Exploring WhatsThat – measuring semantics & possible refinement with Marker Engine

[DYAI2025]

Hi Markrai,

I saw your repo on youTube at "GitHub-Awesome" and those are also a lot of my projrcts, measuring deeper patterns in social dynamics. Make the unspoken visible. I like that! So of course I found your project WhatsThat really fascinating and inspiring. The way it approaches conversational interpretation and visualization caught my attention immediately.

I could not really find out how exactly do measure and interpret the semantic aspects within your analysis. From the structure I’ve seen, the logic seems elegantly minimalistic, but I couldn’t quite identify the mechanism behind the actual measurement or signal extraction.

That’s particularly interesting for me because I’ve been working on a Languages pattern recogntion system based on marker. From which I nearly created 1000 in the last 5 month. Heuristic tools for holistic understanding of situations and dynamics. In theory the system that identifies and arranges communication patterns into four interconnected interpretive layers:

1. Atomic (ATO): The smallest recognizable signal a tone, phrase, or linguistic cue.
2. Semantic (SEM): Groups of atomic markers combine into clear, context-driven meaning.
3. Cluster (CLU): Sets of semantic meanings form relational themes or dynamics.
4. Meta (MEMA): The overarching view of how these patterns evolve over time to tell the story of the interaction.

Each level builds on the previous one, enabling an increasingly comprehensive understanding without hallucinations or noise. This structure serves as a kind of semantic scaffold that language models can reliably follow.

I also have a WhatsApp project, which is more egocentric about the user and their social relationships and dynamics over time. It uses the markers to make the social bonding mechanism visible. Theoretically so far only.

If you are interested in how I might be able to contribute to this, that would be very exciting. The Marker Engine has grown very large and has not been developed iteratively enough, so there is no real front end yet. However, with your extension, you have a very good and expandable one.

I attached some examples how it works in text chats and how those marker are structured.

ATO_CRITICISM_ATTACK_MARKER.yaml
SEM_IDEALIZATION_DEVALUATION.yaml
CLU_AVOIDED_DISCLOSURE.yaml
ATO_INVITING_RESONANCE.yaml
CLU_OFFENSIVE_FLIRTING.yaml
MEMA_ATTACHMENT_PROFILE_MARKER.yaml

marker_chat_visual.html

Best regards,
Ben

[markrai]

Hi @DYAI2025 I'm seeing this so late! my sincere apologies. I will respond as soon as I come back to this project!