Lemon8-XML Public Beta Released

[mj]

The PKP Development Team is pleased to announce the public beta release of Lemon8-XML.

This release represents the latest progress from over a year of re-development in collaboration with a number of partner journals, and the first time the source code has been made available openly to the public. The PKP development team is working in close conjunction with a number of beta testing partners to help bring their extensive scholarly publishing experience and feedback into the final Lemon8-XML 1.0 release later this year.

To download the Lemon8-XML beta, see the download page.

[biomat]

I have tried extensively both the June 20, 2008 beta release installed on my server and the online version that seems to be more advanced on your server. I'm sorry to report that I could not get either of them to work satisfactorily with any of my data (manuscripts). I have three major complains about them.

1) The recognition/parsing algorithms do not work well (e.g., I get figures where they're not and no figures where they are, and the reference parsing algorithm doesn't even find year 2004 in "Cieszewski, C.J. 2004. ...", and so on).

2) The available function selections (e.g., radio buttons, list selections, and even text corrections in the parsing of the references) do not stick - don't get recorded.

3) The design of the manuscript data entry is wasteful and repetitive w/o any learning/correcting options.

I wanted to use this system so badly that I wasted several days on trying to figure out ways to make it work by editing the parsing, editing the manuscript, restructuring the manuscript, and other desperate actions. Mostly because of the above complain #2) I believe that neither of the two Lemon8-XML releases are ready to be classified as alpha versions yet. I believe this because alpha version should be good enough to at least execute existing links and commands that in case of such commands as "Save/Update" should at least save what was typed in. The way these systems are right now they're simply not usable or testable. The are merely kind of preview of some of the features of a future software development.

Based on a painful struggle with these "previews" I have the following suggestions for the Lemon8-XML program changes.

Ad. 1) The parsing program should look from the beginning of each reference for 4 digits and use it as the indicator of the year (the 4 digits), the authorship (left to the year) and the title (right to the year at least until the first period). Then look from the end of the reference for digits, periods, and dashes, there will be pages, issues, and volumes. Left to the ending digits is usually the outlet, and then you can use whatever you've got already, but if you do the above you'll have most of my citations already parsed. Perhaps you could have a switch or at least examples of the format you expect, which would help too.

Ad. 2) All the links, buttons, text entries, etc. should be recorded permanently when the Save or Update is clicked on or they should not be there.

Ad. 3) The program should allow for adding manual override definitions, such as LaTeX commands, with an option to use them instead of the try and fail guesswork on the title, authors, etc. For example, it would be much more intelligent if all the metadata were entered right on the manuscript in a database format, such as \Title{...}; \Author{...}; etc., and then parsed into the form rather than I had to re-enter all the metadata into the disposable disappearing form every time I rerun the same manuscript trying to edit it so that finally Lemon8 can gets it almost right. The same goes for figures and tables. The same thing could be done with the sections defined as \section{...}, rather than fighting the errors in recognition of the subsection font of the same size as the regular text, and finally, the same applies to the references that could be entered in an overriding BibTeX format for example.

Some of the suggested above solutions could be made optional (e.g., a switch between "Mouse Potato" vs. "LaTeX Freak";)


Best regards and good luck,

BTW, does anyone know if it is possible and how to convert a LaTeX, PDF, or PS, file into the XML format?

[mj]

Hi biomat,

Thanks for taking your time to provide such detailed comments. We rely primarily on users' feedback to help improve the quality and reliability of our software, so it is greatly appreciated.

What would be most helpful, if you are willing, would be to provide a copy of the articles that you have been working with so we can examine the areas of difficulty directly and address the specific issues you mention. You can contact me via PM or email if you'd like to send them along.

Best,
MJ

[biomat]

Thank you for considering my comments. I have sent you the link to the manuscript. However, many of my comments are manuscript independent; you should consider them independently of any specific examples. For example, the suggestion of adding provisions for database definitions within the parsed manuscript, such as the LaTeX commands for the title, author, etc., and adding examples of what is expected, are general design issues that are independent from manuscript type all together.

