Yes I know … this is not about MPEG-7 and multimedia, but on the other hand its very interesting Some months ago I started to analyze the structure and usefulness of folksonomies (see here or wikipedia). Especially I was interested in methods for tag similarity computation and the extraction of structure from the chaos of tags and resources. Today I finished an interesting little project, which allows me to create a tag network out of samples taken from del.icio.us. The shown image (click to enlarge) shows a network generated from a list of popular tags from Dec. 28th 2007 in the afternoon. The color of the edges indicates the amount of similarity. Note that the used method does not result in a symmetric measure and therefore an undirected graph, but is normalized based on the source tag and is visualized as digraph.
The similarity of tags is based on the co-ocurrence analysis published by Peter Mika in Ontologies are us: A unified model of social networks and semantics. The visualization has been done with graphviz. The sample from del.icio.us has been taken by monitoring the RSS feed for recent additions.
George L. sent me a nice example of a website integrating image search features: Become.com offers a search ability for color based retrieval of clothing. Just use the “find by color” feature in the right hand side column at the top and you will be impressed how much fun image retrieval can be .
One can also observe how inaccurate algorithms get when the background is integrated in the color histogram (e.g. with this search parameters, where I searched for grey and found also some blue images – or even in the presented screenshot where an obviously not red dress is shown before a red background). However this is a still unsolved problem: How to separate background and foreground in still images?
Today I released Lire 0.5.1: Since Lire already supports Color histograms (with the MPEG-7 ScalableColor descriptor), functions for searching for colors have been integrated by by adding a searcher for color only search operations, a document builder restricted to color and a document factory for fast and efficient creation of documents describing images with one color only.
Download Lire at the sourceforge.net project page.
The – not quite famous but really cool – RSS & Information management suite Omea Pro of Jetbrains has gone open source. As JetBrains announced yesterday the whole suite (not only the free Reader) is now available free of charge in both binary and source.