 So, hello everyone. I am glad to be here for our office hour Actually, this is a part of the hackathon. You have seen this morning the tutorial done by Vigneron about Sparkle At Sparkle actually Kind of Shows us how we can not only actually add knowledge for For to wikimedia projects, but also use it actually For variety of purposes one of these purposes actually is to create web tools for the social good Yeah, so just a brief disclosure here is that this is an office hour of wikimedia funded project Research projects about using wikidata to support clinical practice Yeah, so So they do to explain what we are intended to do so we are mainly making an overview We are showing you how actually we use wikidata to extract knowledge Transform them and load them in web services using a variety of methods Well, we have actually seven tools that we like to show you how to work with Yeah, the first one as we are on said is the wikidata query service And the wikidata query service is wikimedia implementation of the Sparkle server based on place graph engine So what you can do is that you can formulate the Sparkle query to ask about what you need and actually run it and Actually ask for the code to implement that in your application And there are plenty of formats there Yeah We have plenty of formats there where we were Like URL like HTML like PHP Javascript, etc. And this will allow you to embed the results of the query in in your source code whatever it's Actually source language is So for example, if you are developing a tool in Java So you you copy the code the pay you paste it and When you run again your code actually the result will not be the same as this is Actually refreshed in real time So what you will have here actually is the result as you run the query And that's the advantage because you you can actually Do your application with the updated status of knowledge in wikidata The second thing is wikibase integrator And wikibase integrator is a very simple Python library to use actually to kind of Get information from wikidata and add information from wikidata It has been developed by limits a few years ago And actually writing it is quite intuitive by contrast to pi with wikibot That is kind of very hard to have to handle and to manipulate that there are many tokens Actually the process actually with this it is very simple you provide the source You provide the login the password it would generate the authorize for you from wikidata wiki and Then you can just write what you will add or what you will get in the form of triples and get and Add the information that you like The third thing is wikidata hub and wikidata hub is an API Actually That allows you to identify the wikidata item Corresponded to an external identifier as you know There are many external identifiers that are represented for every item in wikidata for example bibliotech nasunat de france like wordcat uclc like Mesh medical subject hidden UMLS there are plenty of resources that are aligned to wikidata items and The thing is that you can actually enter the external identifier and he will find that That that your that API will find for you the right wikidata item corresponded to that So you can you cannot actually Very easily actually Because it is in G zone you can process it using requests That we will explain a bit later. Yeah Another thing is the wikimedia tool for us and the wikimedia tool for us Will allow you to host actually if you have developed an application probably an application based on HTML and JavaScript and CSS web application So you can actually access to tool for us And actually get a token and Host your work there and so it will be deployed automatically on the wikimedia cloud directly So you will define your repository You will share it the SSH key and then you will you will write just some shell scripts and and that's all and there are some for example if if your Website is dynamic using PHP or using flask for example to actually kind of login Do some edits etc. So there are some Python and PHP servers that Actually can be used for implementing websites Yeah concern in flask. Well flask is a white row web framework written in Python and It it is mainly done to actually Gerate APIs, but also to define dynamic web pages So you can use the get or the post method actually to To get arguments from the URL and then using these arguments you can provide surfaces in an automated way So there are some pages you can define In a way on another you can define some static static web pages inside it as well because Not all the pages in dynamic page Dynamic website are are dynamic. There are some static pages. You can define the dynamic pages You can do some code in in Python actually to To process the arguments and then you can run the web tool In in two lines of code. So as you see here Yeah One important thing is the media or key search API Also known as elastic search Well, they this is this is mainly functions as the get requests to perform an advanced search For wiki pages by title or content text match So for example here is JavaScript code for using that API You define the language of the your search And then you define that you I will be using G zone and then using that G zone There is a JavaScript code to convert it into the table actually and and you when you click on the result You will search for it or something And actually this works for wiki data So for example, we have the API for wiki data for for the search and actually we use it Actually to create application to create applications when we need actually To search for for elements for the key data items and then do some something with them Yeah, actually these were these can work use using wiki base at the greater as well So if you would like to do the same with Python you there is a Function search entities that allows you to do the same Automatically without having to do fancy code Yeah, then there is a request and request actually allows you to actually manipulate APIs particularly in G zone and You you that allows you to get to get the The text the the G zone dump of the page that you are requesting for and actually processing it Using Python Yeah, so that's all what we need to show you as The the tools that you can use and we have used as well So now we will be showing you how we we generated some web applications useful a web applications Actually using these methods and nothing but these methods and we are I'm not a computer scientist The MN is not to and we have made successful apps so Actually, I will leave the floor to Yemen to introduce