Wading through mountains of information it soon becomes evident that context is critical.
The problem exists for any type of digital information, including email message, digital photographs, documents, spreadsheets, etc.
As an illustrative example, the average broadband user has a digital camera and uses their PC to store all their pictures. Unknowingly, they’ve created a media warehouse. In my case, I have a huge digital media library containing in excess of:
- 36,000 photographs
- 14,000 mp3 files
- 2,000 movies
Granted, mine may be an above average library, but in no way are my challenges managing consumer-generated media unique.
As an example, what is the best way to find that one great Christmas picture in a pile of thousands? You know, the one where all the family is smiling? Unless it was flagged it at the time, I’ll need to sort through randomly named directories named DCIM0012 / Christmas 2006 / 06/12/25, etc. depending upon how the files were captured and copied. Then, the files names are sequential and not in any way meaningful. The process is neither easy nor pleasant.
Most image management programs today take the easy path. Picasa and iPhoto have a sequential “film roll” concept, and sort by date taken or date imported, respectively. Adobe Photo elements adds a calendar view to sort by data taken. However, these excellent programs create an sequential stream but do not create pr imply any meaningful context to the media.
Professional photographers heavily rely upon IPTC-NAA keywords in the EXIF metadata of their photographs. However, they’re selling their product and must properly package it to make it a marketable commodity for the news wires. As a home user, categorization and EXIF editing tools are not prevalent, cheap or convenient.
A number of programs exist for adding context to email, which is the easiest problem to tackle since each message can be searched, screened and filtered by date, sender, subject and content. Text indexing is extensible to other types of documents.
But what about consumer generated digital media? Here are three ideas for the automatic generation, implication and extrapolation of context info a library of consumer generated digital media.
Face Recognition
Face recognition is an interesting application for generating automatic libraries of home photograph, and has made great strides lately. The exciting part about applying automatic face recognition to a personal library is that present algorithms require training for accuracy. However, a vast majority of personal photo libraries are of close family members.
Geocoding
Another is location or geocoding, or marking the geographic location of where and when a picture was taken. Most camera phones now have GPS capability and some new digital cameras are including this too. This makes is possible to view photos by retracing steps, and adds a geographic reference to photographs of buildings, animals, people and events otherwise unmeaningful without explanation.
Scheduled Events
On unexplored option for the automatic marking of digital media is inferring events or context from a personal calendar. For example, if one has “Junior’s 5th Birthday Party” as an Outlook / iCal / Google calendar event, it’s entirely probably likely that the dozens of photographs taken during that time are related.
Managing your media by hand is a daunting task. However, the judicious application of technology should be able to assist in solving this problem.
– JMC
References: Photographic
References: Geocoding
Programs for embedding IPTC information into digital photos
I just purchased an HP TX2000 laptop. It’s a little heavy but is a truly wonderful piece of hardware. Buy one now at 
Out of the box, Vista doesn’t like Winamp, or Winamp doesn’t like Vista.