Ramesh Jain: "In the recent issue of IEEE Multimedia Magazine, Dick Bulterman wrote a very interesting piece "Is it Time for a Moratorium on Metadata?." Dick is a very well known researcher in multimedia authoring environment and was a leader in the team that developed SMIL. I strongly recommend that people interested in use of metadata read this article. Particularly, his humorous story at the beginning of the article is very revealing. ¶ This article has many interesting arguments that are presented using three well well-known metadata standards: Dublin core, MPEG-7, and Semantic web. I will not go in details here but will mention here two of my favorite points:
1. During the past decade, it's become clear that for electronic assets, locating text is best done using the text itself rather than relying on metadata, because the context of the search is defined at query time rather than catalogue time.
2. Most of the metadata for text files, images and video, if not all of it, is assigned using automatic techniques based on once entered parameters combined with clock, camera parameters and similar information at the creation time. These parameters are usually useless in the context of the search and even when could be useful are usually wrong because people don't change parameters (like clock in the camera) when they move. Granted that one day these parameters may have some correcting things or more correct contextual values like the clock in you mobile phone that is set by the local carrier. But today most of this data is wrong.
..." Link: Ramesh Jain.