Is NASA ADS usefull?

[hubie]

Now that I have it working, if I go to Tools I can see under Installation Details that I do indeed need PHP version 5.2.0 or greater.

[bubblehead74]

I went to try it out and I found that it wouldn't work because there is the call to sys_get_temp_dir, which apparently is only good for PHP versions >= 5.2.1.

If you can't upgrade your PHP, this should be easy to fix. Edit data.php - insert this at the end:

$temp_dir = 'path/to/your/temp';

Unfortunately, Centos 5 runs 5.1.6. I'm not sure what version Ubuntu has in the repos.

Ubuntu definitely has 5.2.something, because I tested I, Librarian with an 8.10. I am not sure at this moment what other functions won't work. Data filtering for sure.

I also had to add include_once 'functions.php' to data.php. With those changes it seems to be running, and I will try it out with our company to see how well it works.

I don't understand why this needed to be done. It worries me a little bit.

[hubie]

First off, you are correct about my install using PHP 5.1; data filtering does not appear to work. No matter what I add or search for, text doesn't show up in the right panel. Oh well. Maybe I'll see whether I can either upgrade PHP, or perhaps install a second copy and run a second apache on a different port.

EDIT: Odd: I didn't see anything in the right-hand frame until I logged out, then suddenly the entry for the paper appeared in the right hand frame.


I've also tried out your demo. I added a paper from arXiv, which seemed to go in just fine. I also tried to add a pdf manually and I think I've found a problem. Just to prove that if you don't keep a user from doing something stupid he'll find a way to mess up, the paper I added did not have any fields filled in (I thought that perhaps the info would be pulled from the paper). The paper that ended up in the library has no title, and so there is nothing to click on to bring up the record to edit and/or delete it.

I'll keep playing around with it. Just using the demo has motivated me to upgrade PHP so that I can install my own copy.

[bubblehead74]

First off, you are correct about my install using PHP 5.1; data filtering does not appear to work. No matter what I add or search for, text doesn't show up in the right panel.

That is not due to data filtering. I actually checked, and I removed all instances of this function due to incompatibility with PHP 5.1. So this problem is something else. I only tested I, Librarian on Ubuntu and Fedora with 5.2.8 and 5.2.4 respectively. Unfortunately, I don't have resources to check all Linux distros.

I also tried to add a pdf manually and I think I've found a problem. Just to prove that if you don't keep a user from doing something stupid he'll find a way to mess up, the paper I added did not have any fields filled in (I thought that perhaps the info would be pulled from the paper). The paper that ended up in the library has no title, and so there is nothing to click on to bring up the record to edit and/or delete it.

Indeed, I, Librarian will attempt to extract DOI from a PDF. Then it will search PubMed and fetch all metadata. In the future, I will add fetching metadata from NASA ADS, plus extracting arXiv ID and fetching metadata from arXiv.

I have tried really hard, but I cannot reproduce this bug. It is definitely a bug, because under no circumstances should a user be able to record an item without a title. If you could reproduce it, please, let me know step by step how you did it.

[hubie]

I am able to reproduce the bug, but as I warned you before, I had to work hard to do what I needed to do to mess it up.

I first went to Add Record and selected Manual Upload. I chose my pdf file then didn't fill in any fields because I was curious whether it would pull out the information. I hit record, it looked like it took the file (because when it came back, it seemed to have assigned the file a number for its name along with the pdf extension). I went back to the Library, and nothing was there. I tried it again, and I got a warning about there being a duplicate file, but still nothing showed up in the library.

The next thing I tried was to do "Import from File". I assumed this meant that this was what one needed to do to import the information from the file, so I hit Browse and chose the pdf file again. The particular article I used (from the journal Optik), and I can send it to you if you need it, had the banner across the top "ScienceDirect," so naturally I chose "RIS" for the import option because it mentions Science Direct. The only box checked was the default "Add to Shelf" and I hit "Import". This is what added the empty entry to the library, and I was able to reproduce it on the demo as well.

