ideambulate

People remember images more easily than they can remember complex strings of words.  Furthermore, images can automatically evoke complex sets of associated thoughts and emotions.  We should leverage the associative power of images to help sort through thick information sets.

I’m talking about web-searching with an unconventional “semantic” twist.

The tools to make this work are all in place.  Several image formats support textual metadata packaging.  This metadata could be used to store words associated with the image and their relative weights.

The metadata could be generated by direct human characterization, image parsing and object recognition, or by trawling “nearby” sources of textual information.  Think web pages that store or embed the image and their own contextual and general metadata.  Google and Flickr, I’m looking at you.

The metadata could be further categorized by type.  To wit, associated words might describe the emotional, allegorical, compositional, or literal content of the image.  Then, users could choose to use any image as a versatile query tool.

For example, consider an idyllic picture of a child and a dog playing happily on a swing (photo courtesy of teresia).

One user could use this image to search for other pictures containing children and dogs.

Alternately, the user could use the images to search for text-based articles relating to “childhood” or “pastoral” or “happy” subject matter, simply by shifting the focus from literal metadata to abstract metadata.

Image-metadata-based-filtration boils down to a rich-content, easily-comprehended “key” used as a pointer to complex, versatile semantic datasets that in turn will aid in navigating the endless ocean of information.