kapuska
Reged: 06/26/08
Posts: 15
|
|
I'm considering moving from Zotero (1.0.6) to Sente for the sake of citation support in Pages '08. Most of the references in my Zotero library are linked to PDFs on my hard drive, but those links are destroyed in the process of importing the library to Sente (Zotero: export references as RIS; Sente: File > Import... > foo.ris). Is there a way to repair those links automatically?
I've tried Edit > Perform > Locate full text locally using Spotlight, but this is really slow, often fails, and so far as I can tell has to be done by hand for each reference. Oof. (There's no sign of the Sente Applescript library in my Script Editor; would that appear if I bought a license?) It looks like there are two other alternatives in theory---either drag the appropriate PDF onto a reference, or use File > Import and let Sente find the citation info. Neither route works consistently for me. A repetitive stress injury rules out *any* amount of drag-and-drop work (let alone ~1000 repetitive drags-and-drops). The citation-finding, while often impressive, is sometimes completely wrong, with no chance for the user to veto the purported match (e.g., I've got a paper by X, which I'd like to link to the existing reference to X, but which Sente determines is actually the completely unrelated Y---and before I can say otherwise, the Y reference has been imported and linked to the X PDF). Even when Sente does find the right reference, it imports a new copy of the reference rather than linking the PDF to the reference already present in my library.
It seems to me that a few basic functionalities are just missing:
- batch processing of imported PDFs;
- keyboard alternatives to drag-and-drop actions; and
- an extra item in the contextual menu that comes up when you click on a reference in your library: "Link to PDF...".
I want very much to like Sente. The idea is great, the look is great, the incorporation of Safari ought to be incredibly powerful, the Pages support is key... please tell me it's not going to take days of dedicated and literally painful work to start using it in earnest.
|
TomEck
Reged: 05/11/08
Posts: 129
Loc: Germany
|
|
I am not sure about this but I doubt that the RIS format includes the (local) path (link) to the PDF... Is there no other export alternative? I have no problems with the EndNote XML format for example... However, I might be mistaken..
|
kapuska
Reged: 06/26/08
Posts: 15
|
|
Zotero's export options are
Zotero RDF
MODS
BibTeX
RIS
Refer/BibIX
Unqualified Dublin Core RDF
Wikipedia Citation Templates
of which Sente appears to be able to import BibTeX, Refer , and RIS. I tried BibTeX, but for some reason Sente couldn't handle the file. The Refer file imports okay, but as with RIS the links to PDF break en route.
|
Michael
Reged: 02/17/05
Posts: 978
Loc: Denver, Colorado
|
|
Please send us a sample RIS file to our support address and we will look at this issue.
Michael
--------------------
Michael
Third Street Software
|
silvo
Reged: 02/28/05
Posts: 389
|
|
I'm not sure whether this would help you much with regards with the stress injury and with the speed....
If you import all your references to a fresh library, you could drop the entire pdf collection (well, maybe in 50-pdf batches) and Sente will go through each of them one after the other automatically. If you are lucky (meaning that Sente is able to match each pdf to its own reference without trying to look online), you should be able to do it using only the keyboard.
best,
silvo
|
kapuska
Reged: 06/26/08
Posts: 15
|
|
Thanks for the suggestion, Silvo. That's certainly an improvement as far as minimizing mousing, but I don't see how to get Sente to look in my refs collection first and not go online---it keeps harvesting fresh (redundant) references and linking my PDFs to them. Any suggestions? Is there a setting I can change somewhere?
|
Michael
Reged: 02/17/05
Posts: 978
Loc: Denver, Colorado
|
|
When importing PDFs, Sente always looks first in your library for a match. If Sente is adding new references, it is because your existing references do not match the incoming reference in some important way. For example, if your existing reference does not contain a DOI, Sente will have to look elsewhere for the citation data. Then, it may find the citation information on a site that uses a different abbreviation for the journal name than the reference in your current library, so the duplicate detection code is not sure that the two references are really the same, so it creates a new reference.
