MVP: Influence cluster graph and repeated amplification path detection #5

New issue

Closed

opened 2026-04-01 06:30:44 +00:00 by chrysolambda-ops · 0 comments

chrysolambda-ops commented

2026-04-01 06:30:44 +00:00

(Migrated from github.com)

Problem

Coordinated narratives often move through repeatable source pathways, including state-affiliated media -> influencer ecosystems.

Goal

Build a graph of source/account clusters and detect frequent amplification pathways with explicit uncertainty labels.

Scope

Nodes: sources/accounts/domains
Edges: citation, repost, linkage, temporal dependence hints
Metrics: frequency, lag, citation depth, direct vs indirect dependence
Output: cluster graph + top repeated pathways report

Out of scope (v0.1)

Definitive motive/intent attribution
Hidden/private-channel inference

Acceptance criteria

Emits graph artifact (JSON edge list minimum)
Computes pathway ranking by repeat frequency + lag
Marks low-confidence inferences explicitly
Supports seed lists (state-affiliated domains, influencer accounts)

Dependencies

Provenance extraction outputs
Trend alerts for candidate topic windows
Propagation benchmark fixture (to be added)

Benchmark target

Pass first propagation fixture with expected path ranking in top-N.

Reference

docs/mvp-use-cases-and-issue-specs.md (Issue-ready spec 5)

Imported from GitHub issue/PR. Originally posted by chrysolambda-ops on 2026-04-01T06:30:44Z.

## Problem Coordinated narratives often move through repeatable source pathways, including state-affiliated media -> influencer ecosystems. ## Goal Build a graph of source/account clusters and detect frequent amplification pathways with explicit uncertainty labels. ## Scope - Nodes: sources/accounts/domains - Edges: citation, repost, linkage, temporal dependence hints - Metrics: frequency, lag, citation depth, direct vs indirect dependence - Output: cluster graph + top repeated pathways report ## Out of scope (v0.1) - Definitive motive/intent attribution - Hidden/private-channel inference ## Acceptance criteria - [ ] Emits graph artifact (JSON edge list minimum) - [ ] Computes pathway ranking by repeat frequency + lag - [ ] Marks low-confidence inferences explicitly - [ ] Supports seed lists (state-affiliated domains, influencer accounts) ## Dependencies - Provenance extraction outputs - Trend alerts for candidate topic windows - Propagation benchmark fixture (to be added) ## Benchmark target Pass first propagation fixture with expected path ranking in top-N. ## Reference `docs/mvp-use-cases-and-issue-specs.md` (Issue-ready spec 5) --- Imported from [GitHub issue/PR](https://github.com/Kyvero-Vexus/truth-kit/issues/5). Originally posted by [chrysolambda-ops](https://github.com/chrysolambda-ops) on 2026-04-01T06:30:44Z.