trust the crowd

[q14]

Whilst the Web has changed with the advent of the Social Web from mostly authoritative towards increasing amounts of user generated content, it is essentially still about linked documents. These documents provide structure and context for the described data and easy their interpretation. In contrast, the upcoming Data Web is about linking data, not documents. Such data sets are not bound to a specific document but can be easily combined and used outside of the original context. With a growth rate of millions of new facts encoded as RDF-triples per month, the Linked Data cloud allows users to answer complex queries spanning multiple sources. Due to the uncoupling of data from its original creation context, semantic interoperability, identity resolution, and ontologies are central methodologies to ensure consistency and meaningful results. Space and time are fundamental ordering relations to structure such data and provide an implicit context for their interpretation. Prominent geo-related Linked Data hubs include Geonames.org as well as the Linked Geo Data project which provides a RDF serialization of Open Street Map. Furthermore, myriad other Linked Data sources contain location-based references. This workshop aims at introducing the GIScience audience to the Linked Data Web and discuss the relation between the upcoming Linked Data infrastructures and existing OGC services-based Spatial Data Infrastructures. The workshop results will directly contribute to the ongoing work of the NeoGeo Semantic Web Vocabularies Group, an online group focused on the construction of a set of lightweight geospatial ontologies for Linked Data. Overall, the workshop should help to better define the data, knowledge representations, reasoning methodologies, and additional tools needed to link locations seamlessly into the Web of Linked Data. Subsequently, with the advent of “Linked Locations” in Linked Data, the gap between the Semantic Web and the Geo Web will begin to narrow.

More information, including important dates, relevant topics, and submission procedures can be found at the workshop homepage: http://stko.psu.edu/lstd2010/

[q14]

I don't want to call it a trend, but the number of websites trying to help in whatever regarding Haiti are soaring. Reminded me a bit on open source, nowadays every company tries to have its own alibi free software project to be accepted in the "community". Maybe this weird first impression was the reason I ignored all these news at slashgeo and the various mailing lists. The Haiti Crisis Map aggregates some of the donated data sources. Very interesting, it gives you a good idea what could actually be possible (and it also illustrates on which low level we operate in our research projects...).

[q14]

we have made our most recent development, the Semantic Annotations Proxy, available as free service. Annotations link from meta-data documents to external, shared vocabularies such as ontologies or gazetteers. The proxy “injects” these references into existing source documents coming from remote Web services without the need to update existing systems. The semantic annotations proxy is available at http://semantic-proxy.appspot.com/

Some its features:

It is fast: We are working directly on the data streams; sophisticated parsing and caching ensures high-performance.

It is robust: Changes to the source document, e.g. due to updates to underlying data models, won’t affect the annotations.

It is reliable: The proxy is running in the cloud. This will ensure scalability and availability.

It is free: This is a free service, and all code developed for the proxy is open source.

It is growing: For now we have support for WSDL Web services and the OGC Web Processing Service. Support for other standards is planned, and we will add new document types on request.

For more information, visit the open source project’s website at http://my-trac.assembla.com/sapience.

[q14]

We are using the DOLCE ontology as framework for most our ontologies, and we use WSML most of the time for writing them down (there are strong reasoners, you are more flexible, and I can do it in Eclipse ;). Well, we always had our own local copy of a WSML Dolce ontology, and I thought its time that we share it. I also created a new project for serving our ontologies (at least temporarily)

[q14]

E-Book Reader

Last week I bought myself a sony e-book reader (the touch edition). Sometimes I just love reading fiction (especially when it's snowing outside like now), but I hate buying these books which are then just wasting space in my shelf. The books in my shelf are supposed to be either references or I read them several times. But many books are just like movies: you rent them, watch them, and forget them. Well, so I bought this reader, and I am thouroughly impressed. Double click on any word, and its definition is displayed. It supports highlighting and basic writing functionality (which is enough for e.g. correcting texts). It's a bit difficult to read complex documents, e.g. PDFs based on Word documents with lots of tables and diagrams. Well, it's not difficult, it's simply not possible. But for everything else, it is wonderful.

[q14]

Sapience & Semantic Annotations in OGC Standards

Our OGC Discussion Paper about Semantic Annotations in Standards is finally online, took us long enough for the final edits. But we are already planning the next version. We discuss how to insert links to external vocabularies or other documentation into existing OGC-compliant metadata (and even data like KML). We realized soon that the theoretical discussion requires some practical implementation. All our implementations, ranging from semantic data integration to semantic validation of service compositions, assume that the Web services or data sets are "already" semantically annotated. We didn't really care where the semantic annotations come from (we of course thought about user interfaces and so on, but not so much about the implications for existing implementations).

