 Ευχαριστώ, Ματίας, και ευχαριστώ, κυρίως, για να εξετήσω αυτή η προσπαθή. Θα μιλήσω για την Βικκή, που είναι ένας εμπανιστικός προσπρόσπρος για να εξετήσω βικικοπαίδια. Ποια είναι η πρόσπρος που ξεκινήσε. Είναι στους τότες, που εξετήσω από την πληθώρα της εμπορυσμής που εξετήσει στην Βικοπαίδια, και να εξετήσει από την πληθώρα της εμπορυσμής. Είναι στους τότες, ότι η Βικοπαίδια είναι σύνοτα για νεοκαμερές. Είμαστε πολύ εμπορυσμές όταν κάποτε θέλεις να εξετήσεις ένα πραγματικό ιδέο στον Βικοπαίδιο. Στον αυτό το προσπρόσπρος, θέλουμε να εξετήσουμε νέες πρόσπρος στην Βικοπαίδια. Και πιο σπιστικά, όπως είπα, κάποτε πρέπει να εξετήσεις ένα πραγματικό ιδέα. Απίσης αν αυτό το Βικοπαίδιο είναι να εξετήσεις όλες οι εμπορυσμές από ένα αυτογραμματικό ιδέα στον ένα σκρομό. Εξετήσεις αυτό ως ένα τρόπο για να εξετήσεις, γιατί όπως θα δω σαν σε τελευταία λεπτά μερικές σκροντές στο Βικοπαίδιο, φύγονται με πάρα εμπορυσμές, εμπορυσμές σε κομμάτι της εμπορυσμής, που είναι ειματζοί, ειματζοί της παραγράφησης, ειματζοί, ειματζοί της παραγράφησης, αυτή είναι η παράγραφη που ειματζοί την αρχαία παράγραφη. Και ειματζοί, θέλω να δώσω κάτι για το μεγάλο κοινό της Βυγυπαίδειας. Είμαι πραγματικός και μεγάλο κοινό. Οπότε, πριν ξεκινήσαμε, τι κάποια από τις κρίντες πηγαίνουν για Βυγυπαίδια. Υπάρχουν άτοιχοι να πω ότι η Βυγυπαίδεια είναι ανοιχημένη και δεν μπορείς να το χρησιμοποιήσεις για τη δημιουργία. Είμαι πραγματικός για την αρχαία παράγραφηση. Υπάρχουν πολλές μεγάλο κοινό της Βυγυπαίδειας, αρχαία παράγραφηση για τη δημιουργία της Βυγυπαίδειας. Και ένας πρέπει να πρέπει να δούμε αυτήν τη δημιουργία και να μην πω. Επίσης, as it's very clear here, I couldn't care less, because we are talking about open source projects and nobody can say to you what you can do or what you can't. So what I mean by visually browsing some Wikipedia page, this is the main GUI of the application. We are making a search for Python programming language here. And what we get is the tiles of the paragraphs, which are these. Here we get the text of every paragraph. So if I click on any of these paragraphs, I get the text. So it's easier to read the paragraph that you are interested and not get a huge of pages with information. What we get here is the links that are found through this article. And most importantly, we get these images. Now, as we get informed here, this page has just one image, which is the first image. This is a screenshot, sorry, it's a small icon of the first image. So what are the other images now? These, the other images are the first images of what in the wiki considers us the most related terms. On our query, which is Python programming. So we have found Ruby, Pearl, this guy there, the creator of Zoop, if I remember well. We get a link to Monty Python here. So these are the first images of all these articles. I'm going to explain to you what will happen in a while if we click any of these buttons. But first let me tell you that if I click next now, I'll get the next tenth of the first images that in the wiki find consider us the most related terms. So I can continue, I can keep on clicking next, I get the next ten icons, which are probably would be something related with Python, Java or whatever. We are going to discuss how in the wiki finds this, how in the wiki scores these articles and considers that Ruby is most important to show than Paris or I don't know whatever link it finds here. So if I click on the first image, which in the wiki considers as a native icon, because it has found this image on the inside, this inside the article that we are looking for, I get the image bigger. So instead of having this small icon, I get the image on its native size, which is this in our case. Now, if I click on any of the other images, I get redirected to the program performs a new query, a new search with this article. So here I have pressed on Zope on this guy and I get redirected to the unperforming a new search with Zope. And I get all the basic data, as I explained before with the titles, the text, links and these images. So the most related items with Zope is Turbogear and all these Python, GNOME, which in most cases are indeed the most related items. So I get a really quick idea of what Zope has to do with or by taking these icons, at least in my opinion. Okay, now the application is written on Python and PyQt4. This power is the GUI. We are still on beta version. The project has three or four months since it was started. It's a small program, as you can see, very few lines of code that Python enables to write a program with a few lines of code. We use this excellent library notebook library to do the Wikipedia parsing, which is the hardest aspect in the application. And we provide executables for Windows users, so they won't have to install Python or PyQt or whatever. Now, these bundles, all the modules needed as PyQt, and it results in 30 megabytes of size, which is a huge size for such an application. But it works, that's what we're interested in. And we don't force Windows