Help with thesis idea - REST and Enterprise Mashups


I have to define a project for my master’s thesis, and I think it will
be about REST and mashups, though the exact research topic is not yet
defined. It would be great to have your opinion about the project
outline that I describe below.

  • Does it sound interesting to you?
  • Have you tried or seen something similar?
  • What would you like to see in the study?

As you may know there are some new commercial tools in the market, that
try to bring the functionality of yahoo pipes or microsoft popfly to the
enterprise (see the banners at to know
some) The idea is to give employees the possibility of building mashups
using data sources from inside the company and from the web.

I think that the idea of having internal mashup tools becomes even more
interesting, if your company’s applications are “RESTful”. For example,
if you made your app using a “RESTful rails” approach, then all your
models would be RESTful resources, that you can set as data sources to
be mashed up with a mashup tool.

I have an application built using Ruby on Rails that is 1.5 years old
now, with about 60 models. I plan to migrate it to Rails 2, convert all
its models to “resources” and feed them to one or many mashup tools, to
see what the applications’ users can do with it.

My research would consist on one or more case studies. The objective
would be to find how great are the advantages that you get when using a
mashup tool combined with feeds from an application built using the REST
architectural style. Specificaly:

  • Do you get more advantages from using feed from a “RESTful”
    application than just using feeds from a traditional database? (i.e.
    let’s remember that building a RESTful web app, implies that the
    database will contain more tables. The question is: are those tables
    useful data sources for a mashup?)
  • Would semantic stuff improve these advantages (for example, using

Other considerations:

  • If I try currently available mashup tools, I could try to figure out
    what kind of mashup tool was more effective in the study? (grafical
    widgets, scripting language, mashup by example, etc)
  • I could do my own mashup tool for this experiment.

What do you think? All comments are welcome.

I appreciate your help.