I gave a talk today on making publications increasingly visible, where I investigated several things, like announcing acceptances on twitter like Jaime Teevan does, and broadcasting news like we, epsrc, mit csail, fxpal and parc do, and generally communicating ideas and discussions online, etc.
One thing I am fairly sure about is the use of document repositories – I think Google Scholar likes them in particular. It’s structured metadata, links to papers, easily scrapable, one point of contact for a department or institution etc. Rather than people maintaining their own publications in many various ways. All sounds likes its heading in the right direction.
Now I already saw that thursday 21st Jan 2010’s papers from Southampton ECS’s EPrints were in Google Scholar today (a week later or so) but that the yesterdays paper was already on Google – scraped ‘19 hours ago’ said google nicely. But I thought I’d do a little test – I added a new recent publication to an EPrints archive to see exactly how long it would take to get onto google.
I submitted it to EPrints at 9:54am – it was on the EPrints ‘latest publications’ by 10:00am – it was on Google by 10:58am. that’s approx 1 hour! Fascinating. It will be interesting to see when it appears on Google Scholar.
The guys at FXPal and Microsoft Research (namely Gene Golovchinsky, Jeremy Pickens, and Merrie Morris) have been driving the Collaborative Information Seeking (CIS) community hard, and doing a great job of it!
Today, Gene announced the accepted position papers for their upcoming CIS workshop at CSCW2010. It looks like there some great papers in their, some from the usual CIS academics, and some new faces too. I haven’t had a chance to read them yet, but it looks like Chirag Shah, and Capra et al, are reporting on a series of interviews about CIS experiences. Some papers are studying CIS in real legal and emergency circumstances, where good collaboration can be truly important. A couple of new CIS UIs are being introduced. I bet Sharoda and Madhu’s paper, a framework for Collaborative Sensemaking, will be excellent, given their history of strong articles and papers. My paper for this is an extension and elaboration on a previous blog post of mine, about the exaggerated affects of perception during collaborative information seeking. I bet it will be a great workshop.
In the background, Gene, Jeremy, and Merrie have been collecting together preprints for their special issue of IP&M. Sounds like they’ve now got all the preprints together, so it wont be long till we see the special issue hit the shelves.
I’ve been thinking lately about wisdom. What makes a wise person? I’m sure we all have people who we believe to be wise. I can think of a few, some who are even still wise after several pints of beer. I felt like I had an incremental realisation of what wisdom really is the other night. I’m almost at the point now, where I remember being ‘younger’ and most decisions in life being easy. These days, I know more arguments, counter arguments, and can more easily perceive other peoples points of view – and decisions these days are less obvious. I then better understood the notion of wisdom, in being able to take into account the wider range of understandings, and still know clearly the right answer. This is of course different to ‘clearly knowing the right answer’ because of ignorance, or lack of knowledge.
A learned person – may be one way to think of it in terms of information seeking. The notion of supporting ‘life-long learning’ is one element of exploratory search. Pictures of wise men are often pictures of intelligent men, but were/are they also wise? Wisdom is more than knowledge, its knowing all the facts, reasons, arguments, and then still clearly knowing the right answer, choice, or action. When I’ve seen wisdom in action, the person can make a tough decision with the apparent ease of a simple one.
Webster defines it as ‘Wisdom is knowledge of what is true or right coupled with just judgment as to action; sagacity, discernment, or insight.’ The knowledge here is not just on fact, but on ‘what is true or right’. In some sense, this could be related to sensemaking. Sensemaking, however, is often associated with making sense of lots of external sources of information, or other people. Wisdom, under these descriptions, may be more related to making sense of all your own knowledge.
Regardless, going beyond discovering facts, support for growing wisdom might mean not only presenting results or answers, but highlighting and communicating correctness, authority, or even moral judgment. Assuming everyone could agree on moral judgment, authority, or correctness that is.
Anyway – the post has become a more ramble than an articulate, insightful and wise blog post.
If you’ve been following the mSpace project over time, you’ll know that we’ve been trying to push the model of how we can browse faceted metadata. We’ve been trying to find out whether faceted browsers should filter left-to-right like with the iTunes columns, or in all directions like many online shops. We believe so, and we’ve been pushing the boundaries of both the left-to-right and general model, with features like backward highlighting (see video) etc.
Now one consequence of pushing on these advanced faceted interactions, is that, with the mSpace beta code available to download here, it takes a little more work to set up an mSpace – well – until now.
