Orange Forum • View topic - GSoC


General discussions about Orange and with Orange connected things (data mining, machine learning, bioinformatics...).


Postby pedromlm » Fri Apr 01, 2011 14:48

Hello. I am pedromlm, a student of IE in the University of, in.
I am interested in working with the "social platform for Orange" idea. I have had past experience working in web services manly written in PHP with a lot of MySQL, so even if I have never develop a social network I have no doubt that it is something I can do. I have also explored, with the help of tutorial and trial and error, Orange Canvas and Orange scripting with Python. How ever I am in the first year of college and I haven't studied, or explored in a personal level, data mining nor machine learning. I wanted to ask if it is possible for me to build the social network with the repository with all the features of tagging, commenting, etc, so I could participate in GSoC without that knowledge of machine learning and data mining?
Last edited by pedromlm on Tue Jan 17, 2012 15:41, edited 1 time in total.

Re: GSoC

Postby Blaz » Fri Apr 01, 2011 15:17

Dear Pedro,

This is hard to say ("I wanted to ask if it is possible for me to build the social network with the repository with all the features of tagging"). That is, we can't even come close to an answer until we see your proposal. I guess beyond PHP and MySQL the knowledge on some CMS like Jomla would come handy, just to be able to better decide if to use these systems or not. Prior to writing the proposal, get familiar with myExperiment just to get the overall idea of the kind of environment we would like to set up. We would, though, like something even simpler, but with potential to be fully integrated with Orange.

Best wishes,

Re: GSoC

Postby pedromlm » Sat Apr 02, 2011 15:20

Before I write my formal proposal:

I was thinking in creating the social network with some modifications to a CMS so it could be simpler to use. Of the many CMSs that exits, namely Joomla, Dupral and Django, I haven't decided yet witch one to use. My decision is going to be based in the overall complexity of the CMS.
The social network would have the repository for the schemas (with all the features of liking commenting, etc.), scripts/code-snippets submitting (with a source control management tool that could work like Mercurial) and the use of OpenID for the login. The "like" part of the repository could function as a "subscribe" feature, like in the famous web site Youtube, in witch users subscribe to other users so they can fallow video releases easily.
Every data transfer from the repository to the user's PC and vice versa could be done using the users web browser without the need to use any extra applications for file transfer. Of course that an FTP server could be setup so the user could transfer many files at once for his user account and publish theme later.
I would like to ask the mentors if what I am getting at is acceptable, and what else can be done for the project?

Pedro Melgueira.

Re: GSoC

Postby Mitar » Sat Apr 02, 2011 19:58

Be careful, I am not sure if we really want a "social network" portal – a portal which would map social interactions between people. What we want is a portal with social dimensions: sharing, collaborating, following and so on. We are not much interested in relations between our uses.

As this portal will be quite heavily specialized to our needs I doubt customizing existing CMS is the best approach, but maybe is. It depends on how well experienced are you with the given CMS. Maybe better approach is to use some generic web framework (like Django) and build from it, combining existing components together. It could be a better approach. But at the end it all boils down to your experiences with a given platform. Somehow I doubt you will be successful with using any technology you are not already familiar with.

I would not use FTP. There are other ways to deal with multiple file uploads, especially with HTML5.

Return to General Discussions