Repository Benefits (again)

A well-respected colleague (who shall remain nameless) recently posted a Tweet which likened "repositories with end-user benefits" to "pigs with wings". Shome mishtake, shurely?

As part of my job I am course co-ordinator for the MSc in Web Science here at Southampton. Last week I ran an Industrial Liaison day for the course, at which the students made poster presentations of their work to the industry representatives. It was a fantastic time, and the attendees all commented on how stimulated they had been by the posters. Consequently, the secretary who had printed out all the posters was able to give me a ZIP file which I then uploaded to our teaching repository. (See http://www.edshare.soton.ac.uk/4790/.) Now I have a record of a key part of the event that I can share with the industrial delegates. It was quicker than me creating a bespoke website, or uploading them to my home page. Plus I can use it as a marketing and publicity resource for attracting other students. And the students are happy because they have learned that their contributions aren't just "marking fodder", but they have valuable statements to make to a public audience. Win, win, win.

Not so much pigs with wings as pigs with snouts: you have to sniff out the opportunities. Or even carpe suem!

How Repositories Can Contribute Linked Data

I've been working a lot with our repository and Linked Data teams (thanks to Hugh Glaser, Nick Gibbins and Iain Millard) on the JISC dotAC project. One of the great things about that project has been the opportunity to really get our heads around the role of repositories in the Semantic Web and the Linked data world. Now that the project has finished, I've finally had the opportunity to sit down with Chris Gutteridge and braindump our understanding so far. The following is a description of how EPrints (v3.2 beta) exports its holdings as Linked Data. It all comes down to how it uses and resolves URIs. (Like it says at the bottom of this posting, please send us comments!)

We ASSIGN a URI to all the significant entities that the repository owns: specifically, eprint, document, file, user, and "subject taxonomy" objects.

1. These URIs will generally be of the form http://repository.com/id/type/id where type is eprint, document etc and idis unique within the scope of the repository.
2. The official URI of an eprint is of the form http://repository.com/id/eprint/id whereas the official URL of an eprint is of the form http://repository.com/id . The latter has historically been the URL of the eprint's abstract or splash page.
3. Where possible, resolving a URI will result in a "303 See Also" redirection to the URL of the most appropriate format export, based on content negotiation of available exporters (disseminators).
4. However, if text/html is deemed most appropriate for an eprint, it is redirected to the standard URL of the abstract page.
5. For a sub-object (e.g. documents and files of an eprint) the URI is redirected to the ancestor object.
6. Similarly, subjects redirect to the top-level subject.
7. Eprint and document objects have special "relationships" fields which allow arbitrary predicates/objects to be attached to the document/eprint.
8. Documents of format text/n3 and application/rdf+xml (which like all documents have their own URLs inside the repository) are linked to the parent eprint via an rdfs:seeAlso statement. This allows arbitrary triples to be associated with any eprint, irrespective of the repository schema.

We MINT (or COIN) a URI for entites whose existence we infer from metadata.

1. Where there is a high degree of confidence from the metadata that two entities are the same (e.g. two conferences, journals or authors) then they will receive the same URI.
2. These URIs will generally be of the form http://repository.com/id/x-type/id where type is e.g. event, organisation, person, place etc and id is unique within the scope of the repository.
3. The unique id is generally a hash generated from the metadata, unless a better value is available e.g. an ISBN.
4. In the case of a book or a serial we can confidently add an owl:sameAs to the URN
5. In other cases, a repository administrator can add a mechanical process for creating the sameAs using specialised knowledge to construct (or map) the metadata to an external URI. An example of this would be looking up an author's URI in a staff database based on an email address in the author metadata. Another example would be a DOI being constructed from a query to the CrossRef database.
6. (The reason for using an x-publication URI as well as the public URN is that it may be useful to provide a local resolution service for non-local entities.)
7. All x-type URIs redirect to an RDF+XML document, describing everything known locally about that entity. Content negotiation for other formats is not currently supported for these non-core entities.

