AI Moderator for DLE — smart moderation of comments and news via neural network

The module is especially useful for sites where users actively write comments or add news. You no longer need to manually check every message all the time — artificial intelligence takes care of the main work.

What the module does

When a user adds a comment, the module does not publish it immediately in the usual form. Instead, a neat informational message appears on the site:

«Our security system is checking this comment; after the check, it will be published.»

At this time, the actual comment text is sent to the moderation queue. The neural network analyzes it and makes a decision: it can be published or it needs to be rejected.

If the comment is normal — it appears on the site automatically.
If the comment violates the rules — it is deleted or hidden, depending on the settings.

User news moderation

The module can check not only comments, but also news added by users.

Before publication, the news goes through AI verification. The module can determine:

• whether the news contains prohibited topics;
• whether it is similar to already published materials;
• whether the text is spam;
• whether the news contains politics, insults, or provocations;
• whether the material is suitable for publication on the site.

If the news does not pass verification, it is not published. The user can automatically be sent a private message with the reason for the rejection.

Checking similar news

One of the useful features of the module is searching for similar publications. If a user tries to add news that is already on the site or is very similar to an existing one, the module can stop the publication.

This helps avoid duplicates, repeats, and junk content in the news feed.

Support for different neural networks

The module can work with different AI providers:

• OpenAI / ChatGPT;
• DeepSeek;
• OpenAI-compatible API;
• other services with a compatible API.

Block «User Opinions»

Additionally, the module can create a short AI summary of comments on a news item or movie.

For example, before the comments you can display a block:

User Opinions

Users mostly rated the material positively. They liked the plot, atmosphere, and acting the most. Some noted that the ending turned out to be controversial, but overall the discussion is rather positive.

This summary is generated based on all approved comments for the current news item. It is convenient for movie sites, news portals, reviews, articles, and any materials where it is important to quickly understand the audience's overall opinion.

The summary is cached, so the neural network is not called every time the page is opened.

Flexible verification rules

You can specify yourself, what exactly should be considered prohibited content. For example:

• spam;
• politics;
• insults;
• profanity;
• advertising;
• links;
• provocations;
• meaningless messages;
• toxic comments;
• prohibited topics for your site.

You can add your own rules and instructions for the neural network so that the module checks content specifically for your project.