User talk:PJvanMill

This is the en-Wikipedia talk page of PJvanMill. If you want to talk to me, you are in the right place (at least, if you want to talk in public about something related to Wikipedia or its sister projects).

Please make a new section at the bottom for each new conversation, and end your posts with ~~~~.

If you post here, I will reply here and ping you.

If you think I did something wrong, I would appreciate it if you phrase it as a "Why"-question rather than an accusation. I hope you can see that I am here to make WP better. If you've come here while angry at me, you should write something and save it on your machine; then come back tomorrow and take another look at it before posting it.

I will archive a discussion when I think it is dead - that is, when it seems the conversation is over. The archives are located under User talk:PJvanMill/archives. I may have archived a conversation that wasn't quite as dead as I thought it was. If you wish to revive a discussion, copy it to here and add your comment, with the edit summary reviving "<discussion name>" from <archive>; then remove it from the archive, with the edit summary reviving "<discussion name>".

---

Favors

Do you remember a few years back when you extracted a book from archive.org for me?

Do you think you could do that for a few other similar books? DS (talk) 14:31, 3 March 2024 (UTC)

@DragonflySixtyseven: A bit of a late response, I'm only seeing this now as I haven't been active lately. Yes, I certainly could; I don't have much time for Wikipedia at the moment, but I could probably do roughly 1 book each weekend? Kind regards from PJvanMill)talk( 23:38, 3 May 2024 (UTC)

That's very kind of you. Finding suitably-formatted texts isn't as easy as it could be. Here's one - "Who's Who in the world", 1912 edition. If you aren't comfortable with putting it in my userspace, create it in your own and let me know when it's ready, and I'll move it over myself. DS (talk) 13:16, 4 May 2024 (UTC)

@DragonflySixtyseven This book seems intractable for the moment. The OCR'ed text given by the IA is of quite a bad quality. To give you an idea, in just the first few names the OCR reads "William:" as "IVIIllamt", "Gordon:" as "CSordoat" and "ABBEY, Edwin Austin:" as "ARBRY, Bdwin Ansttni". Basically, it seems to have trouble with font used for the names in combination with the printing.

I'm looking into doing OCR on it myself, but I'm not sure I can get it working or how long it would take. Maybe you have a different book? Kind regards from PJvanMill)talk( 20:00, 12 May 2024 (UTC)

Incidentally, if you could outline the steps you used in that extraction, it'd be helpful, so that I don't have to pester you further. Thanks. DS (talk) 15:01, 25 September 2026 (UTC)

Hi DragonflySixtyseven, from memory I think I did it mostly with regular expressions. I first obtained the text as a plaintext file, then I noticed that in the text file the lines containing the names followed a certain pattern, so I was able to write a regular expression that selected almost all lines with names and threw away almost all else. Then the rest was done by alternating between manual interventions and more regular expressions. Kind regards from PJvanMill)talk( 23:15, 18 October 2026 (UTC)

Addendum: before the first regular expression as described above, I think I had actually already deleted parts of the file. I don't remember whether I continued with the body of the text or with the index; but I must have chosen one or the other, or tried it both ways. PJvanMill)talk( 23:51, 18 October 2026 (UTC)

@DragonflySixtyseven Also, while I don't often log in on my wikipedia account nowadays, I wouldn't mind doing some more. As long as the OCR has a small error rate as it did with "One thousand men of Vermont", it's not that much work. If the OCR is of poor quality my method simply falls flat, one would have to first run some kind of OCR on it oneself to obtain a suitable plaintext version of the text. Back then I did at some point look into running OCR software on my own computer, but I didn't really get anywhere with this. But if you happen to have some books where the OCR is of high quality, I could do some more of these extractions for you once in a while. Kind regards from PJvanMill)talk( 23:37, 18 October 2026 (UTC)

If you had to write the regex, that does sound like more than I could handle by myself. I'll see what I can find in terms of better-OCR'd text - perhaps the 1916-17 Marquis Who's Who, or the 1906 edition? DS (talk) 01:26, 19 October 2026 (UTC)