Enhanced Moderation and User Experience for `plone.app.discussion`

[rohnsha0]

PLIP (Plone Improvement Proposal)

Responsible Persons

Proposer: @rohnsha0

Seconder:

Abstract

This PLIP proposes significant enhancements to plone.app.discussion to modernize the commenting system with advanced moderation capabilities, improved user experience, and comprehensive administrative controls. The proposal includes intelligent content filtering (both rule-based and AI-powered), streamlined moderation workflows, and sophisticated user ban management systems.

Motivation

The current plone.app.discussion package, while functional, lacks modern content moderation capabilities essential for managing large-scale community discussions. Site administrators struggle with:

• Manual moderation of high-volume comment streams
• Lack of proactive spam and toxic content detection
• Limited tools for managing problematic users
• Insufficient granular control over comment approval workflows

These limitations force administrators to either disable comments entirely or invest significant manual effort in moderation, limiting community engagement and platform scalability.

Assumptions

• Plone 6.x deployment with plone.app.discussion already installed
• Site administrators have access to external AI services (optional, for AI moderation features)
• Existing comment data and workflows should remain functional during and after upgrade
• Performance impact should be minimal for sites not using advanced features
• Integration should respect existing Plone security and permission frameworks

Proposal & Implementation

graph TD
    A[User Submits Comment] --> B{Word Filter Check}
    B -->|Pass| C{AI Moderation Check}
    B -->|Fail| D[Action Based on Filter Rule]
    
    C -->|Safe Content| E{User Ban Status Check}
    C -->|Flagged Content| F[AI Moderation Action]
    
    E -->|User Not Banned| G[Comment Published]
    E -->|Cooldown Ban| H[Comment Rejected - Cooldown Active]
    E -->|Shadow Ban| I[Comment Visible Only to Author]
    E -->|Permanent Ban| J[Comment Rejected - User Banned]
    
    D -->|Warning| K[Show Warning Message]
    D -->|Pending Review| L[Comment Moves to Pending Queue]
    D -->|Auto Reject| M[Comment Rejected]
    
    F -->|Delete| N[Comment Deleted]
    F -->|Move to Spam| O[Comment Marked as Spam]
    F -->|Pending Review| L
    
    K --> P[User Can Edit and Resubmit]
    L --> Q[Admin Reviews Comment]
    
    Q -->|Approve| G
    Q -->|Delete| N
    Q -->|Mark as Spam| O
    Q -->|Apply User Ban| R[Configure Ban Type]
    
    R -->|Cooldown| S[Set Temporary Ban Period]
    R -->|Shadow Ban| T[Enable Shadow Ban Mode]
    R -->|Permanent Ban| U[Apply Permanent Ban]
    
    style A fill:#e1f5fe,color:#000
    style G fill:#c8e6c9,color:#000
    style N fill:#ffcdd2,color:#000
    style O fill:#fff3e0,color:#000
    style H fill:#ffcdd2,color:#000
    style I fill:#fff3e0,color:#000
    style J fill:#ffcdd2,color:#000

Loading

1. Word-based Content Filtering

• Create configurable word filter registry with full administrative control
• Implement real-time content scanning before comment submission
• Provide user-editable filter lists through control panel interface
• Add pattern matching support (regex capabilities)
• Configurable response actions: delete, move to spam, set to pending

2. AI-powered Moderation System

• Train and implement Spam filtering AI models (I will be doing it 😁)
• OPTIONAL - Integrate with external AI moderation APIs (OpenAI Moderation, Google Perspective, Azure Content Moderator) -
• Implement detection for: promotional content, toxicity, restricted terms, hate speech
• Configurable response actions: delete, move to spam, set to pending
• Confidence threshold settings for each detection category

3. Streamlined Moderation Workflows

• Simplify comment states to: Published, Pending, Spam, Deleted
• Create unified moderation interface with bulk actions
• Implement quick-action buttons: Approve, Delete, Mark as Spam
• Add moderation dashboard with filtering and search capabilities

