 Alright, so I could talk one hour about why annotation is important for what we do, but I'm going to just show you one slide. So if you're a scientist these days, you have problems. There's a lot of information that is published constantly, and it's constantly growing. Now if you talk to some of my colleagues that do actual research in the labs, they might very well be dealing with 150 to 250 papers a week. Now that's obviously not what they do, they don't read all those papers every week, but if you do a little math, that would be 34 hours a week of reading. That means nobody in the world would do research ever. Obviously we are smart, so we organize ourselves in groups, and we find other ways of discovering literature, and so we don't go just out there and read everything that is available. But the question is, once we invest that time and we actually read papers, what happens? I find things that are interesting, things that I want to keep for myself and remember in time, and things I want to share with others. But these days most of scientists are still printing out papers, and so their notes are on paper. Some are starting to use technologies, and they have PDF annotators and some HTML annotators, they use Twitter blogs, emails, and so if at the end of the month I ask you, can you aggregate your own annotation? Can you give me an idea of what you thought in the last month? How hard it is to pull that together? Literally when a scientist is looking at a paper, every word in there has a huge amount of connections. And these connections sometimes are already in our heads, and we want to share them with our students. Some other times we find new things and we want to keep track of them. Annotation is an opportunity here. So how do we keep track of these thoughts, these links? Basically it's a new way of linking that it's not really new because before the web has been already considered to have reached linking. CommU is the application that I'm developing since four years at this point. It's a web application, and it allows you to create manual semi-automatic and automatic annotation. We allow unstructured annotations so you can scribble free notes, but mostly we're interested in structured annotation, and in particular semantic annotation because if you mostly we work in the biomedical field, and if you know that field a little, you know that there's lots of ontologies that have been created in the last few years, and those are helpful for identifying content properly and so that you can search for it and do some little reasoning. That's the website, and you can find short videos that are showing some of the features of the tool. The tool has been created as an entire platform, so we have lots of features in there, searching and creating annotations, sharing annotation APIs, and so on. It has been deployed initially in 2010, and it has been used by several groups. In particular, I want to mention NIF, the Neuroscience Information Framework, that is a big project that is NIH funded for collecting resources in neuroscience, and what happens is that when they have to create a notation, basically they jump in our tools and then they get back the annotation through APIs, and some of that annotation gets published into PubMed to give you an idea. So the idea of DOM is to be open to allow all this communication. And another thing I want to mention is the Utopia tool that has been created in Manchester that allows you to annotate PDFs, and what we can do is we can actually annotate HTML, send it to them, and they can display it on top of the PDF. We don't use, we use simple task force selectors and some rule of thumbs. We don't call them fuzzy selectors or anything like that, but it's proven to work at least on biomedical literature. If you're going to legislation or other things, it becomes a little bit more complicated because there's a lot of redundancy and so on. The version 2 alpha is now under testing. There's actually two groups that are using it as if it was a real release. So I'm kind of stressed out about it, but I would consider it now as beta quality. There are some search features that are to be added still, but that is pretty good at the moment. So those are some of the groups, institutions that funded the project. To give you an idea, it's very simple. There is a button that you probably see at the bottom right of the screen that says Domino when you read in the browser and you find something interesting you want to annotate. You click on the button and the page gets loaded inside our Annotator tool and you can, as I said, create notes on images or text. But most importantly, you can start creating semantic annotation. By that, I mean, for instance, semantic tagging. So you can search live ontologies out there and you can actually tag with those terms that are identified by your eyes and then they are connected to lots of other things. So you can actually use all these networks for search. And this is to give you an idea, this is one single protein out there. So this is the network of one protein. That's all the variants of the protein that are called isoform cleavage products and so on. And the scientists are interested in pointing to one of those bubbles because all Alzheimer's research is about APP. If you just say this page is about APP, you find everything. So it's useless again. So you want to point to one of those and you want to be able to search accordingly. You can always create discussions. So you can discuss either annotations that has been created already or you can discuss the full article or sections of it. You can break down the discussions into thread. Humans don't scale very well. Again, 250 papers a week. We might not want to read all of them and annotate all of them. So what we do is we integrate with text mining tools. You can run external text mining tools. You get back the results in the page. They get displayed and you can provide curation. You just say the term is right, wrong, it's too broad, it's too narrow, things like that. And this is an important tool for text mining providers because they get real feedback from users and they can improve their tools. It's a good tool because then when you want to republish the annotation, you republish the good things. And it's also a good tool because it can be used for ontology mappings. Ontology mapping is very, very expensive. Now, if you run tests with groups and you give one group one ontology and the other group the other ontology, they just annotate and you can kind of guess what the connections in between the two ontologies are. And to give you an idea, these features are going to be used at the biocreative competition for text miners. So all the text miners will make their services available. In a particular format, we just suck in those results and we'll display them in the page and so we can evaluate what people do. We work with pharmaceutical companies and scientists' privacy is important. So some annotation is going to be private. Some annotation will be shared with groups. Some annotation is public. And search and everything else depends on that. So you'll find only what you can actually access. Now, you saw free text annotation and tagging, text mining. There are lots of ways you can extend the platform and to give you an example, this is one of the tools that I developed for the NIF, the Neuroscience Information Framework. They're interested in reaching their databases. Now, when you want to reach a database, you cannot write a comment because that comment has to be broken down in pieces that go into tables and that's complicated. So what we do is we agree on a UI and in a few clicks they create an artifacts that is backed up by ontologies because all those pieces have an ontology behind and basically we can serialize RDF out of that UI in a way that can be consumed and inserted into databases. Another similar plugin is for hypothesis management where you can actually highlight sections of the article and say this is a claim, you can attach evidence searching in PubMed linking to articles, you can attach tags to entities that are involved in that claim, you can attach link to data as well. And this evidence, whether it's data or articles, can be supportive or negative because we have to keep track of that. Again, we work with pharmaceuticals and failed experiments are very, very important to save money. But the good thing about this is that with a few clicks in that UI, basically the user without knowing it is creating a very rich model that translates into a very connected network that then can be connected to those entities with other networks out there and you can do crazy visualization things or you can analyze the network, you can find experts of different domains, you can do lots of different things. And again, the idea is if you keep the platform open you have other smart people that can come in and can do that. Another thing that Domio does, it reacts to things. So if you open, for instance, a PubMed article, automatically you can run extraction pipelines and some of these are simple, like references extraction and those are all structured data. We go into PubMed, we retrieve the metadata and we actually annotate the article with the references, with the structured references that then you can use for other purposes. Other things you can do is, for instance, run text mining on some section like method and materials right away when you open the article. So that way you have a sense of what the article is about. As I said, the tool has been integrated with different platforms. The idea is very simple. When NIF is showing a reference, you have button annotate, you can open the annotator or annotate what you need to annotate and then the data get back into the APIs and some of them go into PubMed. We did an experiment with the Drupal platform and we are going to update that to version two very soon. This realization was on top of annotation and methodology before and now we are obviously moving to open annotation standard. We are working on search right now and some other features, but most importantly, we are trying to expand to other areas and we have already collaborators that are waiting for the tool to be deployed in other settings. The tool is sort of independent from the domain so you can strip out plugins and change ontologies and so on. And you can use it for other domains including gastrophysics and others. And I want to thank my principal investigator, Tim Clark, that is the PI of the project. And this is all. And if you want to see a demo, you can ask me or you can see the videos online and they show a lot of other features that I couldn't talk about. Thank you.