarthy, Steele, Abelson, and Sussman (Lisp and Scheme folks), Guido van Rossum (Python), Hopcroft and Ullman. The list goes on pretty far…

What are your personal goals for Cognition’s Semantic NLP?

I would like to extend the technology beyond English, and I would also like to get back to where we were before we decided that the technology has to be fast (we decided some time back to rewrite the system in C++, before which it had been in Prolog), and be able to process a document set and actually have the computer be able to answer, in English, questions about what it learned from processing the document set. I don’t think that kind of functionality is *that* far out of reach…

When you’re out at a restaurant and the host says “Your server will be right with you” do you check your hand-held device to monitor your servers’ performance?

No :). The only thing I use my palm computer for, pretty much, is to quiz myself on whatever foreign or dead language (I like the dead ones) I’m working on at the moment. Besides, I’m bad enough at cooking that I know I have to rely on waiters an awful lot…

What’s the last book you read?

I tend to read a few at a time, but the last few included some old Scrooge McDuck comic books I was reading with my son, a book on Syntactic Argumentation, the book companion to the PBS series on American Roots Music, Pete Seeger’s Handbook on the 5-String Banjo, and the Spanish language version of the last Harry Potter book, to work on my Spanish…

USC or UCLA?

MIT, then UCLA…

Kathy joined Barney Pell of Powerset, Willian Tunstall-Pedoe of TrueKnowledge, and others for a panel on semantic and search technologies on Monday, April 21st. Everyone on the panel did a great job of explaining what his or her technologies did, but as it was the last panel of the day and the crowd growing restless, the panelists decided to get provocative. Kathy had a very funny and thought-provoking comment about how at some point the technology world will have to teach computers language because statistical and math systems are failing in modern computer applications.

Charles Knight of AltSearchEngines.com did a great job of assembling a number of highly engaging panels showcasing many search and semantic companies. We were honored to be asked to speak and tell people about Cognition’s semantic enabling (with no tagging necessary!) technology.

Hi! Welcome to the CogBlog! This is the home of Cognition Technologies’ employees and their musings. Our founder and CTO, Dr. Kathy Dahlgren has been building out Cognition’s Semantic NLP™ for the past 23 years, so she will be chiming in regularly. Plus, we have some other amazing linguists and computer scientists here too. Nearly 25% of our current team has a PhD. Okay, it’s a small team, but still!

Since there are so many linguists here, we thought we’d go straight to the definition of what we mean by Cognition’s Semantic NLP.

• Semantics is the sub-field of linguistics that is devoted to the study of meaning, as expressed by words, phrases, sentences, and even larger units of speech or text.
• Natural Language Processing (NLP) is a sub-field of artificial intelligence and computational linguistics. It studies the problems of automated generation and understanding of natural human languages by computers.
• Cognition’s Semantic NLP^TM is technology that “understands” word and phrase meanings within context in modern computer applications. Cognition’s mission is to make its clients’ technologies and applications more human-like in the understanding of language and more profitable.

The Long Tail of Semantic Understanding

It’s obviously not a unique concept to us. But we feel it’s an incredibly important one when explaining the benefits of semantic enhancement of any technology.

The Long Tail of Semantic Understanding

Most technologies that do any kind of text processing for an end-user’s benefit can only really do acceptable with a set number of words. With real semantic “understanding” going on in a technology product, context is understood; entire sentences are understood; synonyms are understood; you get the picture. It all goes to helping companies better comprehend their users’ intent. We think that’s a pretty actionable asset to any technology.