Landing Pages - Joe Hourcle - RDAP12
•Als PPT, PDF herunterladen•
0 gefällt mir•888 views
Data Citation panel Presentation at Research Data Access & Preservation Summit 22 March
Empfohlen
Empfohlen
Weitere ähnliche Inhalte
Was ist angesagt?
Was ist angesagt? (20)
Ähnlich wie Landing Pages - Joe Hourcle - RDAP12
Ähnlich wie Landing Pages - Joe Hourcle - RDAP12 (20)
Mehr von ASIS&T
Mehr von ASIS&T (20)
Kürzlich hochgeladen
Kürzlich hochgeladen (20)
Landing Pages - Joe Hourcle - RDAP12
- 1. Recommendation “Landing Pages” RDAP 2012 Presented 2012-March-22nd, by Joe Hourclé
- 2. this is last-minute filler, as I only found out the day before that one of panel members couldn’t make it, so did this on the plane ... then found out that I had lost a second panel member For those playing along at home, you have to completely forget the order the slides are in while presenting, and already be flustered from having tried to present someone else’s slides but the projector display it really tiny, so you quit Keynote & came back in, which ended up making the slides bigger, but changed the font making them mostly illegible. You may also want to turn on the presenter notes
- 3. Current Practice • Acknowledge the mission or instrument team w/ standardized text • Cite the PI’s paper on the instrument • Cite a paper describing the data • Cite “the data”
- 4. Acknowledgement • Isn’t tracked by most bibliographic tools • Doesn’t tell : • which processed / packaged form of the data • which mirror was used • which version / edition of the data • where to get the data
- 5. Citing the inst. paper • Can’t distinguish between citing the data, and citing something else in the paper. • doesn’t say that the data was useful • Doesn’t tell : • which processed / packaged form of the data • which mirror was used • which version / edition of the data • where to get the data
- 6. Citing the data paper • Are fixed in time • Has the calibration changed since the data was initially released? • Has the data moved between archives? • Has the data been removed?
- 7. Citing the data • How? • Not all data is formally ‘published’ • May not have a formal title / author / etc. • Different needs from different disciplines • May use data from 100 studies • May use a small portion of a large study • Different subsetting needs
- 8. BRDI Meeting, Aug. 2011 • Two days discussing data attribution & citation • Breakouts made lists of the challenges & issues concerning attribution & citation • Technical, Scientific, Institutional/Financial/Legal/ Socio-Cultural, Main Actors & Roles, Add’l info needed to proceed.
- 9. Technical Breakout • Most of the issues came back to identity • How do I decide what to call the data I’m citing? • What are the essential properties when defining a data set? • How do I differentiate between similar data? • (there were more, but I don’t have that notebook with me)
- 10. We need: • Some official ‘record’ of the data • We need an official ‘title’ for the data • We need the data provider to name who the ‘author’ is. (the instrument? the PI team? the software pipeline?) • Something to persist even if the data doesn’t
- 11. Our Proposal • “Landing Pages” • Serve as a publishing record of the data set • Act as an endpoint for citation • Describe the data so that it can be cited • Provide links to the data • Provide context so the data can be used • Persist, yet can be updated
- 12. Our proposal • “Citation Pages” • Created by the article author • Stored with the paper • Link to the “Landing Page” of the data used • Describe additional processing that might have been applied (‘extended methods’)
- 13. The Challenge • Creating records for all of the data out there • Minting DOIs for them • Making sure they’re maintained & preserved • Getting authors to use them for citation • Can work with journal editors
- 14. Solar Dynamics Observatory (SDO) Atmospheric Imaging Assembly (AIA) 171Ångstrom ; 2010/07/08 17:45:48UT ; 2x2 binned
- 15. Poster, Handout & Links: http://docs.virtualsolar.org/wiki/C i t a t i o n
Hinweis der Redaktion
- Yes, I know, I didn ’ t put my name on the slides ... it was a rush job. I probably should have said ‘ recommendation FOR landing pages ’ (see next slide). This was presented on 2012-March-22nd, by Joe Hourclé It ’ s discussing the poster (see link at the end) which has co-authors (Wo Chang, Franciel Linares, Bruce Wilson, Giri Palanisamy), but they didn ’ t have a chance to review any of this before I presented (see next slide). I also have some stuff in there that came more from the original discussion that didn ’ t make it into the poster, so I should also acknowledge the other people involved in the initial discussion : Paul Groth, Allen Renear, Herbert van de Sompel, and Martie Van Deventer. And part of the notes in here are expanding on questions that people had at the RDAP meeting … so, for the notes portion, I should acknowledge them, but I can ’ t remember who they all were. I remember Karen Wickett, but we ’ ll have to wait for the video to check who else raised issues.
