# quanxiny/rcr-featureenvy architecture

> Architecture diagram of the GitHub repository quanxiny/rcr-featureenvy, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-03). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/quanxiny/rcr-featureenvy
- Repository: https://github.com/quanxiny/rcr-featureenvy
- Video tour (about a minute): https://gitdiagram.com/quanxiny/rcr-featureenvy/video

## Overview

RCR-FeatureEnvy is a research implementation for CG-LSMN feature-envy detection, comparing static contrastive objectives and project-incremental continual learning. The main workflow loads graph-pair data, trains a baseline or contrastive/continual variant, evaluates detection metrics, and aggregates experiment results. Continual streams preserve project/case splits and support replay and distillation; relation-aware losses provide an additional training objective. The README documents experiment orchestration and frozen results, but source excerpts do not expose every runtime connection, so only documented or indexed relationships are shown.

Main entry points: [`scripts/run_contrastive.py`](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_contrastive.py) (Contrastive runner), [`scripts/run_continual.py`](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_continual.py) (Continual runner), [`scripts/run_baseline.py`](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_baseline.py) (Baseline runner), [`src/data/original_adapter.py`](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/original_adapter.py) (Dataset adapter)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Data preparation"]
  node_adapter["Dataset adapter"]
  node_splits["Fixed splits<br/>[fixed_splits.py]"]
  node_metadata["Metadata builder"]
  node_anchors["Directed anchors"]
  node_tokenstats["Token statistics"]
  node_sampler["Contrastive sampler"]
end

subgraph group_model["Detection model"]
  node_modelcore["CG-LSMN detector<br/>[cglsmn_baseline.py]"]
  node_projection["Pair projection"]
end

subgraph group_objectives["Training objectives"]
  node_focal["Focal objective<br/>[original_loss.py]"]
  node_supcon["Supervised contrastive"]
  node_localmatch["Local relation contrast"]
  node_directedloss["Directed relation loss"]
end

subgraph group_continual["Continual learning"]
  node_continualstream["Project stream builder<br/>[stream.py]"]
  node_continualtrain["Continual trainer<br/>[training.py]"]
end

subgraph group_experiments["Experiments and analysis"]
  node_baseline["Baseline runner<br/>[run_baseline.py]"]
  node_contrastive["Contrastive runner<br/>[run_contrastive.py]"]
  node_continualrun["Continual runner<br/>[run_continual.py]"]
  node_metrics["Evaluation metrics<br/>[metrics.py]"]
  node_checkpoint["Checkpoint writer<br/>[checkpoint.py]"]
  node_aggregation["Result aggregation"]
  node_continualaggregation["Continual aggregation"]
  node_summaries["Statistical summaries"]
  node_orchestration["Experiment orchestration"]
end

node_researcher(("Researcher"))
node_dataset[("CG-LSMN dataset")]
node_results[("Frozen results")]

node_researcher -->|"runs baseline"| node_baseline
node_researcher -->|"runs contrastive"| node_contrastive
node_researcher -->|"runs continual"| node_continualrun
node_dataset -->|"loads graphs"| node_adapter
node_adapter -->|"uses split labels"| node_splits
node_metadata -->|"uses fixed splits"| node_splits
node_baseline -->|"loads data"| node_adapter
node_baseline -->|"trains detector"| node_modelcore
node_baseline -->|"uses objective"| node_focal
node_baseline -->|"saves checkpoint"| node_checkpoint
node_baseline -->|"evaluates"| node_metrics
node_contrastive -->|"loads data"| node_adapter
node_contrastive -->|"selects folds"| node_splits
node_contrastive -->|"builds anchors"| node_anchors
node_contrastive -->|"uses token stats"| node_tokenstats
node_contrastive -->|"samples pairs"| node_sampler
node_contrastive -->|"trains detector"| node_modelcore
node_contrastive -->|"uses projection"| node_projection
node_contrastive -->|"uses objective"| node_focal
node_contrastive -->|"uses objective"| node_supcon
node_contrastive -->|"uses objective"| node_localmatch
node_contrastive -->|"uses objective"| node_directedloss
node_contrastive -->|"saves checkpoint"| node_checkpoint
node_contrastive -->|"evaluates"| node_metrics
node_continualrun -->|"loads stream"| node_continualstream
node_continualrun -->|"loads data"| node_adapter
node_continualrun -->|"uses anchors"| node_anchors
node_continualrun -->|"uses token stats"| node_tokenstats
node_continualrun -->|"trains detector"| node_modelcore
node_continualrun -->|"uses projection"| node_projection
node_continualrun -->|"trains stages"| node_continualtrain
node_continualrun -->|"saves checkpoint"| node_checkpoint
node_continualtrain -->|"computes loss"| node_directedloss
node_continualtrain -->|"computes loss"| node_supcon
node_continualtrain -->|"computes metrics"| node_metrics
node_continualaggregation -->|"writes summaries"| node_results
node_aggregation -->|"aggregates runs"| node_results
node_summaries -->|"summarizes evidence"| node_results
node_orchestration -->|"plans experiments"| node_contrastive
node_researcher -->|"launches plan"| node_orchestration