Let me put that in context for a second. Exhibit, developed at MIT, runs quickly over a JSON file, assuming you put the data in the right format. This is fine, for limited sized datasets etc. The next step up is to build a typical cross-filtering system like the standard type of faceted UI that Flamenco provides. This is a little harder than Exhibit, because you have to connect filters to certain database columns etc, rather than just point to a JSON file – but still relatively easy.
mSpace leverages the structure of databases or RDF data to maintain performance and still provide the advanced interaction models. So you had to think carefully and work a little harder to set things up. The latest work by Daniel A. Smith and the rest of the team at mSpace have been pushing hard to remove that barrier.
Now they have developed an mSpace builder that you can point at any SPARQL endpoint, choose a starting class, and set up an mSpace with a few configurations. Then press go – and it’ll work for a few mins and do all the hard work for you.
Anyway, as I, you, we prepare for the next workshop, it is great and exciting to see pre-prints of the articles soon to be part of the Information Processing & Management Special Issue on Collaborative Search go out on the RSS feed and become available online. I haven’t read all the articles myself, yet, in full, but I’m totally excited about the range of work.
Hertzum, from Roskilde University in Denmark, has studied breakdown in collaborative information seeking, in medical circumstances, which should reveal some fascinating elements, limitations, challenges, etc in the way we communicate. There’s nothing like studying where/how an idea fails in order to understand it better.
Chirag Shah, from UNC Chapel Hill, of the aforementioned onion model, along with colleagues at FXPal, including the aforementioned Golovchinky and Pickens, have studied algorithmically mediated collaborative search, where the system helps and facilitates collaboration based on the understanding of the roles they are taking in a team. Sounds like they extend their previous work on Prospector and Miner, to study Gatherer and Surveyor roles.
The most recent one to go up is my own, thinking about using IS theory to evaluate our new search designs from different role-based perspectives. I won’t promote it heavily like the other exciting work – you can judge it for yourself. But I believe there is at least one more to come! oh the excitement.
Anyway – good luck with your CIS Workshop submissions.
After thoroughly enjoying the HCIR’09 last week – I had the opportunity to attend SWUI’09 – which was located at the international semantic web conference this year. The idea being, of course, to understand how the change from a linked web of presentation documents to a linked web of data separated from presentation might change the way we interact with information. For me that means – how can we explore and make sense of vast amounts of data? We are getting so much data coming in from open government data, etc. What are we gonna do with that?
The workshop, which had a diverse range of views (from academia, to start-ups, to industry, from profs to phds, from europe to south america), continued as a pretty high-level theoretical discussion of how you can separate UI and semantic data improvements. What happens when you are building a UI to understand the semantics of complicated information, like relationships in biomedical data or the relationships between cognitive science terms? At that point, you are dealing with semantics, using a semantic web of data, but trying not to show the linked data explicitly. Its a bizarre challenge that blurs the line of hiding UI challenges. Squishy-mediascognitive atlas project is a good visual example of that.
So what were the take-aways from the event? That’s a good question, and to answer it we need to review the history of where SWUI has been recently. Since the SWUI awareness rose to its peak at the ISWC2006 workshop panel with Tim Berners-Lee, Wendy Hall, Nigel Shadbolt and more, the community has still been ’speccing out the issues’, and with, so far, little resolve. The 2007 co-located MIT/Zurich SWUI workshop produced a more formal agenda after the 2006 panel. In 2008, the dialogue continued by discussing the required integration of personal, social, public, and situational data, thus identifying the inter-dependency of these sources. This year, the core topic of the high-level discussion was on the blurred boundaries between Semantic Web User Interaction (SWUI) and Meta-SWUI. In our exploration, the consensus of the debate was that our developing UIs and future UIs need to focus on the seemless switch between finding and organising our finding. That is, our UIs will need seemless interactions to control the process of interaction, as well as enabling the goal of the interaction.
There was one more concrete question in the discussion: are we seeing any examples of people actually benefiting from the Semantic Web? If so – list em!?
We have seen a deluge of UI designs that have benefits for the Semantic Web, with simple interactions that create metadata and the all to common ‘export as RDF’ button. With the linked data being released by governments, novel apps of linked data are being created, such as the stumble safely app, which links crime data with night-life spots. So are we at a critical turning point where we can bring the envisioned benefits back out of the Semantic Web?