- This slide was *not* in there when I presented at RDAP. All of the presenter notes were typed in *after* the presentation, so I ’ ve taken the opportunity to try to make things a little clearer than when I presented. I ’ ve left those slides untouched, and inserted this slide, and two more at the end.
- And it likely would ’ ve helped if I had set the context up correctly ... however, I jumped straight to the material in what ’ s now slide #8, talking about the meeting where these recommendations came from. Basically, we had a group to try to make a list of the issues of the technical challenges to good data citation. So, first, the quick overview of the current practices in different fields.
- An example: “ SOHO is a project of international cooperation between ESA and NASA ” ... of course, there ’ s multiple instruments on the spacecraft, and *lots* of different data products. And it ’ s been running for 15 years and the data won ’ t be made ‘ final ’ until 1-2 years *after* the mission ends (which could be years if they keep getting funding to continue LASCO operations). And the SOHO ack. text has changed over time ... it used to be that you cited SOHO and the specific instrument (with their PI institutions). But that could mean that you ’ re mentioning 10+ instruments, so that faded out over time. And ‘ which mirror was used ’ is referring to when there ’ s copies of the data in more than one place … there ’ s *also* the issue of instruments with multiple optical pathways through them (eg, Hinode ’ s SOT telescope), but I lump all of that into ‘ observing modes ’
- Inst. == Instrument. Ie, ‘ sensor ’ , ‘ detector ’ … whatever your field calls it. For instance, you might cite the paper because you ’ re interested in how the mirrors in a telescope were manufactured or the shielding of the electronics ... we have no way of differentiating between you ’ re using the data vs. you ’ re interested in the design of the instrument. And, as was explained in the Dryad Data talk, the ‘ authors ’ (creators) for the data can be a different group for the research … There are going to be people involved with Phase E (the actual observing portion portion of the mission) that weren ’ t there for the construction of the instrument with in the case of launch delays might ’ ve been years earlier.
- The thing about initial calibrations is ... they ’ re always wrong. They might be pretty close, but there ’ s always going to be *something* that we learn over time. For instance, did you know that if you point a telescope at the sun for years, it degrades? If you search for ‘ EIT calibration ’ , you get : http://umbra.nascom.nasa.gov/eit/eit_guide/calibration.htm But it wasn ’ t the PI team who came up with the best model of how the instrument degrades, it was another scientist : http://umbra.nascom.nasa.gov/eit/eit_guide/offpoint.htm … Should he be the one getting all of the citations whenever someone uses the re-calibrated data, or the original PI team? Or both?
- Much of the data that is ‘ served ’ in the solar physics community is a bunch of files dumped on an FTP server or website. They ’ re FITS files, so they ’ re self-documenting, but because the instrument might be changing observing modes, it ’ s hard to define specific ‘ collections ’ that have titles. Just naming the instrument (or spacecraft, as shown in the SOHO case), is rarely useful to actually determine what data was used in the research. Each discipline has different needs for describing how they reduced the data before analysis -- we only used 30 minutes of a multi-year mission; we only used one measurement per day (vs. averaging each day ’ s observations); We only used the strip right in the middle of each image to generate an alternate projection of the sun before analysis; we rebinned the image as our analysis software was written for 1024x1024 images); we only looked at a smaller region, not the whole sun. Or you might ’ ve done more than one of those. Trying to cram all of that into a citation gets really, really complex.. And that ’ s just for solar physics. Other disciplines use other techniques (eg, the polar folks may only use observations in January, to remove seasonal variations.
- BRDI meeting program w/ links to presentations is at: http://sites.nationalacademies.org/PGA/brdi/PGA_064019
- I should ’ ve mentioned that we were trying to look at the technical barriers to citation. Part of the issue was that we weren ’ t sure if it was the researchers or the data provider ’ s responsibility on generating the citations. We agreed that there were lots of ways to cut things up, but that the data provider should provide a description of what they consider the ‘ collections ’ to be, so that people have *something* to reference.
