Cog Blog

With traditional keyword search engines, such as those used by Google, Yahoo!, and others, finding the best medical research document within complex data sets, such as MEDLINE, is very difficult without the use of complex Boolean equations and a deep understanding of the many permutations of technical synonymy. Cognition’s SemanticMEDLINE has the ability to target and locate these types of data that are otherwise hidden in masses of information because of its comprehensive Semantic Map (particularly deep within the health sciences discipline) and its unique ability to “understand” the meaning behind words, phrases, and idioms.

Read the full article here.

Cognition Technologies is talked about in Merrill’s latest press release about their product:

“To improve early identification and review of key evidence, version 5.6 also features conceptual search tools, offered via a licensing agreement with Cognition Technologies, Inc. Unlike keyword searches, conceptual searching interprets the meaning of words within the context they are used to help computers “understand” concepts, significantly increasing the precision of the discovery results. “Cognition’s Semantic NLP program is revolutionary in the methodology used to conduct complex searching through algorithms. The semantic natural language search component allows our clients to conduct more meaningful content searches and uncover relevant evidence more quickly.” said Harold Leach, senior vice president of IT for Merrill’s Electronic Discovery Services.”

Read full release here.

LOS ANGELES – July 23, 2008 —Cognition Technologies, a next-generation Semantic Natural Language Processing (NLP) company, announces a quantum improvement in the application of NLP technology with the introduction of Semantic MEDLINE™ – the 18 million article abstract database of complex health information published by the National Library of Medicine. This new free service at www.SemanticMEDLINE.com enables complex health and life science material to be rapidly and efficiently discovered with greater precision and completeness. This marks the first time that users can employ a natural, conversational sentence structure to find the most complex studies within the MEDLINE dataset.

Read the full release here.

Saul Hansell wrote an interesting post on the NY Times blog entitled “Google and the Real Search for Meaning on the Web.”

We joined the conversation with this comment (link) because people still don’t understand how critical a vast and complete Semantic Map of the English language is to true understanding of text:

This is a great discussion! However, what we don’t see anyone talking about is what we believe is the key to making Natural Language Processing successful – the Semantic Map.

To advance NLP beyond its current limitations is a very difficult problem because you have to unravel the complexity of language. In particular, individual words have multiple meanings, and at the same time, a given concept or meaning can be referred to by multiple words. To understand which meaning of a word is correct requires an understanding of context, but in order to fully understand context requires that words and meanings be analyzed or “curated” one at a time into a vast Semantic Map.

Cognition’s Semantic Map of the English language, which has been built over the past 23 years, is complete, robust and unique, and represents the key market differentiator for the company and its technology.

Utilizing a Semantic Map differs from Google’s algorithms in that Google relies primarily on string pattern-matching, and does not try to interpret the meaning of the words within context. With a Semantic Map, it is possible to determine word meanings and therefore understand meanings rather than simple string pattern recognition. For example, in Wikipedia, in response to a query “strike oil in California”, Google returns documents about “strike commander” (a flight simulator), striking workers, hunger strikes, etc., in their top 5 results. With a Semantic Map, the most precise and complete retrievals about discovering oil in California are returned, because the meaning of “strike” in context is interpreted as “discover”, as opposed to “attack militarily” or “walk out”.

Cognition Technologies has changed the NLP paradigm through its unique and complete combination of a complete Semantic Map and linguistic elements to optimize semantic understanding:

Morphology

• The various forms of word, e.g. singular, plural, tense

Syntax

• The grammatical structure, e.g. verbs, nouns

Semantics

• Word and sentence meaning, augmented by synonymy and taxonomy

Spelling

• The various ways words are spelled (or misspelled)

We encourage you to try it for yourself at www.cognition.com.

Charles Knight has the story:

Cognition has completed an institutional and individual financing round totaling $2.7 million.  Investors include Draper Associates  Tim Draper), Fingerhut Ventures, and a personal investment by the company’s  CEO, Scott Jarus.

Link [AltSearchEngines.com]

Anthony Ha of Venture Beat does a great write up about Cognition Technologies:

Where Powerset focuses on the structure of a search query, Cognition Technologies has been building a “semantic map” — basically, a big dictionary that allows Cognition Technologies’ products to actually understand the meaning of the words in your query, and therefore understand what you’re actually asking for.

Link [Venture Beat]

Great discussion on SEO for semantic search engines here at SearchEngineLand.com.  They mention Cognition in their write up.

The writer, Pierre Far, compares two queries on Cognition, Hakia, Powerset and Google.  On the first query (Who built the empire state building?), the writer states:

The Cognition highlights of their first hit is probably the best as it’s quite thorough. It also highlights sentences that take a wide interpretation of the verb “build”, highlighting sentences talking about the designers and the construction company.

On the second query (Who makes diet coke?), the writer doesn’t mention Cognition’s result.  Interestingly, if he had capitalized Coke, he would have immediately found the answer!  This is another aspect of semantic search engines that people need to remember:  they are case sensitive!

Link (Search Engine Land)

Ever wonder what top-down and bottom-up semantics mean?  Dr. Kathy Dahlgren explains all to the readers of AltSearchEngines.com.

Here’s an excerpt:

In “bottom-up” semantics, individual pages are hand-tagged with semantic categories.  In “top-down” semantics, a developer models the semantics of a vertical domain (such as, farm equipment, movie cameras, etc) and the kinds of things users would want to do in that domain, and then links Web pages into a meaningful series of information views and user actions.

Link [Alt Search Engines]

Read Dr. Kathy Dahlgren’s post on AltSearchEngines.com entitled “Scientifically Measure Search Precision and Recall.”

This post is to add to the dialog about precision and recall which are standard measures of Search engine performance. Precision is a measure of retrieval accuracy calculated by dividing the total number of relevant retrievals by the number of all retrievals generated by the Search.  Recall is a measure of the extent to which relevant material in the total document base is found.  It is calculated by dividing the number of relevant retrievals by the total number of potentially relevant retrievals in the document base.

Link [Alt Search Engines]

By Lesley Jezierny, Computational Lexicographer

Most anyone reading this will be familiar with the concepts of the Semantic Web, but even among this audience, the answer “I’m a computational lexicographer” to the “What-do-you-do?” question can be met with a blank stare. It’s not like saying we’re Web designers or programmers; most people just have a vague feeling that our work is something to do with words and computers. Oddly, people often assume I’m a translator.

So I’ve come up with a 30-second response and I simply ask this question: “What’s a balloon?”

(more…)