the first two Hello, thank you very much for being here. My name is Yemen. I'm from Tunisia. I'm from the wiki world editor user group And I will present today this application called Southpedia Which allow us to scan QR code and listen to the Wikipedia Articles in the fall in the language of the phone. So what is Southpedia? So it's a tool that enable Enables the generation of a QR code for any for for a monument for example or any building or Museum collection for example when you scan it The the system will detect what is the phone language and will will will trigger Listening to the Wikipedia article that correspond this language on from Wikipedia So first of all, you scan the QR code the system detect what is the phone the language of the phone Go to wiki data from wiki data. It will look for a property called spoken data and Spoken wikipedia file then go to comments to retrieve either the Recorded audio file or to generate text to speech so This is the idea so it's based on wiki data wiki data Then we'll go to either two comments to retrieve the audio file or to generate the user API text to speech to generate Automatic lecture audio of the Wikipedia article Wikipedia page We can show quickly how it's working so basically it can be accessible when south South in Arabic or in Swahili means boy audio. So the URL is southpedia dot wiki It's hosted on tool for the tool that was I'm just presented and For example, if I want to generate we have this monument for Tunisia, which is we used for To generate the QR code. I think there is some photos here. Yeah, so this is some This is for example monument in Tunisia Listed monument. So we have this we printed the QR code region to care code It's it has been printed and posted on and the enters and as you can see here using wiki data We have these properties spoken text audio We have a recording in Arabic and the recording in French and in English. There is no recording So the tool will generate the text to speech for example We'll try here. So this is the Wikipedia article. All I have to do is to copy the URL and go to the tool and This is very similar to the carepedia. If you know queer pedia, this is inspired from it So it will generate the square code. Then you can download this and print it So if I scan the square code, my phone is in English So if I scan it the tool will will go to wiki data But there is no recording nor recorded the audio file on common. So it would 859 Corresponding to 245 a H by the aglabid Prince Abu Ibrahim Ahmad Ibn Muhammad So this is because the phone was in English. So the So when it's going to clear code, it will go to wiki data So here this is the wiki data item of this monument. So here we have these properties spoken text audio So basically it will look for English. There is no English. There's Arabic French So it will trigger the text to speech But if it's in if if my phone is in French With the French As you can see so based on the phone of the language of the phone we can listen to different version Either in English or in French depending on on the on the phone This can be very useful for example to be used in For a museum collection for example a state use mosaics and everything So the museum can put these QR codes and the visitors can use it as an audio Alternative audio guide for free for example or even within the city for the monuments So you don't have to put several QR codes by language only one QR code and the tool will will detect What is the language and automatically generate? the the audio file So currently this has been is being used already in Tunisia But we still have a few things to fix So the remaining to do here So we have a problem with iPhones ipad iOS I mean to read ogg files because the audio files are hosted on Wikimedia Commons on ogg format and currently what if I try to to scan with iPhone and there is to read the audio file from Comments there is an issue and the phone will try to download the file instead of reading it So this is an issue that need to be fixed. It's not fixed yet Also, we would like also to to adjust the API in order to record some statistic For example, how many scan The code has been scanned how many language to detect what I mean the visitors for example for a monument or museum from which country The majority are from so this is we can do some some statistic around that Also, there is a this is can be also added to detect the local languages dialects for example In a country like India most of people will use in a phone in English But English is not their language. So they speak or do or time or or other local dialects So instead of listening to the audio in English the the tool Can propose for example to say there is a local Audio in for example or do would like to listen to it instead of English so it's always better to Listen to the Wikipedia article or to the audio in the local dialect or the local languages So this is can be added and the last thing also Currently if there's no of course Wikipedia article, we cannot generate anything. So it's based because it's based on Wikipedia article but for example if I scan With the English phone and there is no English Wikipedia article We can use something like a resonator to generate brief introduction using wiki data using for example So let's try the resonator for example for this So as well if I don't have an article for example in English So for example the tool we can add this feature to use resonator to something like resonator So currently just small sentence for the Tunisia, but of course this can be improved by adding more data to wiki data to to generate sentence and read it automatically So this is the pending action to fix so this is This is an issue to fix. This is some improvement to the tool that you would like to do So if there's anyone who would like to help on this tool, it's more than welcome So this is what's out here The next tool who Sam will present it is called the medicine. Yeah, so Thank you. I'm so medicine actually Well, I'll know that clinical decisions support systems in the markets are mainly in multilingual and mainly paywalls, etc so the the fact that we keep that wiki data includes a Huge amount of medical information in the multilingual setting in a variety of formats Actually, it is a huge waste not to use that for clinical decisions support and so We actually developed this tool that is mainly based on wiki data query service That actually get a list of items for example a list of drag Actually and create the wiki data spark key data query service Actually to find for