A set of standard triples containing rights information about the exported metadata appears in every single RDF export, to facilitate linked data reuse.

Does this look sensible? Have we got it right? Please let us have any comments and feedback!

Starting out with Hypertext

Hypertext taught me that learning is really fulfilling, it's what helped me to understand algebra and generally get gave me the confidence to succeed at school. And not a computer in sight!

I was about seven (it was 1971) at the time that I discovered "Tutortext: Basic Mathematics" in the village library. It was a volume from an American series of popular education materials aimed at those who "wish to learn ... and yet have to teach themselves". The educational method was devised by Norman Crowder, from the Educational Science division of US Industries Incorporated - an outfit that in retrospect sounds like a front for James Bond!

And well it might have been, because its unusual style seemed just as futuristic as one of Q's gadgets. Each chapter started with a page of explanation and a question with multiple choice answers leading to other pages; some of them explained where you had gone wrong, and one of them congratulated you on your progress and took you on to the next step. It all seems rather pedestrian now with our history of computer-assisted learning and personalised and adaptive hypermedia, but to me at the time it was just magic. It forced me to engage with the problems and consider my solutions and to seek the praise that the text meted out!

I have just managed to track down a copy of the book (from an Amazon reseller!) and it has provided me with a tremendous dose of nostalgia. Still, I might just give this book (copyright dating from the year before I was born) to my youngest daughter to see if it helps her.

HyperCard is Dead. Long Live HyperCard!

I cut my professional teeth on HyperCard, writing VideoDisk XCMDs to allow a HyperCard stack to control a video presentation, back in 1986. Although I was a UNIX system programmer (cut me and I bleed regexp), it was Apple's HyperCard which best let me manipulate data for users.

And now it's back in the form of TileStack, a kind of re-imagination of HyperCard for a Web 2.0 environment. There have been other contenders (e.g Runtime Revolution) but they didn't have proper integration with the Web. Now I can write stacks (in a HyperTalk-like language) that use AJAX Web Services - XML, JSON the lot. I'm as happy as Larry!

The following embedded stack uses an idle handler to periodically make a Flickr API call and then set the icon of button n to media of item 1 of the items of JSONdata. I'd forgotten how simple this stuff was - come back Bill Atkinson, the Web needs you!

Evaluating Expertise Promotion

I thought I'd look further at the issue of effective communication of research impact and expertise. The University of Southampton Communications team made a press release on the subject of "Brain-Computer Interfacing" earlier this term. It's obviously because they believe that we as an institution are good at it and that we have something to promote.

I thought I'd take a look at how effective our communication on the subject is, and as you can probably guess, this equates to how high up the Google ranking do we feature compared to other universities? This is a pretty good measure of the effectiveness of our research expertise promotion because anyone who wants to find an expert on a topic is going to start by looking on Google. I knocked together some scripts to look at how our institution fares in the competition for Google eyeballs (basic web analytics).

The screendump on the left shows the results of a Google query for "Brain-Computing Interfacing". All the results from universities are coloured in red, all those from publishers in green, those from news sources, magazines and blogs in blue and unclassified resources are grey. You can quickly see that there are a couple of university-contributed results right at the top, and then an increasing number further down. Those results with a silver background are from Southampton (yay!)

Clearly, Google looks for resources which are about "Brain-Computer Interfacing" and then ranks them according to their "impact" or "importance". Exactly how it does that (PageRank or Black Magic) isn't really my concern here; however it happens, Google controls the order in which these results are presented, and the effect is that if you appear near the top you are more likely to get visited. The script that I use to generate these annotated pages actually gets 500 results, but most people get 10 results per page and don't bother to ask for more than a single page.

To better compare institutions' effectiveness at promoting their research expertise, I distilled this page to a spreadsheet. Once it's in a data form, then the sky's the limit and I can visualise it in different ways (such as this map of global expertise in the area).

