The form at admin/settings/mollom/manage/% has an option for "Text analysis accuracy" (Strict, Normal, Relaxed), but no explanation of what it means or what the different options do. I'd suggest adding some help text.

Files: 
CommentFileSizeAuthor
#10 0001-1801220-by-sun-Cottser-TravisCarden-Clarified-text-a.patch1.17 KBsun
PASSED: [[SimpleTest]]: [MySQL] 4,449 pass(es).
[ View ]
#8 mollom.ui-strictness.8.patch803 bytessun
PASSED: [[SimpleTest]]: [MySQL] 5,027 pass(es).
[ View ]
#4 2013-03-31_06h11_53.png7.68 KBsun
#4 2013-03-31_06h24_34.png3.78 KBsun
#4 mollom.ui-strictness.4.patch757 bytessun
PASSED: [[SimpleTest]]: [MySQL] 5,028 pass(es).
[ View ]
#2 mollom-clarify-text-analysis-strictness-1801220-2.patch976 bytesTravisCarden
PASSED: [[SimpleTest]]: [MySQL] 4,643 pass(es).
[ View ]

Comments

Yes, what is strict versus normal?

Title:What is "Text analysis accuracy"?Clarify "Text analysis accuracy"
Version:6.x-2.1» 7.x-2.x-dev
Status:Active» Needs review
Issue tags:+Novice
StatusFileSize
new976 bytes
PASSED: [[SimpleTest]]: [MySQL] 4,643 pass(es).
[ View ]

I got the following answer to this question from Mollom support:

About the quickest answer I can give you is that each comment generates a "score" showing its likelihood of being spam. For strict, the spam "boundary" is moved in the direction fo the ham "boundary", so a larger group of posts are regarded as spam. For relaxed, it's exactly the opposite. In other words, the score is evaluated differently depending on the setting—a certain score may register as "spam" on strict, when it would not on normal or relaxed, while a certain score may register as ham on relaxed, where it would not on strict or normal.

That being the case, I propose two changes: 1) Change the field label from "Text analysis accuracy" to "Text analysis strictness", since accuracy connotes error, whereas strictness (more appropriately) connotes rigidity. (The field machine name is "strictness", so apparently that's how the engineers think about it!) 2) Add clarifying help text. I suggest something like the following:

The greater the strictness, the more aggressive the analysis. Use "Strict" if a lot of spam is getting through. Use "Relaxed" if you're getting a lot of false positives.

Here's a patch that does those two things.

Status:Needs review» Reviewed & tested by the community

Patch still applies and clears up the confusion I had.

Status:Reviewed & tested by the community» Needs review
StatusFileSize
new757 bytes
PASSED: [[SimpleTest]]: [MySQL] 5,028 pass(es).
[ View ]
new3.78 KB
new7.68 KB

Thanks for improving the user interface!

This is what I see with this patch:

2013-03-31_06h11_53.png

Observations and considerations:

  1. The first sentence does not seem to carry any message. Also, "aggressive" sounds too scary. I wonder whether we can omit that sentence entirely?
  2. "false positives" is a pretty techy term.
  3. Would it make sense to enhance and clarify the labels instead?

Along the lines of this?

2013-03-31_06h24_34.png

In words:

Text analysis strictness

* Strict: Bias towards spam
* Normal
* Relaxed: Bias towards ham

That looks more skimmable, but at a glance the wording looks to me like Strict might *let in* more spam. Can we spell it out a bit more? Something like:

Text analysis strictness

* Strict: Analysis is more likely to block posts as spam
* Normal
* Relaxed: Analysis is more likely to allow posts as ham

Issue tags:+Usability

Good point!

I think that makes sense. However, note that posts are not necessarily "blocked" when Mollom identifies a post to be spam, since there is a separate option in this form to control that. Thus, leaving that bit out could result in:

Text analysis strictness

* Strict: Posts are more likely analyzed as spam
* Normal
* Relaxed: Posts are more likely analyzed as ham

"more likely analyzed" looks and reads a bit lengthy, but I do not have a better suggestion right now.

Also a good point :) What about this?

Text analysis strictness

* Strict: Posts are more likely classified as spam
* Normal
* Relaxed: Posts are more likely classified as ham

StatusFileSize
new803 bytes
PASSED: [[SimpleTest]]: [MySQL] 5,027 pass(es).
[ View ]

Works for me! :)

Status:Needs review» Reviewed & tested by the community

Looks great.

Version:7.x-2.x-dev» 6.x-2.x-dev
Status:Reviewed & tested by the community» Needs review
StatusFileSize
new1.17 KB
PASSED: [[SimpleTest]]: [MySQL] 4,449 pass(es).
[ View ]

Thanks! Committed to 7.x-2.x.

Status:Needs review» Fixed

Thanks for reporting, reviewing, and testing! Committed to all branches.

A new development snapshot will be available within the next 12 hours. This improvement will be available in the next official release.

Automatically closed -- issue fixed for 2 weeks with no activity.