- Basically, if you want people to cite your data in a specific way, you need to tell them how to cite it. Don ’ t let them pick the ‘ author ’ , or the ‘ title ’ , as there could be infinite variation. If you want to glob lots of stuff under one title (a ‘ collected works of Shakespeare ’ type situation) so you don ’ t have to do as much work, you can do it … but it ’ s not nearly as useful as being able to cite one of the individual parts, and there ’ s an economy of scale … so making descriptions of 10 individual parts isn ’ t going to be 10x the work of describing them as 1 collection.
- Leaving the citation generation up to the individual researcher won ’ t work ... but also, giving a specific ‘ you must use this format ’ isn ’ t as useful, because the author may not be using APA or whatever style guide your discipline ’ s journal uses. (this is where DataCite is useful) There ’ s no reason why we can ’ t describe the data *once* using DataCite, DublinCore-SAM (once that profile is done) or similar, and present it in BibTex or other formats for citation manager software to use. The search engines could even spit out a little report when you download data, as a sort of ‘ receipt ’ , telling you what you ’ ve just downloaded, and how to cite it. (which could provide the basis for the ‘ citation page ’ (next slide) We also don ’ t specifically define the granularity of the collection described -- it could be an individual file; all data for the life of the investigation; broken out by year; divided up by observing mode ... or you could do multiple, as there ’ s nothing to say that a file can ’ t be a member of more than one collection; the provider can define whatever groupings make sense for their data. And they can add new ‘ virtual ’ collections later, if they find lots of people are repeatedly citing one specific grouping of data. The landing pages should also be machine harvestable -- XML w/ XSLT transforms to make it XHTML, RDFa, HTML+microformats, HTTP content negotiation, links to OAI-ORE links, etc. If you have lots (but not TBs) of files to download, it can link to OAI-ORE, MetaLink, sparse bags (BagIt) or similar to automate downloading to those files. We talked about lots of different technical ways of doing it, but we ’ re likely going to need to test them and see which work best for which fields. ... I could keep typing, but it might make more sense to just go read the handout. (see the last slide for a link)
- If the journal publisher won ’ t take it, then in an Institutional Repository, or even on the author ’ s website is better than nothing. As the data provenance models mature (eg, the work at globalchange.gov), we may be able to make the processing description machine readable. In those cases where you can store the data with the publication (eg, the data ’ s small enough), you can do it, but you still want to describe how it came to be for reproducability. The citation page can also be used for ‘ enhanced publications ’ or ‘ interactive data pages ’ , to allow readers of the paper to quickly visualize the data, change the scaling or field of view of a plot; reorder or filter a tabular data, etc. The citation page may just link to a single published dataset, or it may aggregate hundreds of datasets. If you only have a few (<5) sources of data, consider citing them individually rather than aggregated, even if you have limits as to how many citations you ’ re allowed, as without the data, you wouldn ’ t have been able to do the research.
- If we ’ re going to do this efficiently, we likely need to do it by discipline -- each data archive describing the data would be nice, but it ’ s a lot of time reading documentation, interpreting what each field/attribute means for their data, etc. We get an efficiency of scale working by discipline. We ’ d likely need at least two people -- (1) a person experienced with metadata / cataloging, and (2) a discipline scientist who can verify the work, and possibly write missing documentation. New post docs often write the best documentation, as they come at it fresh, having to research things, rather than just assuming ‘ everyone ’ knows about that little quirk that ’ s not worth mentioning to someone in the field, but would likely not be known by people from other disciplines.
- This slide was *not* in there when I presented at RDAP. I usually acknowledge the image I ’ m using in the background ... even had a master slide so I could insert it easily ... but I forgot to put it in. And this time, I ’ ve linked it to the mocked up ‘ landing page ’ for SDO/AIA Lev1 171 Angstrom images. And wow, converting from Keynote to PowerPoint really made this slide ugly due to the link in it. (which I should probably go and modify slightly, to explain that it ’ s a mockup, now that I ’ m giving out the URL to it)
- This slide was *not* in there when I presented at RDAP. The URL goes to the Virtual Solar Observatory ’ s wiki. It has links to all of the references mentioned in the poster & handouts ... and I should probably add ‘ MetaLink ’ and ‘ BagIt ’ and other stuff … but if you follow that link, also follow the ESIP wiki link … they have lots of other references, as do most of the other ones (but those have the same ‘ fixed in time ’ issue as citing a paper.