Reading on Papers

on

I am trying to understand the capabilities of existing iPad applications with respect to active reading. In this spirit, I have reviewed iAnnotate, and have written about e-books in general. Mekentosj Papers is a Mac application for managing academic papers; a version of it has been ported to the iPad. The idea is that you can use it to find papers you need to read, read them, and also manage their re-finding. The app fails on all accounts.

Finding papers

Papers integrates several search engines, including Google Scholar, the ACM Digital Library, arXiv, PubMed, and several others, and allows you to decide which ones are available by default. Unfortunately, you cannot search more than one collection with a single query, and the only way to switch search engines is to clear your query! There is no history of queries either in the app (for review or re-execution), or even within the search box control. Search results are presented in a narrow floating dialog that the iPad UI uses for lists, but not enough space is devoted to each result: the titles of many papers do not fit, only the first author’s name is shown, and the titles of the journals or conferences are so contracted as to make them unrecognizable. (For example, most proceedings retrieved by Google Scholar showed up as ‘Proceedings of the….’ followed by the year; ACM DL results were more informative.)  There is no way to sort the results by publication date or by author.

Tapping on a paper opens the web page associated with  that paper, and shows some of the metadata in the floating dialog. Unfortunately, the name of the publication venue is still not fully visible in this view, and not all authors are identified. Tapping the “Import” button causes the plus sign to change to a check-box, but the paper itself may not appear. This seemed to be the case for publications that weren’t open-access: hits on some journals showed the abstract page, but no link to a PDF; the app seemed unaware that a login was required to download the article.

When downloading files from the ACM DL, the dialog showing the title, authors, and abstract is overlaid by another, modal, dialog asking whether the PDF should be added to the library. Unfortunately, this second dialog obscures the abstract, prevents the loading of the associated HTML page, and prevents scrolling through the shown abstract.  Pressing the “add” button fails unless you have a site license based on your IP address. When accessing from home, with my personal DL account, I have to log in to authenticate myself to the DL. I can do this by clearing the dialog, tapping on the area where the web page should display (which causes it to actually display), then tapping the PDF link. This again prompts me to “Add PDF to Library,” but again fails because I am still not authenticated. On this second attempt, however, the login page is shown. Now when I enter my userid and password, the login process begins. This is convoluted and mildly annoying. What’s considerably more annoying is that it does not remember the credentials for the ACM login, forcing me to go through this procedure for every paper!

Reading

Having downloaded the paper, one can flip through its pages to read it, and use the standard pinch gestures to zoom in and out. The next page is displayed by swiping sideways, but it may take one the order of 1-2 seconds to load. The application seems to cache adjacent page images, but flipping through the paper rapidly (to get to the references, for example) causes the cache to be emptied, and requires the rather prolonged refresh.

No marking or highlighting is possible, and notes may only be associated with the entire document, not with a particular page or passage. Nor does it appear possible to mark or save passages from a paper. These limitations make it unsuitable for active reading.

Re-finding

It is possible to search the set of downloaded papers by title, author, publication venue, or all of the above. It is not possible to do full-text search on the paper contents, or on the notes that you recorded for the papers. There app does provide a browsable index by author or by publication venue, athough ‘et al’ appears in the index as well. One useful missing feature is a count of the number of papers by a particular author in your collection. This kind of metadata could be useful for exploring a largish set of papers.

When browsing by author, I followed a link to a paper, was shown its metadata in that skinny vertical dialog, but could not find a way to get to the PDF short of loading the CiteSeerX page, clicking on the PDF link, getting prompted to download, and then canceling the request. Thus the author-based navigation seems to be tied to the HTML page rather than to the (already downloaded) PDF document.

I could not figure out how to use the Papers folder feature — it seemed like it should be possible to create folders and put papers in them, but I was not able to figure out how to create a new folder.

Comparison with iAnnotate

As I was using Papers, I kept comparing it to the iAnnotate app.  Papers has integrated search, while iAnnotate relies on an embedded browser. The extra bit of structure provided by Papers doesn’t seem to be that useful, particularly given its built-in amnesia. iAnnotate not only allows the use arbitrary search sites available through the browser (I use it to pull attachments out of my e-mail, for example), but also remembers log-in credentials, making that interaction considerably smoother. One thing that would make iAnnotate even better would be a way to save bookmarks of frequently-visited sites.

In terms of active reading, there is no contest: iAnnotate supports it (albeit imperfectly), Papers doesn’t.

For re-finding, there is a bit more parity between the two tools: both allow to browse the collection, and to search by title. Whereas Papers also has author and publication venue search, iAnnotate has full-text search of the papers. The latter seems more useful in most cases, although sometimes being able to browse by author is also desirable. iAnnotate also has the ability to search over the tags associated with a paper, but, apparently, not over the notes attached to specific passages.

In short, while Papers is much better than reading PDFs from a web browser,  iAnnotate is by far superior to Papers for finding and reading papers.  Although Papers has a couple of useful features, their utility is not sufficient given the functionality it lacks. And to top it off, it costs 50% more!

2 Comments

  1. Jon EP says:

    Thanks for an important post. To my mind, a couple of key things must be addressed for active reading to become a reality on ipad and similar tablets, and to be integrated into a truly useful workflow from tablet to desktop computer (necessary for writing research papers, etc.):

    1. Upload from and synch back to a file server are necessary. Once you’ve made annotations, notes, etc., those must be reflected in and updated on the ‘base’ pdf file, so that accessing the same file from your computer or laptop reflects all the work you’ve done marking it up.

    2. Users need to be able to apply tags to individual portions of text, not to whole documents. There is little use in tagging a book-length PDF with a tag–what’s needed is the ability to put pieces of text from different sources into conversation with each other–going through a book like Moby Dick and finding a few key uses of a certain narrative strategy, and then linking them to other examples in other books and to sections of critical articles that deal with the issue.

    3. The user interface that provides access to one’s collection of materials needs to provide access to those tags (a hierarchical tree structure), and needs to allow for comparison of previews of tagged material across documents.

    4. Integration of all of this with reference management software, preferably as part of the same package.

    I-annotate is part-way there, Papers has important features, Zotero also has a lot of what’s necessary, as does Mendeley. Wrapping all of them together, and then using some of the features found only in QSR’s NVIVO — now that would be a software package!

  2. Jon, I agree that we’ve seen a bunch of tools tackle bits and pieces of the larger set that is needed to really move active reading online, but that nobody has managed to produce an integrated solution. Seems like an opportunity, although not an easy one to capitalize on. I wonder what subset of these features would need to be integrated (and done well) to reach the tipping point.

Comments are closed.