NP_SpamBayes 1.1.0 done !

Post your new plugins here!
User avatar
xiffy
Nucleus Guru
Nucleus Guru
Posts: 1194
Joined: Wed Mar 27, 2002 6:37 pm
Location: Deventer
Contact:

Postby xiffy » Wed Nov 07, 2007 11:11 pm

well,
clueless. the array returns empty in nucleus 3.31. For no apparent reason.
This will take some more time.
verbaljam
Posts: 666
Joined: Wed Jul 31, 2002 4:58 pm
Location: Amsterdam, The Netherlands
Contact:

Postby verbaljam » Sat Nov 10, 2007 11:57 am

Using this excellent plugin on two websites that get lots of spam, I have a few suggestions to further improve the usability. This in case there are plans for further upgrading the pluging.

1. I train most spam from within the log. Sometimes there are still spam messages between the ham. At this moment I first have to select them individually, then train them as spam, after that I have to scroll through the whole list again, select them again and delete them.
With hundreds of comments on the site this is a lot of extra work. Therefore I would suggest an option: 'train spam and delete'. Saves a lot of work.
After that I can select all the ham in one mouseclick and train the ham messages.

2. I noticed that the plugin is only capable of training 250 messages in one operation. Or is this amount limited by the server or php?

3. False positives can be published. Very useful! However, I noticed that linebreaks in messages dissappear. Is there a solution for this or do I have to live with that?

4. At his moment each comment appears double in the logfile (as event ValidateForm and as event PreAddComment). I know I can filter them, but as a simple user I don't understand the meaning of these two events. So I also don't understand the consequences of filtering. What I just want to do is: check the list of comments and train the filter.
So I wondered: Isn't it better to just show alle the comments one time and hide all the 'inside tech-stuff' for dummies like me? ;-)

5. Sometimes (in rare cases) a comment (ham) shows up in the list that never appears on the site. Most of the time one of the events is missing then. Therefore a 'publish' option for ham would be nice too.

6. I think it's a good idea to use this plugin in combination with the BadBehaviour plugin (put this one first in the plugin list), because they told me BadBehaviour already blocks spam at 'the front door' and asks less resources from the server. My hosting provider is always complaining about the heavy server load from Nucleus plugins. Don't know if he's right...
verbaljam
Posts: 666
Joined: Wed Jul 31, 2002 4:58 pm
Location: Amsterdam, The Netherlands
Contact:

Postby verbaljam » Sun Nov 18, 2007 10:01 pm

Just upgraded to Nucleus 3.31 and I also get the error:
Warning: Invalid argument supplied for foreach() in /home/....../nucleus/plugins/spambayes/index.php on line 261

I see it's a known issue and Xiffy is working on it.

By the way: I hope this is not affecting SpamBayes anti-spam capabilities, but I guess not. I switched off the log function for the time being.
cyblot
Nucleus Guru
Nucleus Guru
Posts: 399
Joined: Tue Sep 16, 2003 8:49 pm
Location: Netherlands
Contact:

Postby cyblot » Sun May 25, 2008 11:33 am

xiffy wrote:well,
clueless. the array returns empty in nucleus 3.31. For no apparent reason.
This will take some more time.


Have you had any chance to figure this out? It still works great, blocking stuff, but being able to select entries and train the filter again would be be very welcome too.

I'm currently using Nucleus 3.32, but the problem is still there (Spambayes 1.1.0)
Blots of Info
http://www.golb.org
User avatar
matt_t_hat
Posts: 1123
Joined: Sun Aug 21, 2005 4:45 pm
Location: UK
Contact:

Use love it - need one more thing...

Postby matt_t_hat » Tue May 27, 2008 1:17 pm

This is why I hate releasing code - people like me!

I rely heavily on thins plug - between it and bad behaviour I'm 99.9999% spam free these days.

However when a lot of spam get's in (often when the plug is freshly installed I can have up to 1000 pages of spam in my logs.

What I could use are a couple of extra features.

1. Train and unlog - this trains as spam/ham and removes it from the log in one move halving the processing time.

2. Retest / Retest-some / Retest Selected - this would grab upto n most recent logs and retest them based on the latest training. If this could present a list of all newly identified spam and offer me a list of

[Train ALL Spam] [Return to Filtered View] [Delete All Shown]

that would be great and save me some more time.