click node_adapter "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/original_adapter.py"
click node_splits "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/fixed_splits.py"
click node_metadata "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/metadata_builder.py"
click node_anchors "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/directed_anchors.py"
click node_tokenstats "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/token_statistics.py"
click node_sampler "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/contrastive_sampler.py"
click node_modelcore "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/models/cglsmn_baseline.py"
click node_projection "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/models/contrastive_head.py"
click node_focal "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/original_loss.py"
click node_supcon "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/supervised_contrastive.py"
click node_localmatch "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/local_matching_contrastive.py"
click node_directedloss "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/directed_relation_contrastive.py"
click node_continualstream "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/continual/stream.py"
click node_continualtrain "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/continual/training.py"
click node_baseline "https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_baseline.py"
click node_contrastive "https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_contrastive.py"
click node_continualrun "https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_continual.py"
click node_metrics "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/utils/metrics.py"
click node_checkpoint "https://github.com/quanxiny/rcr-featureenvy/blob/main/src/utils/checkpoint.py"
click node_aggregation "https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/aggregate_results.py"
click node_continualaggregation "https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/aggregate_continual.py"
click node_summaries "https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/summarize_task2_detection_sci.py"
click node_orchestration "https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/orchestrate_task2_detection_sci.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_adapter,node_splits,node_metadata,node_anchors,node_tokenstats,node_sampler toneBlue
class node_modelcore,node_projection,node_dataset,node_results toneAmber
class node_focal,node_supcon,node_localmatch,node_directedloss toneMint
class node_continualstream,node_continualtrain toneRose
class node_baseline,node_contrastive,node_continualrun,node_metrics,node_checkpoint,node_aggregation,node_continualaggregation,node_summaries,node_orchestration,node_researcher toneIndigo
```

## Components

### Data preparation

- [Dataset adapter](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/original_adapter.py): `src/data/original_adapter.py`
- [Fixed splits](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/fixed_splits.py): `src/data/fixed_splits.py`
- [Metadata builder](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/metadata_builder.py): `src/data/metadata_builder.py`
- [Directed anchors](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/directed_anchors.py): `src/data/directed_anchors.py`
- [Token statistics](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/token_statistics.py): `src/data/token_statistics.py`
- [Contrastive sampler](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/data/contrastive_sampler.py): `src/data/contrastive_sampler.py`

### Detection model

- [CG-LSMN detector](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/models/cglsmn_baseline.py): `src/models/cglsmn_baseline.py`
- [Pair projection](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/models/contrastive_head.py): `src/models/contrastive_head.py`

### Training objectives

- [Focal objective](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/original_loss.py): `src/losses/original_loss.py`
- [Supervised contrastive](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/supervised_contrastive.py): `src/losses/supervised_contrastive.py`
- [Local relation contrast](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/local_matching_contrastive.py): `src/losses/local_matching_contrastive.py`
- [Directed relation loss](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/losses/directed_relation_contrastive.py): `src/losses/directed_relation_contrastive.py`

### Continual learning

- [Project stream builder](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/continual/stream.py): `src/continual/stream.py`
- [Continual trainer](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/continual/training.py): `src/continual/training.py`