Our new (open source) project sapience has been initiated to address this issue. From the website (I am too lazy to write it all over again):

"The Semantic Annotations API (sapience) comprises libraries giving application developers a simple and fast way to extend their applications with semantic functionality. Existing applications with complex and data models usually lack ways to describe the meaning of the data, which unnecessarily impairs the exchange of data across different applications. This is especially the case for Web services serving arbitrary content which have to be integrated into other applications. Due to our background in Geoinformatics, the libraries have a strong focus on supporting the annotation of geospatial content compliant to standards  published by the Open Geospatial Consortium (OGC). The various libraries have been either developed within scientific research projects, or are results of thesis implementations (ranging from BSC to PhD). The implementations have been refactored and simplified for sapience."

[q14]

Modelling animal spreading to track down Osama?

Look here: N. 33.901944° E.70.093746°

When we applied a distance-decay model to his last known location from 2001, the FATA – or Federally Administered Tribal Area – of Kurram had the highest probability of hosting bin Laden (98%). There were 26 city islands within a 20-km radius of his last known location in northwestern Kurram. Parachinar figured as the largest and the fourth-least isolated city. Nightlight imagery also shows that Parachinar is the closest city to his last known location and by far the brightest city by nightlight intensity in Kurram. When we undertook a systematic building search in the city of Parachinar, this approach resulted in three structures that meet all six of them and 16 structures that meet five of them. source: http://web.mit.edu/mitir/2009/online/finding-bin-laden.pdf

Sounds interesting, right? Geography Professor Thomas Gillespie claims that he can pinpoint Obama Bin Laden, using some serious spatial analysis techniques. I just briefly scanned through the paper. He is using distance decay models, and some knowledge about Osama himself (e.g. his has diabetes and needs a dialysis machine, which makes the stories about him hiding in a remote cave not really believable). Proves how important our work really is...

[q14]

Poster FOIS 2008

On Friday the conference "Formal Ontology in Information Systems" is going to start. This is the first time I've prepared a poster about my PhD for a conference. I am usually having talks. I thought it my be good idea to put my thoughts into a poster, since it is not really difficult (for me) to wrap up my ideas into an article. Posters are kind of more challenging (if you take it serious). You have to omit all the difficult stuff and come to the point. You need to have a catchy introduction, but also some more in depth details which keep the viewer interested. Besides that it's simply more fun to play around with Inkscape.

[q14]

Today was deadline for Google's Project 10^100. They collect unique ideas for projects, with the biggest criteria being that it should as many people as possible. Well - "why not just trying it out" -  I thought, and applied for it. Since there was an (optional) video link, and the accepted ideas are getting voted by the community, I installed Camtasia again and produced the following video. What do you think?

[q14]

According to Wikipedia, desire lines (or social trails) are the trails which manifest on the surface if people heading for a certain destination take a shortcut through the grass. This great post further explains the concept of desire lines, and discusses why landscape architects should take this human behavior into account before planning pathways. The map of the Michigan State University shows an example were architects waited until desire lines emerged and paved them afterwards.

Sometimes we GIScientists struggle with the phenomenon of the rise of geospatial applications in the web (i.e. Google Maps, OpenStreetMaps, Flickr, ...). In the past we were told that research is targeting the theoretical foundations of GIScience, and that the stuff we develop is years ahead of the products available on the market. But the web, Virtual Globes, or the abundance of GPS devices changed everything. This all developed so fast that we have a hard time to catch up. In fact, all we can do is to react and try to understand what's going on.

I like the analogy of desire lines and the ongoing social phenomena of the web. In the past we were the architects who decided how people interact with the concept of space, today we need to study the human behaviour in the web and study the emerging desire lines. And once we know what the people care about we can try to pave the ways and provide the theoretical foundations for the geospatial web.

[q14]

My spatial neighborhood, my music neighborhood, my friends, my ...

I just read an interesting article about a computational model which tries to capture the evolutational aspects of trust within neighborhoods [1] and trust between strangers. Despite some flaws (I just don't like the idea of inventing a computation model to simulate human interaction) it raised the interesting question how neighborhood can be defined, and how this concept conflicts overlaps with the often used term social network. We are usually talking about geospatial information, and that's why we have to be careful applying our (old) view to the nowadays prevailing web-based technologies.

In its usual understanding, neighborhood is determined by spatial proximity. You're neighbors are living in the house next door, or are countries having a border with your country. Your real-life social network on the other hand is not necessarily constrained by distance. Most of your friends probably live close-by, but you might have friends or other contacts actually everywhere. Real neighborhood is defined by spatial proximity, real social network by .., well, that's up to you. Real social network is heavenly influenced by spatial proximity, most contacts are also spatially close.

