# lmarena/search-arena architecture

> Architecture diagram of the GitHub repository lmarena/search-arena, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-02). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/lmarena/search-arena
- Repository: https://github.com/lmarena/search-arena

## Overview

Search Arena is a research-reproduction repository for analyzing search-augmented LLM conversations, user preferences, intent, and citation grounding. Researchers start with the published battle datasets or processed comparison sets, then use annotation scripts and analysis notebooks to produce classifications, grounding labels, statistical comparisons, and visual results. The graph preserves the documented analysis branches as well as the excerpted annotation stages. Notebook internals were not sampled, so their relationships are based on README descriptions rather than inspected implementation.

Main entry points: [`data/search_arena_24k.jsonl`](https://github.com/lmarena/search-arena/blob/main/data/search_arena_24k.jsonl) (Search battles), [`scripts/intent_classification/annotation.py`](https://github.com/lmarena/search-arena/blob/main/scripts/intent_classification/annotation.py) (Intent annotation), [`scripts/citation_attribution_analysis/annotation.py`](https://github.com/lmarena/search-arena/blob/main/scripts/citation_attribution_analysis/annotation.py) (Citation attribution), [`scripts/preference_analysis.ipynb`](https://github.com/lmarena/search-arena/blob/main/scripts/preference_analysis.ipynb) (Preference analysis)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Research data"]
  node_search_data[("Search battles")]
  node_text_data[("Text battles")]
  node_diff_data[("Difference sets")]
  node_intent_labels["Intent labels"]
  node_citation_inputs["Citation inputs"]
  node_citation_labels[("Grounding results")]
end

subgraph group_annotation["Annotation workflows"]
  node_intent_annotation["Intent annotation<br/>[annotation.py]"]
  node_intent_prompts["Intent prompts<br/>[utils.py]"]
  node_citation_annotation["Citation attribution<br/>[annotation.py]"]
  node_citation_rules["Citation rules<br/>[utils.py]"]
end

subgraph group_analysis["Analysis workflows"]
  node_eda["Prompt exploration<br/>[data_eda.ipynb]"]
  node_preference["Preference analysis"]
  node_cross_setting["Cross-setting analysis"]
  node_analysis_utils["Battle statistics<br/>[utils.py]"]
  node_citation_analysis["Citation analysis"]
  node_intent_analysis["Intent analysis"]
end

node_researcher(("Researcher"))

node_researcher -->|"runs annotation"| node_intent_annotation
node_search_data -->|"provides prompts"| node_intent_annotation
node_intent_annotation -->|"builds requests"| node_intent_prompts
node_intent_annotation -->|"produces labels"| node_intent_labels
node_researcher -->|"runs evaluation"| node_citation_annotation
node_citation_inputs -->|"provides traces and content"| node_citation_annotation
node_citation_annotation -->|"uses instructions"| node_citation_rules
node_citation_annotation -->|"writes results"| node_citation_labels
node_search_data -->|"supplies prompts"| node_eda
node_researcher -->|"explores"| node_eda
node_search_data -->|"supplies battles"| node_preference
node_researcher -->|"analyzes preferences"| node_preference
node_preference -.->|"uses statistics"| node_analysis_utils
node_search_data -->|"supports comparisons"| node_cross_setting
node_text_data -->|"supports comparisons"| node_cross_setting
node_researcher -->|"compares settings"| node_cross_setting
node_diff_data -->|"enables comparisons"| node_researcher
node_citation_labels -->|"supports analysis"| node_citation_analysis
node_intent_labels -.->|"supports analysis"| node_intent_analysis

click node_search_data "https://github.com/lmarena/search-arena/blob/main/data/search_arena_24k.jsonl"
click node_text_data "https://github.com/lmarena/search-arena/blob/main/data/text_arena_battles.jsonl"
click node_diff_data "https://github.com/lmarena/search-arena/tree/main/data/describing_diffs"
click node_intent_annotation "https://github.com/lmarena/search-arena/blob/main/scripts/intent_classification/annotation.py"
click node_intent_prompts "https://github.com/lmarena/search-arena/blob/main/scripts/intent_classification/utils.py"
click node_citation_annotation "https://github.com/lmarena/search-arena/blob/main/scripts/citation_attribution_analysis/annotation.py"
click node_citation_rules "https://github.com/lmarena/search-arena/blob/main/scripts/citation_attribution_analysis/utils.py"
click node_citation_labels "https://github.com/lmarena/search-arena/blob/main/data/citation_attribution_data.jsonl"
click node_eda "https://github.com/lmarena/search-arena/blob/main/scripts/data_eda.ipynb"
click node_preference "https://github.com/lmarena/search-arena/blob/main/scripts/preference_analysis.ipynb"
click node_cross_setting "https://github.com/lmarena/search-arena/blob/main/scripts/cross_setting_analysis.ipynb"
click node_analysis_utils "https://github.com/lmarena/search-arena/blob/main/scripts/utils.py"
click node_citation_analysis "https://github.com/lmarena/search-arena/tree/main/scripts/citation_attribution_analysis"
click node_intent_analysis "https://github.com/lmarena/search-arena/tree/main/scripts/intent_classification"