We see this as the core challenge for the year ahead: lets provide a benefit for a real person, with a novel interaction that has been enabled by the increasing volume of Linked/Semantic Web data.
I havent had much to reflect on the HCIR’09 workshop, while staying with family nearby, and have in fact been to another workshop since. But I’ve gotta say the program was excellent. I tried to tweet as much as possible about the event, but even better: the organisers are submitting a report about the workshop to the SIGIR Forum – so look for that shortly. In the mean time – here’s some of the things I thought were interesting.
Ben Shneiderman began by providing a thought provoking keynote speech reviewing past, present, and what might be the future key research agendas. As our understanding of information seeking has expanded from query-response to resolving information needs, our UIs are going to progress to search for a purpose: collaboration, decision making, and societal improvement. Ben reviewed the current UN major challenges, pondering how our research into exploration and learning interfaces will support or resolve these challenges. He emphasised the important roles of mobile technology in developing countries and social media resources.
With Ellen Vorhees in the room, the notion of a Cranfield (which could do with filling out a little on wikipedia) or TREC style evaluation method for HCIR came up many times. Ellen noted how the abstracting out of a user was done purposefully to make a consistent platform. But as the group, noted explicitly by Micheal Cole of Rutgers, came to agree is that the notion of a document collection in the HCIR community was arbitrary. The aim is not to produce an accurately ranked list in HCIR – in fact HCIR work perhaps assumes a well ranked list is a constant in design. So instead of a corpus, query, and ideal result set – what do we have for HCIR? A need, a description of what constitutes resolving that need, and a notion of progress? whats the key abstraction required for an HCIR TREC platform?
There were many interesting posters, but I was pleased to hear Jacek’s active pursuit of aspects of cognitive load in information seeking. Jacek did find a level of UI affect on cognitive load, but also an affect by task stage. He has journal articles on the way about it – those will be the really juicy reads.
There were various interesting projects too from UNC, with Diane Kelly looking further into the notions of interaction-less interface advances for searchers, based on Marcia Bates‘ idea tactics. An idea I have been pressing with my own work. The logical extension of her suggestion that IQE helps users get ‘unstuck’ is that a structured data representation like facets would be even better. Barbara Wildemuth, along with Luanne Fruend, are also studying task structure for user studies, noting that there are many elements to hard search tasks, such as complexity, unfamiliarity, compound requirements, and information availability. It will be nice to see this taxonomy arise – and maybe my own investigation into search needs might be able to populate it with examples. We’ll see.
HCIR’10 will be co-located with IIiX’10 With Nick Belkin and Diane Kelly setting up IIiX’10, in collaboration with the HCIR guys, it’s sure to be an excellent event. Further, I’m planning a provocative poster session to engage attendees in discussion, thoughtful debate, and the planning of future collaborations – I look forward to see you there for both in August 2010.
Im already thoroughly enjoying HCIR. and tweeting about it where I can. Ben S is giving a great keynote, which I am trying to keep up with Follow the event on twitter.
I’m extremely excited about the two workshops taking part in Washington DC next week! First, the Catholic University of AmericaInformation Science school is hosting the Third International Workshop on Human Computer Interaction and Information Retrieval (HCIR’09)! Then, as part of the International Semantic Web Conference (ISWC’09), is the 6th International Workshop on Semantic Web User Interaction (SWUI’09). Both should be fascinating and a great chance to catch up with both the Exploratory Search folks (like Ryen, Daniel, Bill, Gary, Gene, and Ben) and the Semantic Web folks (Duane and LLoyd). Maybe I’ll run into my old friendTim again too.
We’re still waiting for exact details of the SWUI’09 workshop, but having had a sneek peak at the papers being reviewed, it’s gonna be interesting. The flexibility of information on the Semantic Web makes for potentially endless interesting interactions with information, so hopefully a bunch of the HCIR community will attend. Our presentation will be about using our information seeking inspection method (Sii) to assess Exploratory Semantic Web interfaces.
I’ll be in DC late on the 22nd and flying back out of DC on the 26th. A quick but exciting visit.
Hi all, many of you would read my blog posts on the old blogspot address. I’ve been pretty loaded recently, not blogging, but getting my PhD sorted. Last week I successfully defended, and now I thought it was time to get going again, and on better blogging software. I’ve wordpressed it up and put a link from my old address.