example drag interactions For example, if the the the initial items are are are diseases, for example a They it can generate complications if they are symptoms they it can generate probable diagnosis, etc So to show you how it works actually I made a brief demo So let's see it together to see so as you see it is multilingual So just at the first site So this is mainly the search based on elastic search So we can actually add as much drags as as we can and then we choose We have a list about what we need to do with this all these elements Yeah, so which we are fine with track interactions. There are six actually wait is to actually process these elements actually we we run the The tool Sparkle query is generated and actually it returned the results That's all So Actually, as I said earlier for softpedia There is the the QR code for the the URL of the tool and the source code Actually, the tool is deployed on as wiki github page And the source code is actually Hostets on github so what remains to do actually we are supporting five languages now So we need to add more languages that we need naked native speakers to add more and more languages And actually we need some medical experts to add new features because we believe that there are more Things that we can actually generate using medical information in wiki data Yeah, so I will leave the far again for Yemen actually so now we are just going to present to True new idea that you would like to implement and if there is anyone also keen to help The first idea is called wiki festival calendar So this idea also is to use wiki data to generate a calendar of festival around the world for example listed as UNESCO cultural heritage, so The idea is come when when we travel to a country would like to see if there is a festival during that period So we attend this festival to discover the culture So the idea is to use wiki data to generate the calendar Of the festivals by country by period so we can plan a trip based on this information So we can use wiki data for that so instance of festival for example And we have also the date that the occurrence of the festival and generate this calendar Which can be very useful for for people to visit the countries So this is the first idea to that we would like to implement to do the second one the second the second idea is It's called wiki data books also it we would like to use wiki data to generate books online For example, this is a very example use case mean steps for example the first step For example, you choose a cover photo of the book for example from comments step number two Define the structure of the book or the e-book Then for for chapter one for example you input sparkle query the sparkle query will retrieve a list of wikipedia article We'll compile them and we'll compose the first chapter then next second chapter a query from my wiki data that return a list of articles and Create the the book and we can also generate some database also include the database in the book for example ma ma time lines and everything a Practical example of use of these for example, for example, I would like to generate any book for example of the Medina of turns Which is a UNESCO World Heritage So for example the user will choose beautiful photo from the Medina from comments Then next step chapter one for example the city gates the old gates of the city wiki data I can retrieve all the articles wiki pd articles that about the gates so that then I will have the first chapter Then second the second chapter the city palaces for example Same thing instance of palace located in the part of the Medina of turns from wiki data The query sparkle then we have a list of article then I have the second chapter and Go on and then I will generate the the book and we can tell you that download this book as PDF They can be used as educational resource for people to learn about about the history of the Medina this story of the city so it can be Good alternative for education resources for people who doesn't have the hardware to access wiki pd or Or all the educational resources so we can print have these books printed and for any subject for example can be museums history Species so and of course all of this using wiki data Thank you. I'm so as a conclusion Actually, there are plenty of tools as I explained earlier that can be used to build wiki database web surfaces So these tools are easy to learn and we should all disseminate these tools and work with the with them And so the most important actually is to have the idea the idea Is the thing that mostly count actually the tool can be can be learned by everyone But the idea of having use case of wiki data is actually very difficult to to envision and that's why we think that everyone should should actually brainstorm about using wiki data and actually think that they can actually Build their own web applications using this database Thank you. So I think this is the right challenge. I mean, this is the challenge for us for our for the community I mean, it's it's good to have content on wikipedia and comments and everything It's better to how to know how to use this content and this is the real challenge I mean content the creation of content can be very easy very easy in the next future one with one button I can translate all the wikipedia for example in English in any other language create new pages with with the tools Like chat GBT generate photos but with major needs so the creation of content will be more and more easy in the future But the real challenge is to know how to use this content. I mean find the right applications the right Use cases to use this content so it can be useful for people So this is the real challenge and we have as Josem said we need to think To brainstorm how to find new new use cases to use the content on the wiki projects Thank you very much So if you have any questions feel free actually these are my contacts you can reach out to me at every moment Actually, if you are you need further information about the tools or the applications Thank you very much. Thank you. If you have any questions, please Thank you very much. It was super interesting and I have one question and When you talked about the different tools you mentioned for example wiki-based integrator You you mentioned also quite briefly a pie wiki bot and also the the the the API the bear API and I wonder if When and where should we use because they they have a lot of Characteristics in in common and and I I always wonder which to use because I am for example I am starting to be it to build a tool