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_search_data,node_text_data,node_diff_data,node_intent_labels,node_citation_inputs,node_citation_labels toneBlue
class node_intent_annotation,node_intent_prompts,node_citation_annotation,node_citation_rules toneAmber
class node_eda,node_preference,node_cross_setting,node_analysis_utils,node_citation_analysis,node_intent_analysis toneMint
class node_researcher toneIndigo
```

## Components

### Research data

- [Search battles](https://github.com/lmarena/search-arena/blob/main/data/search_arena_24k.jsonl): `data/search_arena_24k.jsonl`
- [Text battles](https://github.com/lmarena/search-arena/blob/main/data/text_arena_battles.jsonl): `data/text_arena_battles.jsonl`
- [Difference sets](https://github.com/lmarena/search-arena/tree/main/data/describing_diffs): `data/describing_diffs`
- **Intent labels**
- **Citation inputs**
- [Grounding results](https://github.com/lmarena/search-arena/blob/main/data/citation_attribution_data.jsonl): `data/citation_attribution_data.jsonl`

### Annotation workflows

- [Intent annotation](https://github.com/lmarena/search-arena/blob/main/scripts/intent_classification/annotation.py): `scripts/intent_classification/annotation.py`
- [Intent prompts](https://github.com/lmarena/search-arena/blob/main/scripts/intent_classification/utils.py): `scripts/intent_classification/utils.py`
- [Citation attribution](https://github.com/lmarena/search-arena/blob/main/scripts/citation_attribution_analysis/annotation.py): `scripts/citation_attribution_analysis/annotation.py`
- [Citation rules](https://github.com/lmarena/search-arena/blob/main/scripts/citation_attribution_analysis/utils.py): `scripts/citation_attribution_analysis/utils.py`

### Analysis workflows

- [Prompt exploration](https://github.com/lmarena/search-arena/blob/main/scripts/data_eda.ipynb): `scripts/data_eda.ipynb`
- [Preference analysis](https://github.com/lmarena/search-arena/blob/main/scripts/preference_analysis.ipynb): `scripts/preference_analysis.ipynb`
- [Cross-setting analysis](https://github.com/lmarena/search-arena/blob/main/scripts/cross_setting_analysis.ipynb): `scripts/cross_setting_analysis.ipynb`
- [Battle statistics](https://github.com/lmarena/search-arena/blob/main/scripts/utils.py): `scripts/utils.py`
- [Citation analysis](https://github.com/lmarena/search-arena/tree/main/scripts/citation_attribution_analysis): `scripts/citation_attribution_analysis`
- [Intent analysis](https://github.com/lmarena/search-arena/tree/main/scripts/intent_classification): `scripts/intent_classification`

### Other components

- **Researcher**

## Connections

- Researcher → Intent annotation: runs annotation (evidence: `README.md`)
- Search battles → Intent annotation: provides prompts (evidence: `scripts/intent_classification/annotation.py`)
- Intent annotation → Intent prompts: builds requests (evidence: `scripts/intent_classification/annotation.py`)
- Intent annotation → Intent labels: produces labels (evidence: `README.md`)
- Researcher → Citation attribution: runs evaluation (evidence: `README.md`)
- Citation inputs → Citation attribution: provides traces and content (evidence: `scripts/citation_attribution_analysis/annotation.py`)
- Citation attribution → Citation rules: uses instructions (evidence: `scripts/citation_attribution_analysis/annotation.py`)
- Citation attribution → Grounding results: writes results (evidence: `scripts/citation_attribution_analysis/annotation.py`)
- Search battles → Prompt exploration: supplies prompts (evidence: `README.md`)
- Researcher → Prompt exploration: explores (evidence: `README.md`)
- Search battles → Preference analysis: supplies battles (evidence: `README.md`)
- Researcher → Preference analysis: analyzes preferences (evidence: `README.md`)
- Preference analysis → Battle statistics: uses statistics
- Search battles → Cross-setting analysis: supports comparisons (evidence: `README.md`)
- Text battles → Cross-setting analysis: supports comparisons (evidence: `README.md`)
- Researcher → Cross-setting analysis: compares settings (evidence: `README.md`)
- Difference sets → Researcher: enables comparisons (evidence: `README.md`)
- Grounding results → Citation analysis: supports analysis (evidence: `README.md`)
- Intent labels → Intent analysis: supports analysis
