Business - Written by Naumi Haque on Tuesday, June 3, 2008 7:56 - 0 Comments
Using the brain to search like a human
The enemy: Information overload
Cost: As much as $650 billion in the US, or approximately 8 hours a day for the average knowledge worker, according to a recent New York Times article.
Our potential savior: Semantic search
The main problem with using regular search engines to find information is that they tend to retrieve a lot of erroneous results. Search engines like Google and Yahoo! don’t do a very good job of taking into account natural language indicators and transition words to derive the context in which search terms are being used. Semantic search—helping computers identify the meaning of the words being searched—is being touted as the leading solution and a key enabler of the knowledge economy. There are a number of players that are working to develop a good semantic search solution, but the following two strike me as particularly neat.
Back in the fall, I interviewed a company called Accelovation (now NetBase) that is working to build structured databases out of unstructured data on the Internet. The main idea is to use semantic search algorithms to create the first ever search engine for innovation and R&D solutions. The process is described in the diagram below. The solution (now available) is called illumin8.
Source: NetBase Solutions, 2008
More recently I came across an interesting article in Ars Technica discussing how scientists are using fMRI images of the human brain to try and teach computers word associations. Images of the brain produced when viewing different objects were used by the computer to guess the meaning of unknown nouns and verbs. From the article:
The basic problem that faces the semantic web is that while a culture generally has an agreed-upon meaning for a word, it is hard to break that meaning up into symbols that a computer can understand. One way to go about tackling this problem determine what symbols our brain uses to convey that meaning. While we’re still a ways off from decoding the internal symbolic “language” of the mind, functional magnetic resonance imaging (fMRI) indicates that meaning seems to be associative. For instance, when a person is shown a picture of celery, fMRI pictures of their brain will usually show the part of the brain associated with taste activating. The general conclusion is that objects are associated with sensory and motor control regions. Unfortunately, it is not quite that simple, because other parts of the brain, primarily the frontal cortex, also light up, indicating that there is more going on. Nevertheless, the associative patterns are strong enough that a trained observer can accurately guess the object being shown to a subject by watching the fMRI pattern.
In a twist on that observation, scientists have trained a computer using word associations and fMRI patterns to see if it could predict the fMRI pattern of nouns that it had never encountered before. This was achieved by creating a neural network with an enormous text from which to gather word associations. The network was then fed 60 nouns and a set of verb classes—from there it searched this text to correlate the nouns with the verb classes, creating a 25-dimensional model for each noun. Finally the network was trained on fMRI patterns for some of the nouns.
Cool stuff for sure. For those interested in semantic search technology, another good read on the topic is the Economist article, “Improving innovation” from a few months back.
Business - Oct 5, 2010 12:00 - 0 Comments
More In Business
- Facebook, Facebook, Facebook
- Survey: How are you using Facebook, Twitter, smart phones, and other technology platforms?
- Will Facebook be your CRM provider?
- Wiki Banking
- The importance of being competent
Entertainment - Aug 3, 2010 13:14 - 2 Comments
More In Entertainment
- Lessons in collaboration from B.B. King’s
- CL!CK – LEGO’s fun social product development platform
- Peer Pressure 2.0: Farmville
- Online gaming more than just fun
- The NFL – The most protective league, attempting to control the uncontrollable
Society - Aug 6, 2010 8:19 - 4 Comments
More In Society
- Balance: customer receptivity vs. customer revulsion
- The Net Gen: Too plugged-in for parenting?
- Are you addicted to social media?
- The privacy discussion we need to have
- “The Data-Driven Life”: Who’s not interested in discovery?