The spreadsheet (reproduced below as a table of institutions) shows me that Southampton comes out rather well in the area - we are the third institution named after Oxford and Gronigen, and that slightly further down the list comes a couple of papers from our Institutional Repository. This seems to be a good result - we can claim to be doing alright on this topic. But what about all the other areas in which we think we have some expertise? There are hundreds and thousands of keywords that we need to analyse to see the effectiveness of our communications overall. I think its time to scale up my scripts to get a bigger picture!

Institution	Country	Server	Title
Oxford	UK	www.robots.ox.ac.uk	Brain Computer Interfacing Project
Groningen	Netherlands	www.ai.rug.nl	Moving Thoughts - A Brain-Computer Interfacing Project at RuG
Southampton	UK	www.bci.soton.ac.uk	The Southampton BCI Research Programme
Ulster	UK	isrc.ulster.ac.uk	Home | Brain Computer Interfacing and Assistive Technologies Team
Lausanne	Switzerland	infoscience.epfl.ch	Anticipation Based Brain-Computer Interfacing (aBCI) - Infoscience
U Twente	Netherlands	eprints.eemcs.utwente.nl	EEMCS EPrints Service - 11091 Brain-Computer Interfacing for .
Plymouth	UK	cmr.soc.plymouth.ac.uk	Brain Computer Music Interfacing Demo
Malta	Malta	www.um.edu.mt	Brain Computer Interfacing - Systems & Control Engineering ...
Carnegie Mellon	US	www-2.cs.cmu.edu	Classifying Single Trial EEG: Towards Brain Computer Interfacing
Southampton	UK	eprints.soton.ac.uk	e-Prints Soton - Brain-computer interfacing in rehabilitation
Southampton	UK	eprints.soton.ac.uk	Cognitive tasks for driving a brain computer interfacing system: a ...
Cardiff	UK	www.caerdydd.ac.uk	Brain Computer Interfacing
Rhode Island	US	www.ele.uri.edu	Brain - Computer Interfacing Mason P. Wilson IV URI department of ...
Hosei	Japan	ijbem.k.hosei.ac.jp	Brain-Computer Interfacing in Tetraplegic Patients with High ...
Colorado	US	www.cs.colostate.edu	Temporal and Spatial Complexity measures for EEG-based Brain ...
UC San Diego	US	inc2.ucsd.edu	"Gerwin Schalk, Ph.D."
Washington	US	www.cs.washington.edu	Dynamic Bayesian Networks for Brain-Computer Interfaces
Birmingham	UK	prism.bham.ac.uk	Kianoush Nazarpour Home Page
Glasgow	UK	www.dcs.gla.ac.uk	A Note on Brain Actuated Spelling with the Berlin Brain-Computer ...
Uni Saarland	Germany	psydok.sulb.uni-saarland.de	PsyDok - Brain Computer Interfaces for Communication in Paralysis ...
Essex	UK	cswww.essex.ac.uk	Feature Selection and Classification in Brain Computer Interfaces ...
U Freiburg	Germany	www.bmi.uni-freiburg.de	BMII: Ferran Galan
Kansas City	US	www.csee.umkc.edu	CIBIT Laboratory
Manchester	UK	mint.cs.man.ac.uk	Microsoft PowerPoint - LeslieSmith
Southampton	UK	www.bci.soton.ac.uk	The Southampton Brain-Computer Interfacing Research Programme - Aims
Brown	US	www.cs.brown.edu	Michael J. Black: Neural Prosthesis Research Projects
U Tuebingen	Germany	www.mp.uni-tuebingen.de	Institut fur Medizinische Psychologie und Verhaltensneurobiologie ...