I do have a question: was Manual Import the proper choice originally? I understand why it wouldn't have found the info for any of the fields because it was a journal article that isn't in PubMed, but it seems to have imported the file but it doesn't show up anywhere that I can find like the files do when I do a PDF Batch Import, where they are given names that start with "Unknown file".

My other question, and this goes to the source of the bug, is, is the Import from File option for importing only the journal metadata, i.e., I assume I was not supposed to have fed it a pdf file?

[bubblehead74]

I first went to Add Record and selected Manual Upload. I chose my pdf file then didn't fill in any fields because I was curious whether it would pull out the information. I hit record, it looked like it took the file (because when it came back, it seemed to have assigned the file a number for its name along with the pdf extension).

That's correct. I, Librarian will attempt to extract DOI from the PDF and search for the corresponding metadata in PubMed (NASA ADS and arXiv coming soon). If nothing is found, PDF is conveniently preloaded, you only need to manually enter the title plus as much info as you want and hit Record in order to finish the process. You can also paste DOI, and hit Record, that will also look for the metadata, but in your case it won't work, because the record is not stored in PubMed.

The next thing I tried was to do "Import from File". I assumed this meant that this was what one needed to do to import the information from the file, so I hit Browse and chose the pdf file again. The particular article I used (from the journal Optik), and I can send it to you if you need it, had the banner across the top "ScienceDirect," so naturally I chose "RIS" for the import option because it mentions Science Direct. The only box checked was the default "Add to Shelf" and I hit "Import". This is what added the empty entry to the library, and I was able to reproduce it on the demo as well.

Aha. Yes. This feature is meant to import reference metadata from files of these formats: RIS, Endnote XML and ISI Export File. I really need to start writing tutorials. I commited a fix; the Import File will no longer try to parse PDF. This should avoid this problem.

I do have a question: was Manual Import the proper choice originally?

Yes. See the explanation above.

...when I do a PDF Batch Import, where they [PDF files] are given names that start with "Unknown file".

This is because it works the same way. PDF Batch Import will attempt to extract a DOI, and fetch the metadata from PubMed (hopefully NASA ADS and arXiv will follow soon). If nothing is found, you can choose to record the PDF as unknown, or not to record it. I admit the usefulness of I, Librarian for folks not using PubMed is limited right now. But I am working on including NASA ADS and arXiv in order to change that.

[bubblehead74]

Just to explain better. There are three ways to populate I, Librarian.

1. You can start with PDF files (Manual Upload or Batch Import). The current limit is that it is not useful for literature outside biomedicine.

2. You can use Import from File to batch upload metadata from RIS file. Almost every online database allows saving references in RIS. Like you mentioned Science Direct: (i) Search for the references of your interest, (ii) check them, (iii) click Export Citations, (iv) choose Citations and Abstracts and RIS format, (v) save the file on your hard disk, and (vi) upload the file using Import from file. Import the corresponding PDF files using the Edit window (click on the record title, to get there).

3. Third way is to seamlessly import references from PubMed, PubMed Central and soon from NASA ADS and arXiv. In case of PubMed Central and arXiv, you will be able to seamlessly import PDF files as well.

[hubie]

In my first case, where I loaded the file without filling in any fields, the file seemed to have been preloaded, but I cannot see the file in the Library or shelf so that I can go back an edit the entry.

Obviously the way to avoid this is to put something in the fields from the start, such as the DOI as you suggested, but how do I get access to the file now that it is preloaded? The clue to me should have been that after the file was preloaded and it didn't fill in any information into the data fields, that is when I should have added some kind of additional information.

[bubblehead74]

I loaded the file without filling in any fields, the file seemed to have been preloaded, but I cannot see the file in the Library or shelf so that I can go back an edit the entry. ...how do I get access to the file now that it is preloaded?

When you upload a PDF without any accompanying metadata, the file is uploaded to the server and is sitting in temp directory and waiting for your decision. You can click on the link which says ddd.pdf in Manual Upload form to download the file back, if that is what you mean by access. At this stage the PDF is not yet recorded in the collection. If you want to record it, you have to enter some title and hit Record again. If you want to abandon it, click on Manual Upload again. The file will be cleaned from the temp directory later.