 As we switch from one sprint to the next, I want to note the importance in this whole effort of thinking through how partnerships will work with existing major projects. So HostyTrust was invoked in this round. Paul Courant and his team have been great supporters and very deeply involved in figuring out how we work with and are interoperable with efforts like HostyTrust, I think, is crucial. So thank you guys for helping us see some of that also through that beta sprint. The fifth is called Metadata Interoperability Services, the Mint DPLA beta sprint. Nassos, Drusopoulos, and Stefano Scolias, over to you guys. Thank you so much. Good afternoon. It's our great pleasure to be here in this event. And thank you very much for the invitation. My name is Stefano Scolias. I'm a professor at National Technical University of Athens, the computer science department. We are heavily involved in the development of digital cultural heritage in Europe the last seven years. And the last four years are very deeply involved in Europe. And what we're going to present today is a system we have implemented during these years. And Nassos, Dr. Drusopoulos, who is leading this activity, will present. Maybe in order to define the scope, I would like to say one word, how it fits with what we have heard in this session. Take, for example, the first presentation where you had three big organizations that would like to share their content and presented through DPLA. What was implemented, if I understood very correctly, was that the two of them ingested, gave some small number of data to the third one who used their own metadata representation schema for making the presentation of the public. So think of the whole country. So you have thousands of medium, small libraries, different organizations that represent their data, their content through different schemas. For example, libraries use Dublin Core or museums use EdoxyRM Lindo. EAD is used by archives. They be used standard, is used for the visual archives. And different organizations may use different variations of these schemas to represent their content based on specific usage of their content. So you really need some mappings, respectively of where you have all the data collected in some place. But what you really use is the metadata representation by given by different organizations. And you really need mappings so that interoperability of access is gained through these mappings. These mappings can be at the semantic level, but certainly you need the involvement of the user, of the experts, at least in the beginning of the organization, to implement the mappings. So based on this, our involvement with all this activity within Europeana, we implement a system that Dr. Drosopoulos will be presenting now. Thank you. Hello, my name is Drosopoulos as you already learned. And I'm going to present our Metadata Interoperability Services that is the core of our Mint DPLA beta sprint submission. I would also like to thank the DPLA steering committee for encouraging us to submit something for the DPLA beta sprint. And we are, of course, happy that it can lead to some cooperation between us. So to start, Mint Services constitute a web-based platform that implements and supports aggregation workflows, as we mentioned, mainly in the European region. And typically, these workflows correspond to three main steps. Namely, the first one is to gather providers' metadata. And in this step with Mint, we are gathering metadata in XML serialization. The second step is, as Professor Collias mentioned, to try to connect or align the different providers' metadata with the aggregations data model. And to do that, Mint offers a visual editor for the XSLT language in order for providers to be able to implement and formalize the crosshooks between their metadata records and the aggregation metadata schema. Finally, the third step is to remediate or host the metadata according to the aggregation strategy and scope. This is a screenshot of the Mint mapping editor. I will briefly explain. On the left, you can see an example input schema coming from a provider side. While on the right part is the aggregations target schema. In this specific case, it's a lead of the museum standard that is being used widely now in Europe. While in the middle of the screen, you see the specific subtrick that corresponds to the selected button. And it is actually the mapping area. And in order for a provider to map his data to the target model, he just needs to drag and drop an element. In this case, we have a simple one-to-one mapping. The title of my metadata is mapped to the title of the target schema. This is done by a drag and drop action. And every mapping action like that corresponds, is translated to XSLT in the back. So that you have a formal way to transform your data between the two schemas and a crosswalk that can be maintained and formalized. All the other buttons and flashy things around will help the user to perform more complex mappings, including conditional mappings or structural mappings, string concatenation, concatenation of elements, and so on. So in practice, Mint has been used by several aggregations in the European sector. One of the first and most important ones is the Athena project that gathered over 5 million of museum items from organizations all over Europe, including Russia and Israel. Under the Lido schema that I mentioned earlier, EUScreen