Some time ago, around the time there was some discussion on why repositories use PDF and not HTML there was this thing called Lemon8-XML announced at the time there was very little information available. I tried to get on the mailing list but nobody got back to me. This week we’re having an invitation-only workshop at USQ, Connecting Repositories to Authoring Tools, and I went back to have another look. Turns out that there is now an FAQ. And a test site which ICE collaborator Ian Barnes knew about.

L8X comes from PKP, same place as the Open Journal Systems (OJS).

The bit of the FAQ I’m interested in is, of course, about using styles.

Q: does it rely on authors choosing styles?

A: not at this point, although we have considered developing (or adopting) a standard set of styles to help provide the document parser more information to work with. We would be glad to hear from anyone willing to help collaborate on a standard style template for authors.

We at the ICE project would be delighted to work with the developers from PKP on a standard style template. We have lots and lots of experience at developing styles andcross platform, easy to use interfaces to apply them. We also have pretty well developed code to go from our generic styles to XHTML. It should be possible to adapt this to target the XML format used by L8X either by writing a whole new converter, or by converting via XHTML-plus-microformats.

I retried registering my interest and I’ve now made contact with MJ Suhonos, who is looking forward to talking more, as am I.

If we were to provide ICE-style templates for OJS then one potential workflow scenario could run like this:

  1. The hard working Journal manager makes a template available using ICE-like styles.

  2. A careless author ignores or misuses the template.

    This is normal for journals. There is typically lots of reformatting to do at the end of the process. I know from the questions people frequently ask about ICE that many people believe that authors cannot be induced to use styles properly, but our experience at USQ is that if you provide a feedback loop so people can see their document converted to HTML, then they do use and even like templates. See my post, Why ICE works for my take on this.

  3. Careless author uploads a document to OJS for the first time.

  4. OJS sends the doc to L8X, which tries to find the structure as best it can.

    L8X has a four-tab interface where you can see how it went:

    1. Extracting metadata

      (another thing we’re working on this week is how to embed metadata in ICE documents so it is easy to extract it reliably if we can get people to use styles then this becomes much easier)

    2. What it thinks the structure of your document is.

      (We find in ICE that showing a table of contents, generated from headings gives people really strong feedback.)

    3. Extracting citations

      (In ICE we’re working on Zotero integration, so that citations are formatted automatically this should make it much easier for L8X to recognize them)

    4. And a preview in HTML and PDF

  5. L8X then round-trips the document back into a word processing doc using ICE styles and passes it back to OJS. The quality of the document has just improved!

  6. OJS uses the styled doc from then on. If the author adds content, or metadata about another author then there is a good chance that they’ll do it correctly, if the template makes it easy. We’ve already got a toolbar that tries as hard as it can to format things using styles, we could extend that to have buttons such as ‘add another author’ just by using autotext.

(I tried out the Lemon8-XML demo site, using the styled Word version of my paper from Ausweb ’06. There were several problems, which you’d expect from alpha software like this I have reported the bugs on the wiki.)


comments powered by Disqus