Many of my other comments don't need specific examples because they seem to apply even to your sample document that you provided on the Lemon8-XML demo site. For example, the radio batons do not stick for registering publication types no matter what manuscript is used, and the parsing cannot find the year ("1956") in "Sheril, R. D. (1956)", which is the first reference of the example document you provided on your site.

Best regards and good luck,

[mj]

Hi biomat,

I've had a look at the link you provided to your manuscripts, and I think a number of the issues you describe have been addressed in the more recent versions of L8X since the public beta release that seems to be running on your server. Your issue in particular about "sticking" settings has already been fixed as far as I'm aware. Have you tried these same manuscripts on the public demo site or the most recent CVS version? I would really encourage you to keep an eye on the more recent CVS builds of L8X, as it has come a long way since the public beta in terms of reliability and accuracy.

Your comments regarding database definitions for metadata elements reflects an ongoing discussion we have been having, both within the PKP group, and with a number of collaborating research groups. In particular, we are discussing technical details with the ICE-RS research group (led by Dr. Peter Sefton) in ways to integrate both ICE and L8X, and allow better marking and detection of embedded metadata in the document editing process.

As always, we are interested in hearing your overall rationale for using L8X for generating XML, editing documents, and so on. The more feedback we get from users such as yourself, the better we're able to improve the software.

Thanks once again for your feedback,
MJ

[biomat]

I have tried the updated version on your server then and again now. The newest version (no. 1) is much nicer and the progress in visual appearance is impressive. The section partitioning seems to work much better now than it did then; although, there still are some major problems confusing equations for graphs etc. I left the manuscript (the same I sent you earlier) on the demo server so you can check it out for yourself. The two major problems that prevented me from trying Lemon8 any further, past some 15min of trying it, were that the references didn’t parse automatically and it seemed to want me to click each one of them one by one (there are over 100 of them), and that the initial metadata were not parsed automatically – keywords, abstract, title, etc., which I was not going to enter manually again.

If the program does not recognize “keywords: keyword1, keyword2, etc.” I’m not going to use it. Lemon8 did not get the title even thought it is the first line in the manuscript; did not get the authors that follow the title; did not get the abstract following the word “Abstract” right after the authors, and so on.

As per my previous postings on this subject I would not mind to type in the manuscript some commands like “\abstract{begin}” and “\abstract{end}” around the abstract, but I do mind copying and pasting the abstract into a form, which may disappear later, every time I want to try a program. Clicking and waiting on every reference is unreasonable too. For example, I’m processing a manuscript right now with over 250 references and I would not wish even the developers of Lemon8 to be forced to click and wait 250 times on each reference.

I hate to say this because I admire your work and really appreciate all your open source developments (especially OJS) – I think they’re all some of the most significant manifestations of the progress in modern civilization – but I really feel that this whole Lemon8 project is misguided and it is build on wrong premises.

I think that Lemon8, or perhaps Lemon8s, should be built on existing strengths, such as TeX and LaTeX that are towers of strength in publishing technology, rather than on human weaknesses, such as major focus on pleasing with GUI, RAD, and mouse use instead of accuracy, consistency, and decidedness. As it stands Lemon8 development instead of joining the best of open source strength base, such as TeX and LaTeX, seem to have an ambition to be another pseudo-Microsoft product, and it seems to have the same performance characteristics as most Microsoft products do.

[mj]

Hi Chris,

Thanks for taking the time to provide such detailed feedback. We're be sure to consider your comments in our future developments.

MJ

[biomat]

You’re very welcome, and if you ever redirect the Lemon8 development onto TeX/LaTeX based structure I’ll be glad to help. You can see at mcfns.com my examples of the LaTeX/TeX based typesetting that I couldn't get working in Lemon8.