that is the audiovisual aggregator that offers a portal with 35,000 videos of Europe's television history. Karare concerning archaeological monuments and sites with a special focus on 3D digital resources for cultural heritage, cross domain aggregations like Judaica Europaana that was aggregating museum, archive, and library content from organizations in Europe, digitization projects, and so on. Also Mint is used outside of aggregators for different scope for supporting the evolution of data models like Lido, prototyping the new Europaana data model, by the European Broadcasting Union for EBU Corp. In order also to publish, since most of those schemas also have the semantic version of this focusing towards the link data publication, one of the first one is the link data pilot by EUScreen. And of course we are releasing the project source code as it finishes as free software, and we are collaborating with other teams in order for it to be reused. Through all these aggregations and projects, we have gathered a growing number of content providers. It's approximately 300 institutions so far and 500 users. And what we are learning together with them and what they help us define and create services for is that aggregation corresponds to tasks for providers and domain experts. It's just a natural next step after annotation and it's a step that is not a one of procedure, it usually is ongoing. And of course aggregators evolution in these past two years has led in the introduction of new tasks and as a result new services for linking and reconciling. To wrap it up since I apparently went too far, I mean, we'll try to help DPLA in order to design, set up and operate appropriate workflows for the delivery of metadata from the diverse institutions that you plan to aggregate. And help providers to interoperate with the DPLA data layer regardless of existing infrastructure in your different institutions or the level of adoption of any DPLA requirements and of course support the natural evolution of data modeling either it is the schema, the metadata schema itself or terminologies being used and so on. So I will not go on about the people, you can find more information not Mint, Image, and TUHG at the URL. You are of course free to contact us and there is also the DPLA beta sprint server that are available where you can use the mapping tool for Lido, EDM, Mark and several other domain startups. Thank you very much. We do have time for a couple of questions for these gentlemen who have come so far. Thank you for traveling such a distance, right in the center. Trying to avoid strikes. We have some Occupy protests for you and other things that you can see here. Yeah, actually I live next to them, it was strange. It was like I never moved. It's another form of interoperability. Yeah, exactly. No lead transfer. John Arnold from Massachusetts, it's just a quick sort of geeky question. I couldn't quite tell from the slides. When you do the mapping from the source to target schema, are you actually doing that mapping at the sort of abstract schema level or do you actually do it by examples and do you think your system could learn by example by having professional catalogers go from one to the other with an example and let it learn what the generic mapping is? Well, it doesn't actually learn since it's really sensitive, at least with the providers that we're used to work with to tell them that anything is happening automatically or it's adapting itself. They want to be really sure that they can validate the end result. So the idea is that usually on the target we have a specific data model, usually a standard in the domain so it's something concrete. While on the input we have provided data that maybe it follows a standard, maybe they think that it follows a standard but they have customized it so it's a standard no more. So usually we work on the instance level. They upload the specific collection that they want to aggregate, they map it and if they use it coherently throughout their organization they can use this stored mapping for subsequent imports. Sir, go ahead. Sir with the microphone, yes sir. Are the mappings themselves published as open data? For example, as RDF triples and can they be reused by others in other applications? Yes, for the time being mappings, of course they are stored and people can download them and load them in other instances and of course they can share it between them. So we had the case, especially with museums that they were sharing the same metadata schema so someone created the master template mapping and they used it as a template to create their own specialized mappings. Now as we see again through the aggregators evolution while back in three, four years ago we had Europeana for example asking give us your metadata records in the format that we are suggesting. Now it seems that people tend in general aggregators tend to ask for the original data together with the formalized mapping and mapping becomes a part of the data that is being transferred between institutions. So yes, currently it's in XSLT form but for schemas and for linked data publication there are subsequent mappings that can be also published as RDF triples. So your question is, the answer to your question is that yes, mapping also becomes part of the contribution of providers towards the community. Wonderful. Thank you both so much for coming this way. Now it says in Spanish.