Cetis OER Gathering
This week i attened the Cetis OER gathering. When i arrived in Glasgow I was not really sure what to expect. It didnt seem as though and clear goals for the event had been outlined. As it turned out the event was quite free form. We began by everyone introducing each other and stating their interests. This was then broken down into “birds of a feather” session to discuss peoples shared intersts in more detail. There were about 25 people in attendence and there were broadly of a technical nature. In the fray of the initial division of interests Pat Lockley (Nottingham) and I grabbed Jenny Gray (Open University). The subject of our discussion was RSS feeds for teaching resources. Our party was about 15 strong in total with a slice of the other attendees pitching in their views or just listening in on the debate.
We used the oppertunity to discuss the various uses of RSS for publishing Open Educational Resources. I have had some problems of this nature in the past when depositing the humbox.ac.uk resources in JorumOpen. Jenny has some strong views on the matter as the creator of OpenLearn and outlined a convention for publishing OERs on in RSS which involved reinforcing the native (loosely defined) RSS tags with their counter part Dublin Core tags. This has worked well for OpenLearn in the past an I will be updating all the oneshare repositories with a feed that confirms to her convention. This keeps us compatible with Jorum and Xpert and will make out resourses more accessible.
The discussion then moved on to whether RSS was a suitable format for large scale injest of OERs. Broadly speaking while we concluded that RSS was suitable for lightweight and small scale transfers we were deffinately using the standard outside of its comfort zone. For the kind of lightweight learning resources we have in EdShare the RSS tags map fairly logically. Other things it would be nice to be able to say were “what resource was this resource based on” and “what resources are based on this”. Also it would be nice to be able to treat a person as a URI rather than a plain text string. The way we think about users in EdShare they are just as important if not more so than the resources they produce. A user is more than just a name. Just as a teacher is more than just teaching resources. Anyway to cut a long story short We decided a linked data format would be a much better way to tranfer data of this nature.
Other topics which were touched on if not fully explored were:
How are people currently finding learning resourses? – we have /some/ data about that
What a people actually do with what they find? – we have very little information on that an lot of hyperbally
How do we monitor the reuse of material? – I dont think we necessarily have to but it certainly makes a stronger case to having open teaching resources if you can show this benefit
should we be embeding the correct metadata in pdfs docx and other open document formats so that the document can be less easily seperated from its metadata?
Catagorisation improves reusablity? – completely disagree with this and gave the people who suggested it a bit of ear bashing. Not pidgeon holing resources into “subject” areas encourages better cross disciplinary reuse.
All in all the day was productive but it was a long days travel and im not sure we really succeeded in doing anything that wouldnt have been better done over skype or instant messenger. However i have now met Pat Lockley in person and a bunch of other good people who im sure will be valuable allys in future work.
Hypertext day 3 – Tagging and Hyperstructure
So iMapping, thats pretty great. Check it out. Personal knowledge management is not the most focussed knowledge management. Most people dont have that much knowledge to manange. However for researchers where we always have large corpuses of material and often have to switch context there is deffinately scope. I would be a good system to try for managing your PhD. Heiko Haller the systems developer is clearly very tallented man and has put an incredible ammount of time into making the software polished and functional. It allows to create mind maps, concept maps and just express relations between data in a spacial hypertext. you can collapse and expand compex concepts and hide/show links between things to hide complexity. The system has a zoomable intface which allows you to focus on area of the concept map very quickly and easy any. Just zoom in an more detail is exposed. I hope at some point ill have time to create a spacial hypertext of map of hypertext 2010 as a means of demonstrating the system. In the mean timesee also: imapping.info
This afternoons talks about tagging are very close to my concerns. A lot of people have used ways to model tagging behaviour but in this section people seems to be looking at what the semantics of making a tag are. What does making a tag really mean? It seems from Christian Körner’s presentation that there are two types of tagger, catagorizers and describers. One is movitvated by personal rediscovery and tend to have small vocabularies (low number of different tags often reused). Others are discribers who are using tags to describe the resource for other users. So for these kind of users presuably their tag means (semantically) something different. This information would be incredibly useful for humbox and Edshare and all their various brothers and sisters. We really arent making enough use of our tags and Hypertext has showed me that there is some real value there to be exploited. But what about reading.ecs where tags are derived by the system? Enter Marek Lipczak:
Marek has been looking at exactly that. How similar are tags to keywords we can derive from the title. I have to look into it more closely but it looks very promising. My opinions of the research we radically escalated when I made a monumental appearence in the presentation. A screen shot of delicious page for the hypertext website was show and Dave Flanders (JISC) was one of the taggers. Furthermore one of tags he used was my name! So im famous by proxy.
Another interesting idea discussed was pivot browsing. The idea that when you return your search results you also return a list of tags or tag cloud which are tags related to the results you return. I like the sound of the idea. A blended approach between searching and browsing. It may help to reduce the problem of retrieving a small number of search results. Even a list of 3 results can make useful list of related tags. There was also some an interesting discussion about the style of tagging on twitter. Hashtags which have become synomous of twitter but their semantics are very different to that of say a delicious tags. Tags are not often designed for rediscoverablilty because twitter’s archive is not that long lived. Often tags will be discussion areas or just for the lols. One resounding by product of twitter tags is what Jeff Huang described as mirco-memes. Tag which is incredibly popular for 1-5 days but then disappears almost completely. One of the examples included #willgetyouslapped. It’s and interesting idea, that something can go into fashion and out again in one beat of the webs heart.