Tsinghua	China	neuro.med.tsinghua.edu.cn	Invited Speakers
North Florida	US	www.unf.edu	UNF Webpage
UCL	UK	www.cs.ucl.ac.uk	"Presence: Research Encompassing Sensory Enhancement, Neuroscience ..."
Washington	US	www.cs.washington.edu	Pradeep Shenoy
Cardiff	UK	www.engin.cf.ac.uk	Cardiff University > School of Engineering > Research Groups > PhD ...
Carnegie Mellon	US	www.cs.cmu.edu	Automated EEG feature selection for brain computer interfaces ...
Ulster	UK	isrc.infm.ulst.ac.uk	About the Group - ISEL - Intelligent Systems Engineering Laboratory
Tufts	US	www.cs.tufts.edu	COMP 250-BCI Syllabus
Essex	UK	cswww.essex.ac.uk	AABAC Publications
Hosei	Japan	ijbem.k.hosei.ac.jp	Evaluation of a Robot as Embodied Interface for Brain Computer ...
Stirling	UK	www.cs.stir.ac.uk	Professor Leslie S. Smith: Research Home Page
New Jersey	US	embc2006.njit.edu	Blind Source Separation in single-channel EEG analysis: An ...
Pace	US	www.csis.pace.edu	www.csis.pace.edu/~ctappert/dps/d860-08/raghu.ppt
Plymouth	UK	www.tech.plym.ac.uk	BCI Papers
Bielefeld	Germany	ni.www.techfak.uni-bielefeld.de	Publications | Neuroinformatics Group
Washington	US	ese.wustl.edu	EIT - Sample Web Page Template
Colorado	US	www.cs.colostate.edu	EEG Pattern Analysis
Columbia	US	www.bionet.ee.columbia.edu	Bionet Group @ Columbia University
Kent	UK	www.kent.ac.uk	Eleanor Curran - Kent Law School - University of Kent
Georgia State	US	www.cis.gsu.edu	EEG-based communication: a pattern recognition approach ...
TU Graz	Austria	bci.tugraz.at	Publications - Laboratory of Brain-Computer Interfaces
Northeastern	US	nuweb1.neu.edu	News
Cardiff	UK	www.engin.cf.ac.uk	Cardiff University > School of Engineering > Contacts and People ...
Plymouth	UK	cmr.soc.plymouth.ac.uk	Computer Music Research
UCL	UK	www.fil.ion.ucl.ac.uk	Journal Articles
Ulster	UK	www.infm.ulst.ac.uk	Final Year Undergraduate Projects 2008-2009
UC San Diego	US	sccn.ucsd.edu	"IN RECENT years, brain-computer interface (BCI) systems"
Ulster	UK	www.socsci.ulster.ac.uk	Slide 1 - Faculty of Social Sciences
Carnegie Mellon	US	www.cs.cmu.edu	Linear and nonlinear methods for brain-computer interfaces ...
New Jersey	US	embc2006.njit.edu	On-line Differentiation Of Neuroelectric Activities: Algorithms ...
Bielefeld	Germany	bieson.ub.uni-bielefeld.de	BieSOn - P300-based brain-computer interfacing
TU Graz	Austria	hci.tugraz.at	Publication list of Alois Schloegl
Uni Saarland	Germany	psydok.sulb.uni-saarland.de	Brain Computer Interfaces for Communication in Paralysis - Eingang ...
Essex	UK	dces.essex.ac.uk	Selected Publications
Brown	US	www.cs.brown.edu	Michael J. Black: Neural Prosthesis Research Projects
UCL	UK	www.gatsby.ucl.ac.uk	Stationary Subspace Analysis
Colorado	US	www.math.colostate.edu	Curriculum Vitae: Michael Kirby (Professor) Co-Director Pattern ...
Oxford	UK	users.fmrib.ox.ac.uk	Reza's Homepage : Resume
Ulster	UK	isrc.ulster.ac.uk	Contact | VGandhi
TU Graz	Austria	www.igi.tugraz.at	"Seminar Computational Intelligence E, SS 2007"
North Carolina	US	catalog.lib.ncsu.edu	NCSU Libraries - Toward brain-computer interfacing / edited by ...

