Day three started a bit later at 0900, as opposed to the not so reasonable 08:30.

First out was Gennady Andrienko, who was going further on with their work on visualization and visual analytics. In short the session was mostly a demonstration of some visual analytics work in a tool developed by his research group. The spatial data in question was spatial movement data, i.e. for instance GPS data and similar. Two demonstration of this were shown. One for traffic situation in Milano and one consisting of GPS data for one persons movement in a car for one whole year. The idea was then to cluster and explore the data in a visual way, allowing the human actor to discover patterns of importance. In essence the idea is fairly good. Clustering together similar data is quite useful to discover patterns, and is very useful for visualizations. For instance to find, and predict what the traffic situation is/will be on certain road segments on certain times of the day. The tool used is developed through research, and is thus only available partly (I haven’t tested it).

After the fairly quick session it was time for lunch break. As far as I remember a regular buffet of some sort.

After lunch it was time for my supervisor, Terje Midtbø‘s, session on web usability and web based research. Midtbø started off with a short introduction on the visual variables then on the dynamic variables, worryingly, very few of the participants had heard about the visual variables, and even fewer had heard about the dynamic variables. The examples shown for illustrating the different variables, I thought, were not that good. Some more practical, and real life illustrations could have been used, perhaps with more success.

Then the session went over to focus on map animations. A few animations  were shown, mostly old animations from earlier research projects. Hence not that “audience capturing”.

After the demonstrations a short tutorial on how to make map animations for use in research were given. In essence it was a very quick tutorial on the use of flash/PHP/MySQL. Probably to short to be of use for participants not already familiar with the tools.

In relation to the session’s title one part of the session was dedicated to methodology in web based research. Here Midtbø gave quite a few interesting tips / guidelines for elements to consider and pay extra attention to while designing and performing a web based experiment. This was probably the best part of the session and could successfully have filled the whole session (from my point of view).

After the two sessions it was time for a buffet (yet again) and an informal gathering preceding the ICA workshop which started next day.

The second day of the summer school started with a session by Gennady Andrienko with the topic of visual analytics and visual data mining. The topic is thus right up my alley. Andrienko mostly demonstrated their own developed visualization software called Commongis. The main feature of Commongis is that it supports several integrated, linked views i.e. several visualizations, both statistical and geographic. Additionally the views are all interactive, which means the visualization can be clicked, selected and dragged.  The tool is fairly impressive, especially the integrated interactive part. It is (or seems to be) impressively fast, as it offers clustering (k-means at least) and other computations on-the-fly.

Andrienko also presented their “invention” of the bi-directional colour scale. Which is very good at visualizing comparisons from one reference value relative to the rest of the data. The bi-directional scale used was mostly beige/brown (high value) and blue (low value). With the reference value being white. This should definitely be used more! An interesting application could be to visualize a conceptual relationship between concepts.

The session consisted of several examples of different visualization techniques. Some of them; Parallell coordinates, time-graph series (?), and bar charts/histograms on top of clustered map. The latter I found to be, well, to cluttered. However, for some applications it could be very useful.

I found the session to be spawning my creative thoughts, and thus, quite good. However, this was mostly due to the screenshots/demos, and not necessarily the “reading of slides”.

Then it was time for lunch. Kjöttbullar with Lingon, which actually was meat loaf with Lingon (cranberry jam). The Lingon was regarded as a weird thing, which it in some sense is.

After this, Jean-Claude Thill had arrived and was going to take his scheduled session.

Mr. Thill is truly a speaker. Not respecting the schedule, at all. However, to some degree this didn’t matter that much. The session was largely on Self Organizing Maps (SOM) which Mr. Thill truly enjoys. And it seems like there are a lot of GISc persons who enjoy SOMs, however, I do not. It is probably useful, however, I cannot see the real use or application for it. Probably due to my interest for interactivity, on-the-fly things and highly intuitive visualizations. Which I haven’t seen SOMs do – yet.

In the end Thill presented an idea of absolute vs. relative space. I.e. space is not inherently geographical, or spatial – it could as well be functional or social. I actually found the idea fairly similar to what I’m thinking on visualization and space as such. Thill did not go very far into this at this session, but was going to have a session later this week on the same subject.

Then it was time for dinner, which consisted of a buffet and was ok.

Dinner was followed by a new session with Mr. Thill on Paper writing and publishing. Which was suitable since Thill is the editor of a journal. The session consisted of guidelines and tips for writing and how to get published in international peer-reviewed scientific journals. Which was useful. However to a large degree common sense does fit well also. Interestingly Thill suggests to write with clarity and keep it at a fair length, i.e. not to write something just to get the paper longer. All in all a good guideline. However, he did not apply the guideline for his presentation, which lasted from about 1800 to 21:30, which is a bit long..

With that, day 2 was finished.

I’m currently attending the NordGIScience summer school and thought it could be of value to add some posts related to it.

I arrived at Gävle late sunday evening and found the hotell fairly easy.  The hotell is ok, but nothing more.

Monday started off with registration and some opening notes with an emphasis on the networking opportunities that the summer school offered.  Which is nice, in my oppinion. The original plan was that Jean-Claude Thill was going to hold a lecture , unfortunately Mr. Thill was stuck in Atlanta airport due to bad weather. This resulted in Bin Jiang stepping in and taking the first part of the lecture.  The lecture consisted of a (very) brief introduction to Knowledge Discovery, (Spatial) Data Mining, Self Organising Maps (which later should be of huge importance) and the importance of the notion that “Spatial is Special”(!). Mr. Jiang is an energetic person, which held a lively presentation.

Following was lunch break in the cantina. The lunch consisted of a pasta salad with chicken,  green salad and tzatziki. Which was quite amusing for the Italian guy:)

Then it was time for the afternoon session which Vasily Popovich had responsibility for.

This session consisted of a theoretical lecture followed by a practical session using the Intelligent GIS tool developed by the oogis.ru team. In short the intelligent GIS idea consists of integrating techniques from the field of artifical intelligence (AI) with GIS techniques, mainly computational techniques.

The tool develop implemented this by the combination of using an Ontology, Rule base, Rule engine, traditional relational database and a map visualization tool. In short the tool is an extension to the ontology/rule engine Protégé. I’ve been using Protégé before, and find the principle idea of ontologies fascinating and potential very useful. However, I do not see the potential of success for ontologies with the current human interface for them. The current interfaces I’ve been looking at (such as Protégé) have issues, especially regarding usability. It is almost impossible to create and maintain a large, stable ontology without dedicating a large portion of your life to it. Unfortunately this was the case for the intelligent GIS as well. Although the ontology was created for us, and we only flickered with some instances and some rules associated with them. I strongly believe this needs better visual interactive interfaces. Probably the most needed feature is, in my opinion, to be able to easily get a quick, and correct, overview of the complete ontology, knowledge base and rules – in one view. Similar to what we find in conceptual modelling for large enterprise systems. However Intelligent GIS is  a good idea, which I definately see the potential and use of, however, if humans are supposed to use it effectively, then it needs to satisfy the humans requirements for usability and interaction.

The first day of the summer school ended with dinner at a local restaurant (greek or persian).