- Home
- Skills
- Research & Analysis
- landscape lineage review
landscape lineage review
Map a research field, then trace how its key approaches evolved, for literature reviews in any discipline.
Free
Works with the AI Agents you already use
landscape lineage review
Example session with this skill installed
Perform a standard literature review on "Graph Neural Networks for Fraud Detection", tracing how they evolved to handle class imbalance.
- Read your context and instructions
- Compiled the landscape lineage review
I have completed a standard two-phase review of Graph Neural Networks (GNNs) for fraud detection, specifically focusing on the evolution of class imbalance handling.
Framing and Scope
- Task: Detecting minority fraudulent nodes in large-scale graphs.
- Design Axes: Sampling strategy (topology vs. feature), Loss weighting, and Neighborhood aggregation.
- Mode: Standard (22 core papers identified, 3 lineage hops).
Landscape
| Paper | Year | Core Contribution | Why Core |
|---|---|---|---|
| Wang et al. | 2019 | Semi-supervised GNNs | Establishes the baseline for GCNs in fraud |
| Liu et al. | 2020 | PCGNN | First major attempt at label-balanced sampling |
| Zhao et al. | 2021 | GraphSMOTE | Synthetic node generation for GNNs |
Lineage Map
graph TD
A[GCN/GAT, 2017-18] -->|Relaxed: Homophily assumption| B[PCGNN, 2020]
B -->|Mechanism: Label-aware neighbor filtering| C[GraphSMOTE, 2021]
C -->|Mechanism: Interpolation in embedding space| D[Pick-n-Choose, 2022]
D -->|Relaxed: Fixed sampling ratio| E[Frontier: AoE-GNN, 2024]
Transitions
GCN → PCGNN: GCNs assume neighbor nodes share labels (homophily). Fraudsters are often "camouflaged" (heterophily). PCGNN relaxes the homophily assumption by filtering neighbors based on label similarity, at the price of increased computational overhead during training.
2.
PCGNN → GraphSMOTE: PCGNN handles existing nodes but fails when the minority class is extremely sparse. GraphSMOTE introduces a synthetic generation mechanism to create new fraud nodes in the embedding space, assuming that the feature manifold is locally continuous.
Frontier and Gaps
Mechanistic Gap: Most current frontier models (AoE-GNN) still assume static graph topology. Evolution into temporal-dynamic class imbalance handling remains an open gap.
Transfer Opportunity: Transferring "Cost-Sensitive Reinforcement Learning" from the robotics field to GNN neighbor sampling could address dynamic reward scaling in fraud detection.
Next steps
- Review the generated
fraud_detection_references.bibfor full citation details. - Examine the "Cross-domain transfer" section in the attached document for the RL-based sampling proposal.
- Verify the unverified 2024 preprints listed in the audit log.
Connects securely to your tools. The creator never sees your data.
About this skill
Landscape → Lineage Review is a literature-review skill for any research field. It combines a broad map of the field with a deep, principle-driven account of how its key approaches evolved.
Phase 1 – Landscape (breadth) Expands the search vocabulary, retrieves and deduplicates widely, and selects core papers using several signals (multi-query hits, age-adjusted citations, survey citations, use as a baseline). It then maps themes, a timeline and open debates, self-checks coverage, and searches again if anything is thin.
Phase 2 – Lineage (depth) Starts from anchor approaches chosen on the map and chains forward through citations, limitation-driven queries and framing-driven queries. Every step is explained by the assumption it changed, the mechanism, the price paid and the conditions under which it wins. It also abstracts each approach into a domain-free problem to find better versions in other disciplines and assess whether they transfer back.
You get a search log, a core-paper table, themes and debates, a lineage map, a comparison table, cross-domain transfer cards, ranked research gaps, a coverage audit, and references labelled by how deeply they were read. The skill never invents sources.
Four modes: Quick, Standard, Deep, Trace. Answers in the language you write in.
Works with Claude (Claude.ai, desktop, Claude Code), OpenAI Codex and other SKILL.md-compatible agents. Web search should be enabled.
About the included script: scripts/scholar_tool.py is optional. It makes read-only requests to the public Semantic Scholar and OpenAlex APIs, reads only the optional S2_API_KEY environment variable, and uses only the Python standard library. It writes nothing outside the working directory and sends no user data anywhere else.
MIT-licensed. Source, documentation and updates: https://github.com/Louis-Chenyang-Lu/landscape-lineage-review
Requires access to scholarly search tools or web search. Effectiveness depends on the availability of open-access metadata (DOI/arXiv). Limited by current tool rate limits and database coverage.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account
Requires
Permissions
Allowed Hosts