A simple critique of TILE

I would like to echo Annie’s thought that I feel a little awkward critiquing DH projects of this kind given how new a lot of this is to me. I was drawn to T.I.L.E. as a user, and so while I may not yet be able to discuss the project’s inner workings, I can assess its usability. I cannot come up with any single idea that would improve the project other than that it must improve the exposition of how it works and what its purpose is. I will begin by listing a number of general mistakes that occur in the “Sandbox” version of the TILE tool:

  • There is no home button to return you to the TILE site once you have clicked on the Sandbox
  • OCR appears to be a critical aspect of the project, yet there is no explanation of what attention was paid to improving its results.
  • Often but not always, when attempting to create an “Image tag” box, the selected area appears off the scanned page instead of where the user’s cursor is creating the box or ellipse.
  • Once this incorrectly created image tagging square appears off to the right of the scanned image, there is no way to move it back to its proper location.
  • After performing a line recognition, there is no way to zoom in to judge how well this was done.
  • A composite scan containing a text/image composite would have been very helpful as a demonstration of the image tagging tool.
  • I have trouble understanding the choice of the phrase “Semi-Automated line recognizer.” Especailly when combined with the shaky functionality of the Sandbox, it gives the impression of an “almost working line recognizer.”
  • There really MUST be some visual connection between the dialogue box that pops up when image tagging, and the specific location of the image-text being tagged. I would advise looking at software such as Nowcomment for the functionality/UI for TILE.

To take a wider scope, there appears to be a general lack of exposition. MITH must be aware of how traffic will get to their site. I think it is not uncommon for users to discover the site in just the way I did; by clicking on one of a series of links relating to text-based DH projects. With this in mind, the problem that TILE is attempting to solve must be placed in the foreground. A lot of these problems could be solved with a simple video demonstrating the project at each stage; from scanning to OCR, to line recognition and finally image tagging.

TILE introduces a lot of issues in terms of exposition of a DH project. More text explaining what inspired the project, and what it does is not always the best way to introduce and “hook” new users. Still, MITH’s decision to use such minimal text does not help. It should also be pointed out that the small screenshots devoted to the various tools are poorly chosen and fail to demonstrate anything. TILE easily could accomplish its stated goals, yet it seems the presentation and functionality of the sandbox would scare off or confuse most users before they realize the strengths of the project.

Ed was a 2011-12 Scholars' Lab Fellow and Praxis Fellow, and is a PhD candidate in the McIntire Department of Art.

Comments are closed.

Archives