Digital Resources Dep. People report from ECDL2010
Have been meaning to write about our fascinating conference experience for days! This is it.
The first keynote was by Susan Dumais from Microsoft Research on a study in retrieval of dynamic web information. It was interesting to hear about users reusing the same queries and revisiting the same pages but in different re-visitation patterns (fast, hybrid, medium, slow). Esben has some very nice drawings in his notes 🙂 Also 43% of search activity is repeat clicks or repeat queries. Also have handwritten notes with nice table. The DiffIE plug-in highlights the changes since the user last visited a page, and the study showed that users noticed more changes using DiffIE and revisited pages more. The DiffIE plug-in highlights the changes since the user last visited a page, and the study showed that users noticed more changes using DiffIE and revisited pages more. This could also be used to see new results on the same query in a search interface.
After the keynote, Esben went to the session on metadata and Bolette went to a steering committee meeting.
User contributed descriptive metadata for libraries and cultural institutes:
Easing user interaction by using known “portals” for allowing users to contribute knowledge.
Exposing hidden collections, with verry little or no meta data to the public, in order to use crowd sourcing to enrich the data.
1.Personal & historical
3.corrections & translations
Crowd cataloguing NOT professional, will be entered into meta data with flickr.com as source… (is flickr.com known in 100 years)
Steering Committee Meeting
The SC meeting raised some interesting discussions. The TPDL charter discussion was mostly on organisational details. The discussion on how to get good quality papers in an interdisciplinary conference was more interesting. The challenge in cross-field conferences is that we try to write papers, that can be read by everyone in the community. This means that we leave out the important technical parts specific to our own field and the paper ends up more superficial and somehow less scientific… We also talked about introducing multiple themes in next years conference: TPDL2011. And about a number of other things…
Next there was a panel session.
Developing services to support research data management and sharing
Mandatory data management plan in research grant applications, are more and more common. There is a need for a “creative commons” approach to the curation and preservation of research data.
There are trust issues in who has the data that is curated, and preserved. Both in relation to confidentiality and the data being preserved.
Transparency in data records.
Peer review hardly ever touches on the gathering of data and it’s relation to the results…
What is being done to alleviate the sharing of person-sensitive data? not much at present and it is beginning to become an issue, recommendations are to consider this early on in the grant application process.
Measures mus be taken in order to insure consent for data sharing and archiving! At present ethic advisors recommend destroying the datasets upon the end of person-sensitive research.
up front prepare for sharing!
Problems exist in the preservation and curation of datasets, how will the “user guide” to the data be preserved in complex datasets.
This area needs to write and talk about the successes of data sharing.
After the panel it was time for an extended coffee break, meeting an old friend and relaxing before the evening program. The welcome reception at Glasgow City Chambers was amazing! Well, the reception was nice as receptions are. Talked to nice people, had a bit of wine, listened to an official city representative welcoming the conference to his city and talked to more nice people, but the building was absolutely amazing! And the nice Indian student helper gave us directions and reserved a table at a nearby Indian restaurant for us, and we had a great dinner!
The second keynote by Liina Munari from the European Commission was on EU-supported DL projects and included a lot of slides with a lot of bullet points…
Next Bolette went to the Digital Preservation session and Esben went to the Web 2.0 session.
Digital Preservation Session
Then there was a presentation on estimating digitization costs and one on a vocabulary for preserved new media art. New media art with both digital and physical elements or maybe interactive present interesting preservation challenges, and maybe with this media rather than preservation we should talk about facilitating recreation or re-performance.
Web 2.0 session
Privacy aware folksonomies
The problem regarding privacy in folksonomies are, to a large extent, related to the fact that the provider has all the knowledge about all the users and all their tags. This issue can become a real privacy nightmare when tags are analysed in order to associate users in groups, it is likely that all users who tagged images with “
bobs_wedding" also attended said wedding, imagination and varying degrees of evil seems to be the limiting factor in the possibilities of data mining.
A possible solution combining cryptography and data split over several databases was shown, this allowed a granularity of access, between users, friends and providers.
SEAmless WEb EDiting for curated content SEAWEED
A project to release the user from the chains of the
edit->review->publish process, it’s basically just a new take on web editing, but looks promising, especially in conjunction annotation.
Details are available here(WP plugin) and here.
Before lunch the somewhat nervous Esben did a very good poster spotlight presentation ending with “And this was the…” Look at watch… “50 second presentation of my poster.” And he did get the audience to laugh 🙂 After lunch was the poster session and quite a few people came to talk to Esben about his work. Some people also came to talk about his fascinating fabric poster, especially the people who were presenting posters themselves and had been travelling with big paper posters rolled up. Bolette thinks Esben got a number of good ideas for further work but unfortunately has too little time… Bolette also went to see some of the other posters and talk to some people.
Once again the late afternoon session lost to the need for relaxing. We went shopping on the way back to the hotel. Whisky, book, dvd. Slept for 45 minutes and walked via the botanical garden to the conference dinner, which was at an old church converted to restaurant complete with whisky bar in the cellar. It was very impressive and we talked about whether it would be a good idea for some Danish churches, since we don’t seem to be using them all that much for their intended purpose anymore. Oh, and the bagpipe was fun and the food was good.
By Thursday everyone was tired and the Theatre 1 didn’t seem quite as full as the previous days. We went to the session on query log analysis and the one ontologies and TPDL2011 was announced. And then we used 11 hours travelling home!
Query log analysis session.
Determining time of queries for search of re-ranking
This talked was focused about how the time of search and the implicit information in the search terms can be analysed and used in retrieving the correct data entries, the talk focused on how the terminology changes over time. The term “
Hillary Cllinton 1997-2002” should return many of the same results as “
New York Senator” and “
The team had found that of searches 1,5% eg. “
Presidential election 2008” uses explicit time queries, and 7% use implicit time queries eg. “
German world cup”
As many events occur at a specific time, or at least over short periods, the researchers introduced Time granularity which returned relevant search results for the period of granularity
Questions were asked regarding problems with events at or near the borders of time granular, this was an issue they where aware of, a possible solution could be double overlapping time granulars
[01/06/2004;01/06/2005] allowing for inclusion of search results otherwise omitted.
Query transformation in cultural metadata.
User contributed descriptive metadata for libraries and cultural institutions.