4. Comprehensive Ban Management

• Cooldown Bans: Temporary restrictions with automatic expiration
• Shadow Bans: Comments visible only to author, hidden from others
• Permanent Bans: Complete commenting restriction
• Admin interface for ban management and history
• Appeal system for contested bans (configurable)

Configuration and Administration

• Comprehensive control panel for all new features
• Feature flags for enabling/disabling components
• Performance monitoring and reporting tools
• Import/export capabilities for filter lists and ban data

Deliverables

Core Package Updates

• plone.app.discussion: Enhanced with all proposed features
• plone.app.registry: New configuration schemas for moderation settings

New Components

• AI moderation service integration utilities
• Enhanced comment workflow definitions
• Moderation dashboard views and forms
• Ban management interfaces

Documentation

• Administrator guide for new moderation features
• Developer documentation for extending moderation capabilities
• Migration guide for existing installations
• Performance optimization recommendations

Risks

• Existing comment data and workflows remain unchanged
• Default behavior preserves current functionality
• AI moderation may introduce some more CPU usage (can be configured to off to save this)

Participants

• TODO

cc/ @plone/framework-team

[davisagli]

Hi @rohnsha0. I'm curious, are you running a Plone site that uses the discussion features, or are you just excited about implementing a more complete set of commenting features?

I don't want to be discouraging, but this is a complex proposal and I think we need to think carefully about whether it makes sense to add these features to Plone core. There are a lot of good ideas here, but we can't just add every feature we think of, or we risk making the software harder to understand and maintain.

We can apply the Pareto principle here: if a feature is needed by 80% of Plone sites, then it should probably be in Plone core; otherwise, it may belong better in an add-on. plone.app.discussion as a whole is something that we've already been moving in the direction of becoming an add-on, since many Plone sites do not use it. Of course, that could be partly because it is missing some important moderation features like the ones you are proposing. But I also think it's less common these days to enable commenting on a public website, and intranets are less likely to need complex moderation. If a client came to me with these requirements, I would first see if Discourse would be a better fit, since it already has a lot of these features.

I appreciate that you're already planning to avoid disturbing existing functionality and make the new features optional. That also makes me suspect that a lot of this could be implemented as a separate add-on which works together with plone.app.discussion, instead of adding it to Plone core. Do you think that would be a possibility?

Frankly, my biggest concern is the effort to fully review, test, and maintain the features. I don't think I can personally commit to helping with that this summer, since I'm already feeling kind of busy with the existing set of features that is in the roadmap for Plone 6.2, and I have some holidays.

Perhaps I am being too cautious. If there are other folks in the Plone community reading this who need these features and are disappointed I'm saying they may not belong in core, then I invite you to step forward and volunteer to help implement, review, test, and maintain them. It would make me feel better about it if there was a team of people working together on it instead of just one person.

[tisto]

@rohnsha0 wow. this is an impressive PLIP Rohan! I agree with @davisagli here. I'd suggest splitting this PLIP into slightly smaller PLIPs so that we can add them one after another to the core. Some features might be better off as optional features or should go into add-on products.

I was a GSOC student in 2009 and I wrote p.a.discussion in that year. I also split up my work into Plone core and a set of add-ons that can be installed on request. I am rather busy these days but I would love to help make this happen!

[rohnsha0]

Thank you for the thoughtful feedback! @davisagli, I really appreciate the strategic perspective on core vs add-on decisions and the 80/20 principle. @tisto, it's incredibly encouraging to having you interested in helping make this happen!

I'm working with few Plone sites that struggle with comment moderation, which sparked this proposal. The manual overhead is significant enough that some have disabled comments entirely, which seems like a missed opportunity for community engagement.