PS It did occur to me after I had published an earlier draft of this post that I should also have checked the results for "Brain-Computer Interface". It turns out we come further down the league-table for this variation on the phrase (7th institutional position rather than 3rd) but that this phrase is much less-commonly used. As long as potential funders, students and media researchers know which phrase to use we should be alright. Otherwise, we will have to become a bit more canny about our use of synonyms. (I'm not sure whether it's significant that an EBay sponsored link appears only on "Brain-Computer Interface"!)

Life is a Conference (Oh Chum)

Since EPrints has now celebrated its 10th birthday** I have been chewing over where this decade of repository activity is leading us. Bigger repositories? More repositories? Faster repositories? Better repositories? Well, yes to all the above, but collecting, curating and sharing data/documents seems to be only part of the picture.

At the same time, I have become a director of the Web Science Doctoral Training Centre at the University of Southampton. Its five year mission (no, really, it's there in the EPSRC grant letter) is to build up a cohort of interdisciplinary scientists who can understand the impact of the Web on our society - its economic activity, political exchange, social interactions, scientific knowledge transfer - and predict the future benefits and downsides of different kinds of Web technology.

For the last few days I have been trying to pull some of these pieces together: the Web, the Social Web, the Data Web, repositories, open access and open science. In recent years, the community has built a Web infrastructure for e-research that handles research outputs, research data, research process and workflows. This infrastructure has many desirable properties - it is dynamic and persistent and supports managed curation and auditable provenance.

One thing I believe is missing from the picture at the moment is research people, research careers and research meetings. Researchers engaged in human-oriented research activities, rather than research artefact and research experiments. Research after all isn't just about individual scientists turning dials on a piece of laboratory equipment, but about many individuals debating and evaluating their ideas in scientific discourse and scientific debate. Part of that discourse and debate happens through journal publications, but much of it happens in conferences and workshops, through face to face interactions. The proceedings of these meetings become part of the literature, and so part of the personal, dynamic, face-to-face engagement is captured for posterity, but the questions and answers, the ad hoc discussions, birds of a feather sessions and arguments over dinner - all the normal human interactions that generate inspiration as well as larger scale knowledge transfer - have not been captured.

Except that they are starting to be exposed beyond the boundaries of the conference meetings by microblogging services. The low barrier to communication afforded by a Twitter client on a smart phone means that ideas, controversies and emerging consensuses are broadcast beyond the immediately present delegates in a meeting. These communications are not edited, published and catalogued for posterity (and are only searchable for a short time), but they do (potentially) increase the efficiency of the meeting.

A decade ago, the only way to facilitate social networking in the research world was by face to face meetings; flying hundreds or thousands of people half-way across the world for a week in order to be able to talk to each other (or perhaps even to listen to each other). This is still the way by which much research business is conducted, despite being more aware of the environmental consequences of our conferences. There must be a better way.

Twitter, blogs, web, phone, email, papers, workshops, meetings, projects, texts (SMSes) are all ways of mediating engagement between knowledge generating people. In point of fact, conferences are not very efficient engagement mechanisms - most sessions are full of people doing email. Virtual conferences (whether held in Second Life or a rather more prosaic video/audio conferencing environment) also have shortcomings in fostering participation and engagement.

We need to redesign our social interactions to make them more pro-human, pro-diary, pro-budget and pro-environment. We need to use technology not to ape our large-scale face-to-face meetings (using enormous video walls of dozens and hundreds of virtual delegates), but to support us as we try to achieve scientific debate and argument with loose synchronisation across a dynamic community of individuals spread over a number of years.

It's ironic that a university is supposed to be a community of researchers, but none of us know what our neighbours do until we accidentally meet them at a conference on the other side of the world. This is no longer acceptable as the importance of interdisciplinary research increases! Let's instead use technology to improve the social transfer of our knowledge capital with our international research community and our institutional research community too!

