August 31, 2004

Searchable Photos: Part IV - XHTML Questions

The full “Searchable Photos” thread now includes:

  1. How To Make Photos More Searchable (Part I)
  2. Searchable Photos: Part II
  3. Searchable Photos: Part III
  4. Searchable Photos: Part IV

I have finally gotten my Web Photos Pro product to a place where I feel like I can breathe for a bit. The last three or four months have been an endless cycle of new features, bug fixes, and beta releases. So much so that I’ve felt a bit like Judy in Punch in Judy (ergo the photo that adorns this entry) – though I suppose it could also be due to the broken rib I got last week while mountain biking, the rib that’s been making me yelp with pain whenever I sit, stand, lie down, roll over, or rotate in just about any direction.

With this latest release, 1.0b12 (beta 12), out the door, I feel like I can get off the treadmill for a minute and do a couple of things that don’t involve programming – things like pay the bills, clean up my office floor, write documentation for Web Photos Pro, write some entries for this weblog, and get the PhotoRSS web site fleshed out. And oh yes, spend some time with the family.

Never one to do first things first, I decided to spend some time on the PhotoRSS web site. But I got stuck pretty quickly because I wasn’t sure how to go about extending XHTML. I know it can be done, but I’m not sure of all the details, and I don’t know if it’s possible to do it the way I’m thinking about. I looked at the docs, and I looked at various examples, and I found my eyes glazing over. Very quickly I came to the conclusion that it’s just like doing my taxes – it’s a heck of a lot quicker, and a lot less painful, to call in the pros. So here I am, calling all you XHTML pros.

Before My Questions: Some Background
(If you’d rather skip straight to the chase, you can jump down to “The Four XHTML Questions” below.)

Those of you who have been following this blog know that I’m very interested in making my photos more searchable. There are various ways that this can be done, and I’ve started to write up three ideas on the PhotoRSS web site.

Of the three ideas, the first two are relatively straightforward. The first involves creating a photo-centric XML namespace for RSS. I’ve done something similar before, when I was working   with   some   interesting   characters at Wired Digital. One of my skunk projects involved adding RSS 2.0 output to Lycos New and Wired News searches. One of the things I wanted to support was the ability to page through the RSS search results, and to do that I defined an XML namespace xmlns:WiredNewsSearchResults which provided things like number of results, next and previous links, etc.

The second proposal involves creating a new RSS-like format for photo syndication. While at Wired Digital I did a lot of work with XML. Not only did I architect, design and implement the first all-XML-transport site at Lycos (all communication between the front-ends and eight different back-ends used with Lycos News was via XML), but I was instrumental in getting Lycos to change all of its middle and back-ends to serve up XML, so that data could be reused anywhere on the network. No longer was adding a stock ticker to a web page a major project involving cross-group coordination and development, rather it became a simple 15 minute exercise to get and format some XML data.

But it’s the third proposal, the one about defining an XHTML tag that’s giving me trouble.

What Do I Want?
I want search engines to stop having to guess what my photos are about, I want them to know what they’re about. When I search for “Sunset over Daymer Bay” on Google I want to get back this and this, not this.

The way things are today, the best any search engine can do is look at the text and html near an image, and from it hazard a guess as to what the image is about. When you look at this photo search, I’m not sure whether we should be pleased by how many correct pictures were found, or disappointed at how many unrelated pictures there are. Whatever your thoughts, I think we can all agree that not only would it be nice if photo search results were more accurate, but it doesn’t seem as though it should be a particular hard problem to solve.

Two Different Ways To Make Photos More Searchable
To make photos more searchable the information about the photo – the title, description, keywords, location, etc. – must be made explicit. And there are only two ways that this can be done: 1) you can embed the data in the photo, or 2) you can bind the data to the photo’s <img> tag.

Many cameras already embed data in your photos, using the EXIF portion of the JPEG header. The big benefit of EXIF is that the data stays with the photo – link the photo to a web page and it’s there, email the photo to me and the EXIF data gets emailed too. But there are so many downsides – the data is invisible to the human eye, the photo has to be opened to see if it has embedded data, EXIF data is not as structured as it needs to be, the format has some severe limitations, most resizing programs do not keep EXIF data intact, EXIF editing is difficult to do, and the data cannot be context dependent – that it’s pretty obvious that EXIF is not the right way to solve this problem (at least not in the foreseeable future).

This leaves us with only one way to make photos more searchable – by finding a way to bind a photo’s data to its <img> tag. And this problem, how to best bind photo data to the html <img> tag, leads me to ask the following four questions about XHTML.

Continued at »

Posted by: Frank @ 12:40 pm — Filed under:

Comments »

No comments yet.

RSS feed for comments on this post.

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>



In an effort to control spam, please fill in the result of the equation below