with pie wiki bot But I wonder for example the if the speed is different in between Those different tools if using the the bear API is faster for example, then then using and some Python modules, I don't know In your experience what could you tell us? Yeah, so actually all of them are based on Particularly The media wiki API or the sparkle query surface. So that that's for sure. So they will run at the same rate in in some way however If you you would like actually not dated wiki data You you would like to only do get information. Actually you can go through for example the wiki data have API And that's all Concerning the wiki base integrator. It is mainly useful if you are you will be interacting with wiki data This means that you will get it you will get information process it and Go back to add it to wiki data again Or for example getting some information from other resources process it and then Get it to wiki data. So that that's it Concerning wiki bot okay by wiki bot actually if if the the user actually has experience in building wikipedia bots and It is actually he is actually familiar with actually doing some some requests and and some Edit automated edits using py wiki bots. So why he should learn something new He can adapt himself in some way But for the newbies Actually, this is the problem Because they don't they don't know how how to use py wiki bot It is difficult to learn There are few tutorials There are many some fancy like beginner tutorials not Going straight to the points Actually the technical community Is actually very hard to reach considering the py wiki bot question. So here wiki base integrator will play It's a it's real real function Thank you Thank you very much for your presentation I would like to know all that you have developed. How do you think to spread this in the community? Because I think the tool that you develop is not for you alone It's for it to be used Do you have a plan to spread this With the users? Thank you Well, of course We have a little bit to answer Very good So To spread We will answer the questions before in English and then we will answer So the question was I mean these tools are are very good, but how you are willing to communicate about these tools so the community can know about them and use them Yeah, so the answer here is that Actually we participated to to many wikimedia conferences actually to share this with the wikimedia community Actually, Yemen was a wikimdabab We went to Wiccon Convince on Francophone wikimania last year wikimania this year And actually we are showing up actually to people to actually Show them what we did And if they have some concerns about our tools and how we develop them As well We are actually Reaching to the to to the public in general Actually, we are showing this to them for example people from the civil society mainly on social networks And actually see we have we have developed this tool And let's look what's behind that. Actually, there is a project a very interesting project called wikidata That is working behind this And people will will be surprised and actually is that open? Yeah, it is open. We can can we contribute it? Yes So they will be interested to contribute. Actually, there there is a more rich For for wikidata and the wikimedia community in general. So that's the main thing if Yemen cannot something So first yes, first of all, I just to confirm that currently we are mainly Communicating through this is during the conference that we are participating And I I should confess also that it's not easy to find these tools when you there's no Like central repository where you can browse all the available tools so you can know about them So maybe this is something that we can work with the foundation to create like a place where you can find this Common application that you can use and this is the different use cases So so for example you you can go to this place and find these tools and maybe search for one that you can use in your country So I confess this is It's not really clear. I mean that maybe we need to work on it with the foundation to to centralize all these tools in one place otherwise everyone will build new tool And no one will know about it, which is not so not good. The second thing also these tools are not only used for the I mean for the community also they I mean for civil society like us and we can go to the civil society NGOs to tell them about these tools so they can use them by themselves so They're not meant also to use only by the community. So it can be used by anyone So we need to communicate more and more about these tools around Like for example, if if it's about monuments out we can go to Cultural NGO to tell them about these tools and everything So you mentioned working with the foundation to document tools like these out of curiosity like Do you need the foundation to do like is there a reason we can't just create a wiki page on wiki data and start documenting them there? I'm just curious what the barrier that I'm asking because I'm at the found I work at the foundation So if there's like something we can help with Great, I'm very interested in what that would be But I'm also just wondering if it's like that may actually not be necessary Like we can probably just start building wiki page with those resources, but maybe there's something I'm missing So that's why I'm asking So basically all these tools they should have their page on We have a page on meta for example for these tools to document them But these page also are I mean, you know, you need to know how to go there But there's no just one place like a single point of of entry like to say this is tool for glam This is tool for medicines This is tool for that you can browse by subject and find the list of the tool that you can use And this is the use cases that can be used for example So for for example our friend from toku. He would like to use something about monuments He will go to this place and see glam tools and Yes, something like that Yeah, just the link will be yeah, it will be like Yeah, it can be very useful to do something for external audience not for the community For example an NGO in cultural heritage. He would like to see what are tools available for them to be used Thank you Thank you I am just going to start with this It's based on wiki data It's to know if you do a good job of cleaning To give it to people through wiki data Many people contribute And it's not really proper For example, people can use properties that are not necessarily the right ones To characterize an element Or to