Monday, August 29, 2011

Summer Progress

Boy have I had a summer!
In between moving into my first apartment and a number of summer classes, things have been exciting around TWC. I've been collaborating with Bo, another undergraduate at the laboratory on presenting an interface for the BCO-DMO project, a representation of the efforts of a number of oceanographic researchers.

Our task was to present remote data, from the BCO-DMO servers, on our internal webpages here at TWC. The data was accessible through a SPARQL frontend set up by our local guru, Patrick West. The first, and by no means insignificant step, was to learn how to utilize the SPARQL query language, an interface for accessing remote semantic data. After some close study of the SPARQL Specification and some fantastic tutelage in the form of Semantic Web for the Working Ontologist, a book co-authored by our own Jim Hendler, we had a set of queries that would fetch the remote data in a form that we could begin to work for presentation.

The next step was formatting it for display. For this we had to deal with the somewhat more complicated beast of XSLT, or Extensible Stylesheet Language Transformations. The first step was to use the basic capabilities to format the data into a simple table, presenting the information a row at a time such as RESEARCHER, CONTACT EMAIL, and DEPLOYMENT LOCATION. Unfortunately due to the fluid nature of semantic data, especially in comprehensive ontologies like BCO-DMO's, the information was not always presented in a consistent manner. Furthermore, in the midst of research I was dragged away (okay, maybe a little willing) to the Californian coast, where I am now writing this post, unable to return home due to the recent hurricane.

Upon my return I am eager to get back to work solving this problem, as the presentation of semantic information is key to any future projects I may want to undertake.

Tuesday, July 5, 2011

Google+ And the Semantic Web

Google Plus and the Semantic Web

Google+, Facebook's newest and strongest competitor has just been released, bringing with it some interesting architectural changes. There are two changes in particular that caught my interest, circles vs. groups/categories and following vs. frending. I believe each of these changes to be both conceptually and technologicaly closer to a more semantic and stronger social network model.

The first of them is Circles. To manage privacy google employs the concept of social circles, overlapping, nested, or disjoint conceptual categories in which we place friends. For instance, when categorizing Alice, Bob, and Charlie, I might be using the groups "Office", "Friends", and "Kayaking Buddies", with Alice in Office, Bob in Office & Friends, and Charlie in Kayaking Buddies. From there you can output messages or share information on your profile with specific groups, such as only sharing your cellphone number with Friends.

While this kind of categorization has been done before it's often been in non-overlapping groups, where someone cannot be in both "Office" and "Friends". The architecture of Google+ instead encourages users to conceptualize contacts as members of a kind of schema of semi-overlapping classes, a definite step towards a semantic understanding.

The second is the follow vs. friend change. In facebook and many other social networks friendship is a binary reflexive property between two individuals (i.e. A friend B & B friend A, or neither are friends). Google+ however employs a different model, in which one follows individuals by putting them in one or more circles. This is a single directional relationship, a more general class than before. Secondly, it is far more conducive to a semantic representation, in which the object of the relationship is instead referred to by URL, an address to their location. With that simple change you could follow individuals on google+, facebook, or even someone who hosts their identity on a local machine.

With the schema-like circle system, and the more semantic friendship model, I think that Google+ could be a step in the long journey of social change necessary for widespread adoption of semantic technology.

Wednesday, May 4, 2011

New Direction

Something I've become more interested in recently is the consumer end use of semantic technology. In particular, personal semantic databases have gotten my attention as general purpose stores for structured information. Looking into this further I had a pretty great conversation with Dominic here at Winslow about work he's anticipating on ontology construction and linking. It's really pushed me to think about what kinds of interfaces and views one would have in a desktop semantic database. Furthermore he directed me towards Li Ding and Jie Bao who he says are doing/have done work in this area, suggesting I do a literature review before undertaking any kind of project on my own.

Wednesday, April 6, 2011

First work session

Having familiarized myself with the basics of XML (the carrier for our rdf data) and XSLT (the language used to translate XML into HTML) I got together with Patrick and Evan to start contributing to the website. The first thing we did was to begin enhancing the Tetherless World ontology, the document defining the hierarchical classes and their properties used everywhere in the site. After hashing out what we planned to do we finished adding in the different kinds of deadlines we had come up with previously. Patrick and Evan are adding content creation forms for events and organizations but I have to head out. Homework: learn vim and make the person form.

Wednesday, March 2, 2011

Starting web work

I've just joined the web team, fiddling with the code for the Tetherless World website and the semantic system it uses to dynamically generate web pages. The short version of it is when you request a page on Rob Greer instead of serving you a static prewritten HTML page the webpage does a little bit of magic. The webpage has a kind of template that it fills with data from our triple store, our semantic information database. It basically asks the database "What's Rob Greer's position title? His contact information, picture of him, etc.?" then uses this information to assemble a webpage in a manner much more aware of the fact that it's using actual information instead of just serving a document some human wrote.