Finally for the day was the Frontiers strand. This struck me as being “big science”. I was incredibly impressed by the research. The talk I was really taken by was by James Goulding and his talk on Dimensional Hypertext and hyperstructures. He made some very interesting points. So interesting that it is my aim to read his thesis (the first i have ever read out of interest). Sadly it has proved very hard to get hold of but theres some really good information science in there im certain. Also his presentation was absolute dynamite so kudos!
Hypertext Day 2 – Keynote, Search, Recommendation and E-Lit
So today was the first offical day of the Conference. This morning started with the opening keynote. As with all keynotes the range of topics covered was broad. There was some really strong emphasis so far on both days that what computer science is doing is not really new. There seems to be a lot of psycological and soloilogical theorys which could be applied to the particularly the social aspects of the web. The research then becomes discovering how existing theorys map into a web experience. This would mean doing some really deep genuine interdisciplinary research. It strikes me that the sort of interdisciplinary research we currently do is very superficial, it tends to be – build a system for some Language teachers. A deeper piece of research would be “take what language teachers know about teaching languages and apply those theories in a web context” thats not a very good example because im not convinced language teachers really know about how their students learn. That means we need some learning theorists which probably leans on psycology. We need some hard metrics to measure whether learning is actually improved and then weve got a really solid piece of research where theories from two disciplines combine to their mutal benefit.
Following that inspiring start to the day there was a very interesting piece about the <title> tag of a document and how persistent it is. It turns out about 40% of web pages have a crap title and it makes them pretty difficult to google. Thats quite a useful piece of information to know you can also do some studies about how well the title reflects the content of the document and how the document changes and the title changes or not over time.
Then Charlie’s talk was on. He had a very well polished and quite stylish presentation compared to what I’ve seen before and it was backed up by his usual sturdy presentation style. He woke the audience up a bit, it was the first talk of the day to really get any questions and they were pretty tough but at least people were listening. Another presentation of note was Automatic Construction of Travel Itineraries using Social Breadcrumbs by Munmun De Choudhury et al. Again no copy of the presentation online yet but the paper is currently self archived. The plan was to use geotags from images in flickr in order to construct virtual tours of Cites based on where users actually went. What they produced looks incredably successful and I don’t think there choose method of evaluation really does it justice.
So by now i was a bit tired sitting and being talked to so me charlie and bunch of the other delegates grabbed a break out room and had a discussion about e-lit and hypertext fiction. These are things im not particularly familiar with but they that made the discussion all the more interesting. I’m told hypertext usually has a strand for e-lit and so forth but this and the 10 or so people we were in the discussion with felt a bit put out it had been dropped from the program.
To finish the day we headed over to the poster reception where there were some quite interesting presentations and some really good sushi. Tomorrow is more social web focused so should be more in my usual workspace.
Hypertext Day 1 – Workshop on Modelling Social Media #MSM10
So today is the workshop day of Hypertext. The talks are fairly informal and its nice to be eased into the swing of the conference. I have been attending the Modeling Social Media workshop (#MSM10). There were some pretty neat talks and some interesting discussions. I’d like to touch on the workshop keynote by Ed Chi. At the time of writing Ed’s slides are not online which is a shame but the workshop organiser tells me all will be eventually. He has A LOT of interesting models of social media very scientific and lots of statistics and other good science stuff. Being a practical man I took away this citation on “Online Reputation Systems: How to Design One That Does What You Need“. It is a nice piece analysis that basically explains that the way you you build reputation systems in your site will encourage different user behaviors. It outlines 4 different types of behaviour you might want to encourage and what to do and not to do in order to encourage it. I will definately be baring closely in mind as im building MePrints widgets for EdShare.
Another aspect, discussed by Else Nygren, was social grooming. While the research is quite early on the initial findings are certainly intuative. If a person grooms another it encourages the groomed to participate further in network and also increase the likely hood that the groomed will later become a groomer. That certainly reflects the experience i have had on humbox.ac.uk. When the users started receiving comments they start to comment back and also upload further resources. I should get some stats together to prove that but the point is it seems very important to think about groomability when designing your site. How can you make a system where it is easy for one user to groom another and let a user discover they have been groomed.
I guess that ties a little bit into what my friend Clare Hooper back in ECS is doing on TAPT. Its a software engineering technique for recreating the emotional experience of an activity rather than just making a replica of the physical form (example word documents basically emulate paper rather than recreating the experience of writing a letter). Im sure Clare and Else would agree that the reason rich text input on commenting and forum systems are useful because they facilitate some of the emotional experience of talking to person (even tho they dont emulate it). It allows people to express some of the things you would do in a face to face conversation like put empasis on a word or make facial experessions.
I think thats enough for today. I may post a little video when i get back to the uk and work out if any of the footage is any good.