SpeakinAbout

From SpeakinAbout

(Difference between revisions)
Jump to: navigation, search
Current revision (11:32, 19 February 2008) (edit) (undo)
 
Line 24: Line 24:
* Finally, you can give a look at the [http://davide.eynard.it/sa/index.php?title=Special%3AAllpages&from=&namespace=100 Wiki templates] used to link the search string with the search engines which are specific to the particular concept chosen. As they are saved inside a normal wiki, you can modify them and add new features which are more useful for you. Any contribution is welcome!
* Finally, you can give a look at the [http://davide.eynard.it/sa/index.php?title=Special%3AAllpages&from=&namespace=100 Wiki templates] used to link the search string with the search engines which are specific to the particular concept chosen. As they are saved inside a normal wiki, you can modify them and add new features which are more useful for you. Any contribution is welcome!
 +
 +
== Acknowledgements ==
 +
 +
This work couldn't have been possible without the help of Marco Polonio and Matteo Rebulla, two students of Politecnico di Milano who developed the very first Speakin'About prototype for their bachelor thesis. I hope they will allow me to put their work (both the app and the thesis, which is almost ready) here as soon as they graduate... In the meanwhile thanks a lot, guys, you've done a great work!!!

Current revision

SpeakinAbout is a new project aimed at "exploiting user gratification for collaborative semantic annotation". In simple terms, it means I'm researching on ways to make you happy so you'll be glad to collaborate in annotating Web contents with semantic tags. Here's the abstract for a paper I'm going to present at SWUI-CHI2008 about this project:

Semantic annotations could improve the Legacy Web by adding semantics to information which has already been published in form of untructured text. Semi-automatic annotation tools seem the most viable way to obtain a contribution from users without requiring them to have a deep knowledge about semantics, however the effort to make them work is still, most of the times, not worth the reward for using them. This paper presents a collaborative semi-automatic annotation approach for Web pages which requires almost no knowledge about semantics on the user side, but nevertheless provides an immediate advantage for the whole community: annotated data become automatically linked to a whole set of online services and resources specific to their related concepts, thus providing an instant reward for users in the form of additional available information.

Contents

What does it mean?

Oh, well, it's easy as 1-2-3:

1) You'll have a firefox extension that allows you to select a piece of text inside a Web page, open a menu and choose the "Speakin'About" option.

2) Some concepts will be suggested as possible choices of what they're speaking about when referring to that particular text you selected in the page. For instance, if you select "Harry Potter" you'll be able to choose between the book, the movie, the videogame, the fiction character and so on.

3) Once you choose the concept that best fits for you, the selected text will become a link, providing you with lots of new information related to that name, considered as that particular concept. For instance, look at what happens when you search for Harry Potter as a book...

Where is the extension?

Currently, it's still in alpha stage: this means it's installed on my PC and I'm testing it... Actually it works, but I still want to check some details before I release it (heh, who said "The perfect is the enemy of the good"?)

So what can I do now?

  • You can test how the searching part of the tool works by clicking here: you'll be directed to a "manual" meta search engine, where you can enter a search string and specify the matching concept to have better, more related results.
  • You can also try the automagical classification part, which uses Freebase to automatically detect what you are speakin'about (that is, which tries to classify the string using the freebase taxonomy). This is provided as one of the possible plugins for suggesting concepts to the users, and from the very first tests it behaves incredibly well! Click here to try it!
  • Finally, you can give a look at the Wiki templates used to link the search string with the search engines which are specific to the particular concept chosen. As they are saved inside a normal wiki, you can modify them and add new features which are more useful for you. Any contribution is welcome!

Acknowledgements

This work couldn't have been possible without the help of Marco Polonio and Matteo Rebulla, two students of Politecnico di Milano who developed the very first Speakin'About prototype for their bachelor thesis. I hope they will allow me to put their work (both the app and the thesis, which is almost ready) here as soon as they graduate... In the meanwhile thanks a lot, guys, you've done a great work!!!

Personal tools