fix(irm): Improve model evaluation

1 files changed, 4 insertions, 0 deletions

diff --git a/src/umabot/rules/intelligent_moderator_base.py b/src/umabot/rules/intelligent_moderator_base.py
index 7839f58..f3a5721 100644
--- a/src/umabot/rules/intelligent_moderator_base.py
+++ b/src/umabot/rules/intelligent_moderator_base.py
@@ -46,6 +46,8 @@ Guidelines for Content Needing Development:
 - Only remove content that is genuinely low-quality, not just short
 - High confidence (0.7+) for clear cases, lower confidence for borderline cases
 
+IMPORTANT: Media presence does NOT automatically make content acceptable. Evaluate the TEXT CONTENT quality regardless of whether there's an image, video, or other media attached. A post with media can still be low-effort if the text content is brief, lacks context, or doesn't add meaningful value.
+
 Examples of ACCEPTABLE content (should NOT be removed):
 - Character status updates with context and future plans
 - Posts with OOC notes explaining absence or context
@@ -59,6 +61,8 @@ Examples of content that NEEDS MORE DEVELOPMENT (should be removed):
 - Posts that are just "I think..." with no elaboration
 - Content that lacks any meaningful character development
 - Posts that are clearly spam or off-topic
+- Posts with media but minimal text content (e.g., "Whaddya looking at?" with just a link)
+- Posts that are just questions or statements without context or development
 
 Post to evaluate:
 Title: {title}