### Experiments and analysis

- [Baseline runner](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_baseline.py): `scripts/run_baseline.py`
- [Contrastive runner](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_contrastive.py): `scripts/run_contrastive.py`
- [Continual runner](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/run_continual.py): `scripts/run_continual.py`
- [Evaluation metrics](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/utils/metrics.py): `src/utils/metrics.py`
- [Checkpoint writer](https://github.com/quanxiny/rcr-featureenvy/blob/main/src/utils/checkpoint.py): `src/utils/checkpoint.py`
- [Result aggregation](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/aggregate_results.py): `scripts/aggregate_results.py`
- [Continual aggregation](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/aggregate_continual.py): `scripts/aggregate_continual.py`
- [Statistical summaries](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/summarize_task2_detection_sci.py): `scripts/summarize_task2_detection_sci.py`
- [Experiment orchestration](https://github.com/quanxiny/rcr-featureenvy/blob/main/scripts/orchestrate_task2_detection_sci.py): `scripts/orchestrate_task2_detection_sci.py`

### Other components

- **Researcher**
- **CG-LSMN dataset**
- **Frozen results**

## Connections

- Researcher → Baseline runner: runs baseline (evidence: `README.md`)
- Researcher → Contrastive runner: runs contrastive (evidence: `README.md`)
- Researcher → Continual runner: runs continual (evidence: `README.md`)
- CG-LSMN dataset → Dataset adapter: loads graphs (evidence: `src/data/original_adapter.py`)
- Dataset adapter → Fixed splits: uses split labels (evidence: `src/data/original_adapter.py`)
- Metadata builder → Fixed splits: uses fixed splits (evidence: `src/data/metadata_builder.py`)
- Baseline runner → Dataset adapter: loads data (evidence: `scripts/run_baseline.py`)
- Baseline runner → CG-LSMN detector: trains detector (evidence: `scripts/run_baseline.py`)
- Baseline runner → Focal objective: uses objective (evidence: `scripts/run_baseline.py`)
- Baseline runner → Checkpoint writer: saves checkpoint (evidence: `scripts/run_baseline.py`)
- Baseline runner → Evaluation metrics: evaluates (evidence: `scripts/run_baseline.py`)
- Contrastive runner → Dataset adapter: loads data (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Fixed splits: selects folds (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Directed anchors: builds anchors (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Token statistics: uses token stats (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Contrastive sampler: samples pairs (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → CG-LSMN detector: trains detector (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Pair projection: uses projection (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Focal objective: uses objective (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Supervised contrastive: uses objective (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Local relation contrast: uses objective (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Directed relation loss: uses objective (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Checkpoint writer: saves checkpoint (evidence: `scripts/run_contrastive.py`)
- Contrastive runner → Evaluation metrics: evaluates (evidence: `scripts/run_contrastive.py`)
- Continual runner → Project stream builder: loads stream (evidence: `scripts/run_continual.py`)
- Continual runner → Dataset adapter: loads data (evidence: `scripts/run_continual.py`)
- Continual runner → Directed anchors: uses anchors (evidence: `scripts/run_continual.py`)
- Continual runner → Token statistics: uses token stats (evidence: `scripts/run_continual.py`)
- Continual runner → CG-LSMN detector: trains detector (evidence: `scripts/run_continual.py`)
- Continual runner → Pair projection: uses projection (evidence: `scripts/run_continual.py`)
- Continual runner → Continual trainer: trains stages (evidence: `scripts/run_continual.py`)
- Continual runner → Checkpoint writer: saves checkpoint (evidence: `scripts/run_continual.py`)
- Continual trainer → Directed relation loss: computes loss (evidence: `src/continual/training.py`)
- Continual trainer → Supervised contrastive: computes loss (evidence: `src/continual/training.py`)
- Continual trainer → Evaluation metrics: computes metrics (evidence: `src/continual/training.py`)
- Continual aggregation → Frozen results: writes summaries (evidence: `README.md`)
- Result aggregation → Frozen results: aggregates runs (evidence: `README.md`)
- Statistical summaries → Frozen results: summarizes evidence (evidence: `README.md`)
- Experiment orchestration → Contrastive runner: plans experiments (evidence: `README.md`)
- Researcher → Experiment orchestration: launches plan (evidence: `README.md`)
