Quarta-feira, 28 de Setembro de 2011
First Meeting

On the first formal (yet improvised) meeting with my thesis advisors, Professors Luis Pedro and Carlos Santos on September 20th, we discussed some ideas for what I intended to do on my MCMM thesis.

I prepared a low-budget keynote that you can see below (meaning I wrote something on a piece of paper) that explained the path I intended to follow on this early stage.

But first, some context on what we’re already doing at labs.sapo.pt/ua. We are currently working on Sapo Campus Escolas (SCE), a new and improved version of Sapo Campus UA for the general public. I am developing the news stream and notifications layer, which is that piece of the system that allows a user to see what has happened on the platform recently. Much like Twitter’s timeline or Facebook’s newsfeed, each user’s timeline is different depending on who he follows, or what schools he belongs to. If you want to find out more on how we’re doing this, read this post here (portuguese only).

This sets the mood for the first low-budget keynote’s slide, which is a simple model of the system that we have been currently working on:


Slide 1


Basically, it says that we have a wide set of notices (platform contents and events), divided on two big sub-sets: human content (articles, videos and photos generated by the users) and feed contents (imported from external sources). With our current system, we can easily retrieve the sub-set of notices pertaining the user. We can also retrieve a sub-set of this sub-set. For instance, all “published articles pertaining to user X that where published by another user”.

So this is done, and its working, but is also flawed. Not technically, but conceptually. And solely because of SCE’s main purpose which is to bring to portuguese schools a social platform that is engaging to students and teachers, and that improves their productivity and learning levels. Let’s assume I’m a highschool student using SCE. I’m very likely to follow some of my classmates, a couple of my teachers, and even people on other schools. But being a rebel teenager, there are lots of people that I don’t want to follow, specially that unfair teacher that gave me a C when I thought I deserved a B+. But even if I don’t follow him, the SCE platform should notify me whenever he posts any content that concerns my class.

This brings us to our second slide...


Slide 2

...which, basically, states two things:

· There can be no urgency without semantics. Which means that we cannot design a system that filters or ranks information without having a set of semantic rules that can be applied to all pieces of information on that system.

· User semantics complements content semantics. The urgency degree of a content pertaining a user X can only be asserted if we also apply semantic rules between the users. We must extend the follower/non follower relation between users to a more complex relation system, in order to understand the urgency of any given piece of content to any given user. Also, the system would greatly benefit if we extended the post/update/delete content actions to a wider set of actions (asked, warned, replied, assigned, etc).

What all this means is that, ideally, being an highschool student, I should be notified when someone in my class asks a teacher a question on the platform, even if I don’t follow any of them. For this to work, the inter-user relations (which is NOT the same thing as user hierarchy) should somehow be explicitly defined on the platform, allowing it to make decisions on what’s a content’s urgency degree for any user, depending on who published it, with what intent, directed to whom, and so on.

All this makes us quickly identify three different urgency degrees (there are certainly more):

· Something happened publicly - the user is probably not even notified, maybe he’ll see the notice on his school’s timeline or searching the platform, or maybe not.

· Someone I follow did something - the user is notified, and will see the notice on his timeline.

· Someone I don’t follow, but that has a special relation with me (teacher, school’s principal) did something on a context that pertains me - the user is notified, probably on a more explicit way than usual, and will see the notice on his timeline.

 There is a lot of research and analysis to be done before trying to define a model for this behaviour, and I specially have to be really careful to not let this thing get out of hand, because when you start talking about semantics, you jump right to AI, and what could be done in a year becomes a life’s work. And we won’t have any of that, thank you.


P.S: There are also a third and fourth slides, but they don’t really add anything to this post. Check them out if you feel like it.

.mais sobre mim
.pesquisar neste blog
.Maio 2012






.posts recentes

. Topography of Terror

. Elevator Pitch: Proposta ...

. Triplestores

. First Meeting

. Review of Tom Stonier’s I...

. Monografia Aphex Twin

. Processing + JMyron + OSC

. VJing with Processing and...

. Nunchuk Norris

. jMyron e Processing


. Maio 2012

. Novembro 2011

. Outubro 2011

. Setembro 2011

. Maio 2011

. Abril 2011

. Março 2011

. Fevereiro 2011

. Janeiro 2011

. Dezembro 2010

. Outubro 2010


. todas as tags


. participe neste blog

blogs SAPO
.subscrever feeds