3. List by score - this would allow me to order the log list by the score low-to-high or hgh-to-low reducing the chances of ham/spam being ont he same page and thus the ammount of reading I nee to do (or the ew threshold I need to set).

4. Train Spam All Logs Scoring Above x - this would allow me to train as spam everything in the logs above a given score which would save me piles of time. Train ham all below in the same way might be good too. Optionally being able to remove them from log in the same action nice but not essential (see 5).

5. Retest if trained - this would mean that if a log item is trained as ham/spam it would be retested and rescored and marked as appropriate. This should in most cases mark the item as belonging to the group it was used to train.

6. Common list - (weaker idea) it should be possible to generate a lsit of words so common that they show up in both groups so regularly as to be not worth much points. Would it be worth listing these words? I think it might but to auto ignore them would rob the plug of some of it's subtle edge.

7. Nuclear Option - Everything in the logs is obviosuely spam train it all and then remove it. After a massive spam run on a new blog (lacking any readership) this would be a godsend.

8. Import/Export - the same spammers hit all my blogs so being able to create a master training block (with, say, a honeypot to help) would make hardening blogs much faster.

9. SkinVar: ThisIsSpam - adds a smll admin only button to all comments and allows the comment to be optionally deleated and trained as spam while the commenting email/ip/user is blocked, repored or black listed depending on what other plugs the blog has running. (Might want a simple API for that).

10. SkinVar: TrainHam - press to train as ham makes a pop-up box appear saying "done" but allows you to stay on page.

11. SkinVar: ThisIsAllSpam - an atomic version of 9. Spammers tend to target commentless posts. also good for honeypot systems.

12. SkinVar: ShowScore - allows the spam/ham score to be sown optional paramiters allow this to be shown to everyone/members/blogteam/blogamin/superadmin.

Now you know why I hate people like me. I actually figure that some of those wants could be calls tot he same core set of fucntions so I've probably made it seem more demanding than I could have.

I love the plugin - keep up the good work.
sluggo
Posts: 14
Joined: Wed Jun 04, 2008 2:20 am

Postby sluggo » Wed Jun 04, 2008 9:25 am

???

eleventeen pages of blather, all for the sake of "I refuse to utilize a CAPTCHA" ?

(I'm asking) Maybe there's an implication, beyond preventing spam comments?
Does placement of a CAPTCHA interfere with trackbacks, or pingbacks?

(I've just started reviewing the NucleusCMS app & haven't yet checked whether it handles 'pingbacks'.)
ftruscot
Nucleus Guru
Nucleus Guru
Posts: 7430
Joined: Wed Feb 22, 2006 6:19 pm
Location: Massachusetts
Contact:

Postby ftruscot » Wed Jun 04, 2008 1:38 pm

There is a Ping plugin that comes packages with the 3.3x version of nucleus, though it is not enabled by default.

Some people use things besides captcha to find spam because they would rather the spam fighting happen on the admin side and not affect the user experience. And while captcha is pretty good, it's not fail proof. Many people combine captcha with something else.
ftruscot
Nucleus Guru
Nucleus Guru
Posts: 7430
Joined: Wed Feb 22, 2006 6:19 pm
Location: Massachusetts
Contact:

Postby ftruscot » Wed Jun 17, 2009 4:40 am

Xiffy, if you're still watching this thread, could you take a look at the issue described by wgroleau in this thread?
Is your question not solved yet?
Search our FAQ,
read the Documentation, or
browse the list of available plugins.

Check out my plugins
User avatar
matt_t_hat
Posts: 1123
Joined: Sun Aug 21, 2005 4:45 pm
Location: UK
Contact:

Possible bug

Postby matt_t_hat » Sun Jan 29, 2012 9:52 am

I've noticed that there may be a problem with 1.1.0

When I click "select all" and choose the option Train Spam and press "Uitvoeren" (which I guess means submit) the probability and wordcount do not change however when I click the individual train spam link the values change as expected.
User avatar
LordMatt
Posts: 33
Joined: Wed May 01, 2013 7:46 pm

Postby LordMatt » Wed Nov 20, 2013 6:17 pm

I've created a repo on hithub for xiffy's fine code.

https://github.com/Lord-Matt-NucleusCMS ... _SpamBayes

I plan to add a help file so if you want to contribute to the help file feel free to fork the repo and do just that.

Return to “Plugin Development”