I believe that's where our infrastructure needs to grow - supporting our research engagement as well as managing our research artefacts.

** Technically, it is 10 years since Stevan proposed EPrints at the first OAI meeting in Santa Fe at the end of October 1999. We will have a more tangible anniversary in June 2010, celebrating 10 years since the first release of the software at the second OAI meeting.

"Getting" Twitter

Recently I sat in on a demonstration of Twitter to a University research group that included our PVC for research. Because of his presence I was quite self-conscious about justifying the Web tools I normally take for granted, and although the demo itself was fine, it didn't seem to answer the question "is this really useful or just some gratuitous teenager technology?" I have always claimed that twitter is a fantastic tool for keeping up-to-date with the spread of ideas and debate in the community - lots of micro-comments keep me in the loop about which speakers have raised what issues at which conferences, even when I can't travel and engage directly. However, I have been worried recently that the twitter output that I see has been less technical/academic/professional and more personal/informal/gossipy. So I thought I would do a quick investigation to see if there is any evidence to support my positive experience of twitter. I chose to look at Twitter activity surrounding CETIS 2009 as several of my Twitter contacts had mentioned it in the run-up to the conference.

The CETIS conference (http://wiki.cetis.ac.uk/Conference_2009_Programme) is run by the JISC Centre For Educational Technology and Interoperability Standards, and attracts many people from the E-learning community. It took place at Aston on 10th and 11th November 2009, attracting 146 delegates according to the open list on the conference website.

Over the period that the conference had been mentioned (from the afternoon of Nov 5th up till midnight on Nov 12th) 566 tweets were sent by 89 separate contributors. The large majority of these (440, 78%) were sent during the conference sessions, with 255 (45%) on the first day and 185 (33%) on the second day. Outside the conference hours, 32 (5%) were sent in the break between the two days of the conference, 57 (10%) were sent before the start of the conference and 37 (7%) were sent after the end of the last session.

How many of these tweets are merely "backchat" or "electronic gossip" and how many of them are broadcasting helpful information? I used the twitter API to download all the tweets and individually categorised them as "informational" or not. An informational tweet contains some information about the conference that is useful to an external viewer (a non-delegate such as myself). It may contain a quote from a speaker, a URL to a relevant resource, or a brief (microsummary) of an issue raised. By contrast, an example of a non-informational tweet may be a complaint about the wireless network, a comment about the quality of the food or a message of thanks to the organisers. This categorisation requires some judgement on my behalf, but the criteria are resonably straightforward and repeatable.

The distribution of tweets over time can be seen in the following figure (click to see a bigger version), which also shows how the number of "informational" tweets (red) compares to the total number of tweets (blue). In total, 324 tweets (57%) were in the informational category.

During the conference sessions, the informational tweets account for most of the Twitter activity (307, 70%). In other words, the effort expended in twittering during conference sessions is not wasteful and distracting effort from engaging with the conference agenda. It is mainly valuable to an outside observer - which I would claim extends the impact and influence of the conference beyond the cohort of local delegates. Of course, this works best if the tweets can refer (and link) to a rich set of online resources to direct observers to.

Back to my obsession with showing that Twitter isn't just an electronic stream of gossip - the figure on the left shows how people break down into different Twitter categories: those who only twitter useful information (or did on this occasion), those who never twitter useful information (not useful to me anyway) and those who mainly or partly twitter useful information (those whose information rating were more or less than 50%).

But those who stick strictly to the facts don't provide the biggest chunk of information. This figure on the left shows the contribution of the various groups of twitterers to the total information content of the tweets: most of the useful twitter information is provided by people who mix "information" and "comment".

So perhaps I shouldn't get too worried by the criticism that Twitter is full of people telling us what they have for breakfast and what happened on their trip to work. Perhaps it is precisely those kinds of people who are more likely to let us all know what the key themes are emerging from that high profile conference that we couldn't attend.