The internet is challenging this view, and interesting examples exist which support this. Last.fm is a website which allows you to upload (scrobble) the music you are at listening to (at the moment). The website then generates, based on your history of music,  suggestions (music you might like) for you. One feature is the creation of a list of people having similar music taste (which makes it possible to scan through the favorites and discover new music). This list of people is you neighborhood, because their interest music is similar to yours. The thematic distance is minimized here.  Last.fm provides also the option to identify friends, which means to create your own social networks. Experience has shown (and this is reflected in my Last.fm contact list as well) that online social networks are usually overlapping with real life social networks (you tell your buddies about this great new platform, they register, they tell their buddies, and so on..). As already said: your social network consists mostly of friends living close-by. Online neighborhoods are defined by thematic proximity, online social networks overlap with real social networks. And real social networks are highly influenced by geospatial proximity.

Now a special case: online platforms which handle information which is, in some way, related to real objects. Example: experience/rating sites like Dooyoo allow for commenting on restaurants. Before rating a restaurant, one should (preferably) have visited it. And one can assume, that mostly people from the real (spatial) neighborhood of the restaurant are among this list. So, the neighborhood of such online platforms overlaps with your real neighborhood, because the theme here is something within the geographic space. People with same interests (the thematic neighborhood) are mostly people coming from the same region (the spatial neighborhood). Inferring, for example, trust from neighborhoods has to have this in mind before applying any computation models to websites.

[1] The Evolution of Trust and Cooperation between Strangers: A Computational Model, by Micheal W. Macy and John Skvoretz

[q14]

Tools for desktop sharing

Ontology Engineering requires deep knowledge about the domain of interest, which means you have to communicate with the domain specialists to put things right and have, at the end, valid conceptualisations. In our current project our project partners are sitting in Paris, which makes face-to-face collaboration a bit cumbersome. And relying on e-mail is definitly not sufficient when we are in doubt. Today we tested out some desktop sharing tools to have a simple (and bit more cost-effective) alternative to traveling to Paris. The concept is fairly simple: Open a conference call, share your desktop and let all (or one of the) others control the application. In our case, the applications in questions where the concept map editor of IHMC, and eclipse (namely the WSMT editor).

We tried Yugma, Dokeos, ShowMyPC, Lotus Unyte, Mikogo, FastViewer, CrossLoop, and WhiteBoard Meeting.

Our first candidate (it was a recommendation) was Dokeos. One requirement was a simple, server-less installation, and real desktop sharing (not just a shared word processor). Dokeos looks like a great platform if you go for e-learning, but for our case it's just overhead. In its features, it lists videoconferencing. But the screenshot just shows a shared whiteboard.

ShowMyPC (and Cross Loop) are nice tools, but work only between two peers. They are perfect if you have to configure your mom's pc, but we looked for an application which is able to support up to ten peers. Anyway, both tools are great and if you ever think about remote administration (or teaching) I suggest to have a look at one of them.

Fast Viewer was the most expensive candidate in our list, and, according to the homepage, it won quite a number of awards. We managed to install the 5-minutes demo, but the quality and responsiveness was, well, below average. Compared to the mostly free alternatives this was quite a surprise (perhaps we just failed to find a proper configuration).

We didn't test Mikogo. You can't read anything about its features without logging in (and giving away your e-mail address). For us that meant we don't care.

Most of our contacts we have to collaborate with use skype, and having the collaboration tools integrated into skype is a neat feature. Well, if you have a current skype version and have a look at the installable extras (in the actions menu), you will see several tools which provide this functionality. We tried Unyte(clumsy, not very responsive) and Yugma.

Yugma is the application which did really impress us. It has an intuitive and enjoyable interface, and it is very easy to open new sessions and let other people access your desktop. Its pricing is reasonable (around 10$ per person per year) and its performance is acceptable (still less than we expected, but it sufficient). Its annotation tool might need some improvement, though.

But for having a collaborative sketch book, WhiteBoard Meeting is a far better option anyway. It is really responsive, nice looking, and intuitive. Definitly worth a look.

[q14]

Finally Moved‽

Moving from office to office kept me busy the last couple of weeks (it started all in the middle of August). We were forced to leave our usual building due to PCB contamination. My working group had its offices in one of the more infected affected parts of the building, so we moved right away into a new building close. But no time to rest, after some more weeks we had to move again, and now we are sitting in the 5th level of the RvE building. The building is neat: we now have air conditioning (useless in Münster), electronic blinds (useless in Münster), a lovely view onto the Weseler Straße (famous for its respirable dust), a fiber glass connection (which didn't work the last two weeks) and it is easy to reach (if you arrive by car, preferably from outside the city).

Btw, the sign in the headline is called the "Interrobang", which I found today on digg.

