VBPro is back in circulation

I’m happy to announce that after a gap of several months VBPro will again be available for download. I have taken over the task of distributing VBPro from Mark Miller. Thanks to Mark for making this possible.

Mark and I continue to work on the details of opening VBPro’s source code, so this distribution is of binaries (executables) only. Nevertheless it’s great to get the program back into circulation.

The app at APSA

This year the American Political Science Association meeting was in Philadelphia. The Yoshikoder went too, under cover of a Content Analysis working group organized by the irrepressible Stephen Purpura.

Attendees were asked for a short presentation, and a brief document covering either the content analysis methods they were developing, or the research problem they thought might benefit. I took the opportunity to write 4 pages on the Yoshikoder. You might find them useful. In particular, there’s motivation for, description of, and a guide to interpreting the relative risk ratios that the new statistical comparison report provides.

There’s plenty to be said about the other methods we heard about at the workshop, but I’ll save those for another entry.

Hosting changes

The Yoshikoder currently has its tray tables stowed and its seats in the upright position, ready for the move to sourceforge, an open-source project hosting site. You shouldn’t feel much difference: hopefully not much more than a change in the process for downloading new versions.

Famous last words…

Laver and Garry dictionary update

Some of you may be using Laver and Garry’s policy positions dictionary, available from the resources page. This is a dictionary of english-language political terms designed to capture issue content in party manifestos. After some investigation it appears there were actually two different versions of this dictionary floating about the academic end of the web. And it seems that the one I translated for the Yoshikoder was not the one used in the final article.

Tush.

I’ve unlinked the old one and added LaverGarryAJPS.ykd to the page. Thanks to John Garry for sorting out which was which. Practically speaking, they’ve much the same content, but the real one is smaller.

Tying up loose ends on the preview

Although it may not look like it, there’s a fresh preview of 0.6.3 for download. This should fix two issues:

In the first, folk were running out of memory with large projects. The new preview allows, but does not require, 256M of RAM to be used by the program. This is four times as much as before, but comfortable in the context of most machines’ specifications I hope.

The second issue is the notorious ‘Already running’ problem that usually appeared after a sudden program shutdown. That shouldn’t happen now.

Yoshikoder 0.6.3 – a preview

A preview release of version 0.6.3 of the Yoshikoder is available from the home page. I described the changes in a previous post a while back.

Remember that this is only a preview: online help is not up to date, and I’m sure there will be some other things to iron out. Nevertheless statistical document comparisons are ready. And you saw them here first…

Pesky drive names (Version 0.2.1)

There's a new release of the YKConverter that fixes a rather nasty bug for Windows users trying to convert webpages. It seems the old C:\ shuffle tripped me up. The moral of this bug is: do not construct local URLs by hand, use file.toURI().toURL() instead.

This release also attempts to remove <!— thing —> sections that the html parser tends to leave in. These are usually useful but deeply uninteresting chunks of javascript. The removal code might not always work, but it might save you some post-editing.

The new release is available from the converter's homepage.

YKConverter for Windows

Just a quick note to say that the YKConverter is now available as an executable for Windows.

The application default is to save all converted documents to a particular directory, shown in the preferences. If you’d prefer each file to be saved as text next to the original, switch that in the preferences too.

Sometimes the converter will fail to convert PDF documents. Most times this is because they have been locked by the author. Locking a PDF document means that no text can be extracted, either manually or automatically. Unfortunately, there’s not much I can do about that.

In the pipeline

Just got back from visiting to the friendly folk at Penn State’s Political Science department. Yoshikoder figured in my day course on text analysis.

Anyway, as a result of conversations in the lab, I’ve added some more stuff. When it’s finished a round of testing, it will be part of release 0.6.3. In the meantime here’s the list:

• Unified frequency report – Word frequencies for all the selected documents in the same table
• Multiple document addition – Select a large number of documents and import them all at once
• Statistical document comparison – How much more of each category is there in document A than in document B, with confidence intervals

There is one more goody in the collection, but that will get a post all of it own.

If there’s something you’d particularly like to see, let me know.

VBPro goes open-source

Mark Miller has decided to make VBPro open-source, under the Gnu Public License. VBPro is a classic computer content analysis program and a de facto standard in the field, so it’s great for the scientific replication standard to have the algorithms available. But it’s also great to have a code base that can be updated (VBPro runs only in DOS), enhanced, and ported to other operating systems.

VBPro was always free, but I’m proud to have been part of the crowd presenting the arguments for making it open too. And I’ll be helping Mark sort of the licensing and hosting arrangments for VBPro’s new life. Frankly, I’m pretty excited about the whole thing.