It would be good to first check out the original data import problem, so please send us the sample RIS files and we will look to see if our importer is doing something incorrect with them.
Michael
--------------------
Michael
Third Street Software
|
silvo
Reged: 02/28/05
Posts: 389
|
|
Feature request: could you allow the user to force looking just the library?
silvo
|
kapuska
Reged: 06/26/08
Posts: 15
|
|
Michael, I sent the RIS file a little while ago (just got confirmation from TSS). Thanks for looking into this so quickly.
There are definitely some cases where there's a mismatch between reference versions---I haven't done an exhaustive comparison, but there seems to be a problem with papers with advance publication of accepted manuscripts. I frequently grab abstracts and refs at that stage but wait to get a PDF until the final, typeset version is available, so the apparent publication date can change (but shouldn't the DOI stay the same?).
I'm baffled by cases (there have been many) where Sente is able to pull the correct DOI from a PDF but then finds and suggests several mismatches in Google Scholar. If a unique identifier is neither unique nor identifying, the DOI system is in trouble...
|
Michael
Reged: 02/17/05
Posts: 978
Loc: Denver, Colorado
|
|
Thanks for sending the file. I have followed up directly to you with a more detailed response, but for those watching at home, here are the key points:
-- Zotero is not exporting the attachment paths in the RIS file, so Sente cannot import them. Too bad, because if this worked we would be done.
-- Zotero is exporting DOIs in a field called M3 (for "Miscellaneous 3"). Sente is looking for DOIs in a field called "DO". A search and replace on the file, changing "M3 -" to "DO -" and then importing the resulting file, gives records with DOIs in Sente.
Importing PDFs will then be able to match existing references using the DOIs. In this case, this covers about 50% of the records. For the others (those without DOIs) the reference found externally during the PDF import does not always match the imported reference (e.g., due to differences in journal name abbreviations), so this will still result in some "duplicate" records. Not great, but this is about as well as Sente is able to do at this time.
Michael
--------------------
Michael
Third Street Software
|
kapuska
Reged: 06/26/08
Posts: 15
|
|
I just imported the modified RIS file you sent, Michael (and responded over email as below). The import seemed to go without a hitch, Sente found (obviously) lots of duplicates, and a post-import dialog indicated that, for any pair of duplicate refs, any information that had been missing from the existing copy had been harvested from the new copy. Perfect, just the behavior I'd want to see. But I've just had a look through the updated library, and not one of the references appears to have a DOI.
I then imported the modified RIS into my temporary library, and the DOIs appeared correctly. *Then* when I copied the temp library to my library, the DOIs came along too, supplementing the existing references as expected. The workaround involved exactly one extra action, so not a big deal, but I wouldn't have thought that step should be necessary.
I've tested a couple of imports; some are smooth, some not. E.g., in one case the newly updated ref included the DOI jxb/ern091 and the paper was tagged with DOI 10.1093/jxb/ern091, so they weren't recognized as being the same and a new copy of the reference was imported. (No idea whether the truncated DOI my ref had was Zotero's fault, or the journal's, or what, but there it is, and I'm sure it's not a unique issue in my library.) It would be really, really great if there were a chance for the user to intervene before that happens. I know there's a check-box in the Quick Add dialog for "Automatically link to obvious match" or something, and that it defaults to being checked, but when you're only importing one PDF and you have a reasonable internet connection, that option has barely appeared onscreen before the whole dialog disappears. The feature that Silvo suggested on the forum, forcing Sente to look in the library and ask permission before it ventures out to Google Scholar, would be a really nice option to have.
Thanks so much for your quick response, Michael. There may still be some kinks to work out, but I feel a lot better about diving into this project when I know TSS is paying this much attention to its users.
|
Previous
Search Results
Next
Threaded
Edit
Reply
Quote