[q14]

Geostatistics & Evaluation of Ontologies

Ulrich Streit, the founder and oldest member of my department, gave his last lecture of his career today. He was responsible for teaching the geoinformatics basics: introduction into geographic information science and GIS in general (with strong focus on the Arc* family), geostatistics, and numeric modeling. Today's lecture was the last session of the geostatistics class. The content was not relevant for the exams, therefore only a few student showed up. But many coworkers (mostly former students of him) compensated this lack of motivated listeners.

And it was interesting, again (it has been 4 years now since I had to deal with geostatistics). He talked about Zeitreihen (timelines), trends, autocorrelation, and much more. I liked his statements about inconsistencies and inhomogeneities (measurement implies always uncertainty) within the data. He said, (geo)statistical methods can give you only a clue about inconsistencies and inhomigenities.

I dare to say that the same is valid for ontologies (the subject of interest) and the evaluation of them (the statistical methods). Ontologies, especially if the ontology population is based on (semi-) automatic methods, include inconsistencies and inhomogeneities (if we use them over a certain time, which means we have changes of, e.g., context). We can run our evaluation methods only for detecting error, but not to repair them. Here we still need ontology engineers.

[q14]

No Oil. So what?

There’s an oil shock going on. It started April 30, 2007. The world oil supply is falling short of demand - by 1.5% at first, but it’s expected to increase to 3% or more throughout 2007.

I am living in Münster, Germany, which is usually called the city with the highest bike density in Germany. I actually have a car, so have many friends. But its only purpose is to get out of the city once in a while or do some serious shopping. Gasoline was always quite expensive here, and the now occurred oil shock is just making use of car not only an costly, but exclusive hobby. I will continue to use my bike, and expect now that even more of my fellow citizens will do this as well.  I am working at the university as researcher, which includes traveling to international conferences and project meetings. The increasing price of jet fuel does therefore affect me. But flying is very cheap here anyway  and the airlines will try everything to keep their customer base happy. The competition will sort things out, and if it's getting more expensive trains might (hopefully) be an option again. I am, of course, concerned about the global impact of the oil crisis. Upcoming wars about the last drops of oil will affect everyone. I can drive a bike and stop using heating. But most people don't have this choice, and they will react accordingly. And this reaction will change my life, too.

Well, this was interesting to think about. The "World without Oil" - study tries to assemble facts about the possible effects a massive oil shock could have. Another great application of public participation. I took part of this by writing down my thoughts here. What's your opinion?

(via Very Spatial)

[q14]

How Google Earth came into existence.

Mark Aubin, co-founder of Keyhole and now software engineering at Google, speaks about the creation of Google Earth. I like how he describes how the idea of Google Earth emerged, and which sources they use. Sure, after having visited remote sensing classes I know all about satelite and areal imagery. But using kites to collect images was new for me. I found this link via Slashdot, so there, as always on Slashdot, some interesting comments bringing new insights like  this one.

One of the comments did also discuss the "Australia Day Flyover", which was pre-announced by Google (via Google Sightseeing). As reaction, many people took the opportunity to advertise by adding big signs and have, as consequence, a quite cheap and permanent ad in Google Earth. Sounds familiar, huh? It seemed Google reacted fast enough and changed their flight path to avoid map spamming. (Yeah, another new word coined).

[q14]

Emergent Semantics within Social Networks

User in the web are permanently producing information. They either interact with other users, giving us feedback about the grade of the user relations (e.g. trust).  Or they create content, which gives us the opportunity to gather information about the content, for example its relevance or its popularity.  The derived information has been coined "emergent semantics" [1].   It lies within the tags [2] people use to classifiy content like bookmarks, images, or even e-mails. It lies within the ratings people give to the products they bought or to the people they bought from.

Semantic Annotations, the process of enriching ontologies on the application level with information of the domain level, can benefit significantly from the semantics emerging within a web community. A platform which makes it possible for a community to produce geospatial information might use it to constantly improve and extend the domain ontologies, which again can be used to efficiently search content which otherwise dissappeares in the massive amounts of user-generated information.

One of the open questions is, how we can capture the emergent semantics. Semantics is nothing we can measure or automatically extract from log files. It requires users again who are able to anticipate changes of meaning and change the domain ontologies accordingly. This requires a very deep knowledgeof the social network and the platform the network is operating with. Experience is the most important requirement, and experience can only be acquired by using the system. Our approach therefore is the creation of a reputation system, which awards active and loyal users by giving them more responsibilty, and the highest rates user in the reputation system have access to the semantic annotation facilities.

[1]  Karl Aberer, and many many more, Emergent Semantics Principles and Issues. (link)

[2]  Peter Mika, Ontologies are us: A unified model of social networks and semantics (link)