Researchers at MIT have created a pictorial image of the English language and categorized it is a Visual Dictionary. Now Visual Dictionaries have been around for over 20 years and were mainly used to teach vocabulary and reading to children and immigrants used by parents, teachers and translators.

This latest online version is not really a dictionary in the usual sense, but simply a massive visualization of all the nouns in the English language arranged by semantics. The list of nouns was obtained from Wordnet, a database compiled by lexicographers, which records the semantic relationship between words.

Each of the tiles in the mosaic is an arithmetic average of the most commonly associated images relating to one of 53,464 English nouns. A total of 7,527,697 images were obtained using Google’s Image Search and other search engines, with each tile being the average of 140 images. The average reveals the dominant visual characteristics of each word. For some, the average turns out to be a recognizable image (some say it looks like a Jackson Pollack painting); for others the average is a colored blurry smudge.

Using the combined database, the MIT Computer Science and Artificial Intelligence Lab extracted a tree-structured semantic hierarchy to arrange tiles within the poster so that their semantic distance gives the proximity for any two tiles. The relationship between the visual and semantic correlation of our language is shown by the extent of visual clustering within the poster as can be seen at: 80 Million Tiny Images.

Since MIT rarely does anything purely as an arts-n-crafts exercise, there was a highly technical goal behind the process. Computer Vision researchers are currently investigating methods of training computers to recognize and localize thousands of different object categories in complex scenes using a key component of the algorithms.

While the dawning of the digital visual dictionary does not seem too far in the future, thus far the MIT Linguist researchers have not talked about the commercial applications of the project, but the use of Google as the medium by which the images were sourced, smacks of a impending Google application of the work. The MIT research explores how the billions of images available on the Internet can be used to train computer models for object recognition and relate that image to a word without a word tag.

Interestingly the use of the Google in this study probably holds some of the Keys for the future of Search technology, as Images are already a part of the search engines logarithm. Imagine a search engine that would look truly at all of the elements of a post or page, not just the words and their semantics, but all imagery as well, to rank that pages usefulness to the searcher. It would mean that Webmasters would have to update their SEO policy to include a visual element on web sites as well, and the job of SEO writer will truly become one of SEO Designer utilizing the visual dictionary to design relevant and searchable images.


Want more Info to make money online ??
Subscribe to the Expert Earning feed for more great tips by RSS | by Email



Similar Posts