However, after considering feedbacks, I think a hybrid approach makes the most sense:

• Start with essential moderation features in core (basic word filtering, streamlined workflows and sophisticated ban management)
• Develop advanced features (AI moderation) as add-ons
• Build community adoption and prove demand before considering further core integration

Following your suggestions to split into smaller PLIPs, I'm thinking:

Phase 1 - Core Essentials (PLIP for core):

• Basic configurable word filtering
• Simplified comment states (Published, Pending, Spam, Deleted)
• Enhanced moderation dashboard with bulk actions
• Simple ban management (temporary restrictions)

Phase 2 - Advanced Moderation (Separate add-on):

• AI-powered content filtering
• External API integrations (OpenAI, Perspective, etc.)
• Sophisticated ban types (shadow bans, appeals system)
• Advanced analytics and reporting

Phase 3 - Community Extensions (Additional add-ons as needed):

• Integration with external comment systems
• Advanced workflow customizations
• Specialized industry-specific filters

@davisagli, I'm confident I can handle the technical implementation solo, though I'll definitely reach out for feedback when I hit challenging architectural decisions.

@tisto @davisagli , I'd love to get your technical guidance on the current p.a.discussion architecture and which features you think would have the most impact. which of these features do you think would benefit the majority of Plone sites using comments vs. which should remain as optional add-ons?

@davisagli, does this phased approach address your concerns about maintenance burden?
Regarding long-term maintenance, I'm open to volunteer as a maintainer for these features if they make it into core. I understand this is a significant responsibility and I'm prepared to handle bug fixes, compatibility updates, and feature evolution over time.

[mauritsvanrees]

Nice!

@ksuess has implemented the current extended moderation of comments in Plone 5.2. See mainly plone/plone.app.discussion#166. She made more contributions, for example captcha validation. So she may have some input.

From your Phase 1, two points are already in core due to her work, if you enable comment moderation:

• Simplified comment states (Published, Pending, Spam, Deleted) -- Plone currently has 'Rejected' instead of 'Deleted'
• Enhanced moderation dashboard with bulk actions

What would you want to change about these parts?

The other two points, basic word filtering and temporary bans, sound useful.

About temporary bans:

• Do you envision doing this only for authenticated users?
• Or also for anonymous visitors? Then you would have to be careful to get the correct original IP address, instead of the address of an intermediate server.

[gforcada]

For the basic word filtering, you can reuse parts of https://github.com/collective/collective.contentalerts which is what we are using in production at work.

Given the size of our website and number of comments (over the half a million mark, though we peaked at some point at +1 million), we added some conditional dependency on collective.taskqueue, to run async tasks, which I would love that to be part of Plone core (again, as it was already in with old Zope 2 on python 2.7) but it gets a bit out of scope for this PLIP 😅

[rohnsha0]

Simplified comment states (Published, Pending, Spam, Deleted) -- Plone currently has 'Rejected' instead of 'Deleted'

@mauritsvanrees, I meant the moderation buttons... see plone/plone.app.discussion#273

Enhanced moderation dashboard with bulk actions

To encorporate the changes fromm the same moderation buttons

About temporary bans:

• Do you envision doing this only for authenticated users?
• Or also for anonymous visitors? Then you would have to be careful to get the correct original IP address, instead of the address of an intermediate server.

At this moment, I find planning for only authenticated user, as for anonymous there emerges the space I am unaware of...

For the basic word filtering, you can reuse parts of https://github.com/collective/collective.contentalerts which is what we are using in production at work.

I will see it while implementing it... thanks @gforcada

[tisto]

@rohnsha0, your proposal makes lots of sense to me. Sorry for not replying sooner here. I am currently on holiday and a bit out of the loop when it comes to actual development. However, I am always happy to help and share insights of why we things were initially implemented in the way they are today.

[rohnsha0]

Sorry for not replying sooner here. I am currently on holiday and a bit out of the loop when it comes to actual development.

@tisto I have pushed all the PRs, ready for review!

#276 | #277 | #279

[tisto]

@rohnsha0 sorry for not getting back to you sooner. I will have a look at the PRs next week. Is this something that we would also want to finish for Plone 6.2 before the end of the year? Or shall we rather give us some more room and time for this?