users to download Python, install it, blah, blah. So these are the core modules we are using, PyQt4 for the GUI. This library is native Python libraries to do the Wikipedia parsing. You are a lib and a notebook. And we use threading. I didn't explain before that as soon as we perform a query, we get the images on asynchronous mode, that is, we might get the first image, the fourth all asynchronously. We don't expect them to be loaded seriously. A few details about the GUI. It's composed by several classes, functions, and it displays the image on an asynchronous way. And there are many emits. This is for Qt programmers, obviously, this information. Many emits, because of the first rule of Xlib, you don't make any GUI changes. You make them only from the main class. Otherwise, your program will crash all the time. This, we're keeping on with several information about the program. Now, what's important to my opinion, this is a homemade algorithm which definitely can be improved. It's how in the wiki gets the links, how actually it scores the links that it gets. So if we perform a search on a Wikipedia article, we have many useful information we can get from Wikipedia itself. This is the links, obviously. What are the links, what articles are linked from our page? These are the bug links, a very important piece of information. It is which sites, which Wikipedia articles point to our Wikipedia article. Now, if we perform a query for a place, it might easily link to a date or to a place or whatever, or to CIA, I don't know. But if we get this page, the CIA date, whatever, linking back to us, this is a strong relation. We have a bi-directional relation. So we score these links with four. A plus B equals four. Now we also get from the Wikipedia app another very useful piece of information and that is what our search query, what categories it belongs. I named them CD to X. A Wikipedia article might belong to many different categories. Now we are looking, if we find that our article belongs to one category, we are looking what other articles belong to this category and if we find one of them and we score each one of them with one. So we create a big list out of this and we score them. If we have one article that it's linking back to us and it belongs to one category that we belong, then it gets a higher score and the program then looks at the highest scores and it displays these images. That's why when we are performing for Python, for example, we get Ruby and Java and we don't get 2001 or Paris. This definitely can be improved. This is subject to ideas and discussion. Now, this is the newer version that we are planning to apply. Replace all the, make the GUI much friendlier. Replace all the ugly and boring bottoms with icons like this. And instead of showing a clear text, so that display the HTML of the article. Now, this is much better. The only problem I see here is that it distracts the eyes. You are supposed to make a tool that it makes things easier. Now, if we fill it with all these fancy buttons, we might distract the eyes and don't get it very clear. Actually, this, I'm supposed to add this bottom is, this bottom will display, will open the article on a web browser and this will show the translations that our article belongs. If we are looking for Paris, for example, on the English Wikipedia and we press this bottom, we will get the information that Paris is also translated on French or Greek or whatever language. And from a click, you might be, you might get the extra information. Okay, and that's what I mean visually browsing Wikipedia pages. I'm looking for a keyword here, Greek islands, French islands, whatever. And if I click on whatever icon I like out of this and I get transferred to the new article. Several, these information are, whoever downloads the presentation will see all this information about how the GUI is powered. There's no point to, we are lacking, we are lacking the time to explain all this. Briefly, I discuss several of, how several of the bottoms are enabled and when I are powered and when I click for a title, I get the paragraph on the main text browser. So that's the most important thing for me. What can be done for the in the wiki project? One could try to port it to port the application for the web so you don't have to download the GUI. You just visit it through a website. The display of HTML is not as easy as it sounds because there are very few tools to transform, to convert the information from media week markup that we get from Wikipedia to HTML so that we can be displayed. I've contacted several developers on this issue but still haven't found an ideal solution. Overall aesthetics can be improved because this application is supposed to help people and not, so people should have a nice GUI to use. Whatever. Get responses from multiple Wikipedia sites. As I said, find a way to add all this extra information. Okay. The presentation is on the web so anyone interested can see it. Thank you very much for your time.