give the element of what it really is So do you do a good job of cleaning the data before launching your requests? I will translate the question So what he is saying that all these tools are based on wiki data And all we know that wiki data is a lot of information and everyone can edit So we can have at the end some messy data So he is saying what we can do to make sure that the data are correct I mean have a wrong result as you know garbage in garbage out So this is the question for him And what we are doing to do to make sure that the data are correct Yeah, so actually there are several ways actually to do that But actually we have not used in this in this particular use case Because actually the data is too specific But what we have done is that research for statements have references So that's the basic thing that we do that But beyond that actually there are several other methods Actually based on semantic alignments between wiki data and external resources that are Actually curated and maintained by experts Actually to actually verify that the information in wiki data is correct So we actually get in the items that are corresponded to the wiki data item in external resources And we are just comparing And if it matches it is a good information If it does not match that's rubbish So that's how we do that Yeah Right, it's a great conversation I have another question We have been talking about retrieving data from wiki data But what about facilitating the edits to uploading information to wiki data By people that haven't got skills on wiki data Because we all know that wiki data is not so easy to edit And I wonder if what the community thinks about building tools that facilitate Seamless additions to wiki data Is it something that the community approves Or is it something that Because I think there could be concerns about vandalism and so on So that's why I ask Yeah, actually For this project actually For the project that was funded by the wiki media foundation Actually we are leveraging mesh keywords of public publications Actually to generate a list of relations In the form of triples alongside the references And we are envisioning to build a tool actually that Like wiki data came That gives the statement Actually the ready statement And they ask the user if it is true or false And if it is true at wiki data click it is at If it is false click on discard It is discarded from the initial listing on the tool But actually there are plenty of challenges here The first challenge is that We need to the token of the user The authorized token And so Will Be Would that be actually Kind of accepted by the community actually to give Their token Although they do that for many other tools like quick statements like But actually we are developing these tools now And we are minor community members So we will give us the same privilege as high standard developers The second thing is that If someone gets into there And it is auto confirmed But it is not an expert Actually the thing Looks very credible The formulation is look very credible So we can add that to wiki data Without knowing anything And actually it will pass there And it will not be visualized And actually it is the wrong information So how we should give that right to the right person To validate the thing The third thing actually And the last thing Is that There are some other ideas about actually Instead of Like the query builder Instead of letting the user do all the stuff Actually formulate and get stuck with things For example we can tell people You can choose from a list of properties For example And when he clicks on the property We can actually Refine the elastic search To include the elements that correspond to the Subject and the elements that correspond to the object So that this will not cause any inconsistency So that's a way of thinking But actually this is expensive From a computation point of view And that's what makes it Very hard actually to implement in a real world There were a few initiatives in the past Mainly around games So for example to add some What is the gender Male or female For example instance of When you answer the question It will edit wiki data and add the information There were also some initiatives around structured data So you have for example Painting and the game So this painting depicts for example A man or a tree or something like that And of course will update wiki data So currently this is the few initiatives that I recall It's a bit around like gaming But it will update wiki data But I would like also to show you something Not about editing wiki data But about retrieving data in an easy way Because as you know also Not everyone can master Sparkle or the wiki data query So in our user group wiki word heritage This is our website So we try to provide some interfaces For visitors to interact with data Without manipulate Sparkle So for example here there is a project about World heritage cities So for example I would like to generate a map So here basically We have a map of all the cities around the world But if I would like to go inside the city For example I chose Medina of Trenus Then without putting any Sparkle query automatically By selecting the city we will have a zoom into the map On the city and the different monuments of it If I would like to generate a timeline of a city For example I would choose for example a city in Morocco And wiki data will generate a timeline automatically Without manipulating Sparkle So we can think about for example This is a graph about for example people In the city of Kasbah of Algeria So what are the people that are related to these old cities For example and we can navigate through this data Without manipulating Sparkle So also we can think about different use case How to display data without manipulating Sparkle Using this kind of interfaces where the user can select a subject Or a topic and then send this parameter Of this for example here when we select the city The website will retrieve the Q number of the city And will build automatically a query that will retrieve the data So the query is behind the scene And the end user will not have to know about the query Yeah that's very simple actually to implement So it is basically HTML with JavaScript That's all it works And this can be adapted to any subject This is a word heritage but can be also doing to Education to for example wiki women For gender gap for everything So we can think about queries and just lay with the parameters Any more questions? Not then Thank you very much Thank you Bye bye Thank you