citation-network-builder

本条目面向文献与学术工具场景,提供“citation-network-builder”技能的目录入口。公开标签:learning、引文网络、文献计量、网络分析。请通过固定上游版本核对实际功能、依赖与权限;此页为 Skillπ 独立编写的用途介绍和原站链接,不代表运行安全保证。原文与附件依各自许可开放。

仓库
brycewang-stanford/Auto-Empirical-Research-Skills
仓库最近推送
Stars
5,023
Forks
578
Issues / PRs
0

安装方式

先用 Prompt 核对来源,也可切换到安装命令或下载可用的安装包。

安装方式

用 AI 助手安装

复制提示词,让助手先核对固定版本和依赖。

查看 Prompt收起 Prompt
先检查当前 Agent 是否已安装 SkillPI Skill Finder。如果尚未安装,请按官方仓库的说明安装并启用:https://github.com/Findskills/skillpi。然后使用 Skill Finder 打开此 SkillPI 详情页并定位 "citation-network-builder":https://skillpi.com/zh/skills/deepskill-0d498ef993f96084fad88a135b11b83270279aa4c5c0203b622b0afac1c1d83a。安装前核对上游来源和版本:https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills/blob/bfa0a6e5a843babbf45a1ef1663fd5dc4f07c0ba/skills/43-wentorai-research-plugins/skills/tools/knowledge-graph/citation-network-builder/SKILL.md。来源不一致时停止并说明差异。核对通过后,用 Skill Finder 安装目标 Skill,并报告安装路径和来源版本。

此命令从 GitHub 来源仓库获取指定技能。运行前请先查看源文件。

运行工具
npx skills add https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills/tree/bfa0a6e5a843babbf45a1ef1663fd5dc4f07c0ba/skills/43-wentorai-research-plugins/skills/tools/knowledge-graph/citation-network-builder --skill citation-network-builder
许可协议: —

来源与配置

查看 GitHub 源码

brycewang-stanford/Auto-Empirical-Research-Skills固定提交 bfa0a6e5a843babbf45a1ef1663fd5dc4f07c0ba

配置方式待核实

请查看原始来源中的适用工具、依赖和权限要求。

安装实测待补充

本条目面向文献与学术工具场景,提供“citation-network-builder”技能的目录入口。公开标签:learning、引文网络、文献计量、网络分析。请通过固定上游版本核对实际功能、依赖与权限;此页为 Skillπ 独立编写的用途介绍和原站链接,不代表运行安全保证。原文与附件依各自许可开放。

↓ 0 下载 ↗ 0 分享点击

文件浏览

在原站浏览文件 ↗

原始 SKILL.md

打开原文 ↗
---
name: citation-network-builder
description: "Build and analyze citation networks from academic reference data"
metadata:
  openclaw:
    emoji: "🕸️"
    category: "tools"
    subcategory: "knowledge-graph"
    keywords: ["citation network", "bibliometrics", "graph analysis", "co-citation", "bibliographic coupling", "network visualization"]
    source: "wentor-research-plugins"
---

# Citation Network Builder

A skill for constructing, analyzing, and visualizing citation networks from academic reference data. Covers data collection from bibliographic databases, network construction using direct citation, co-citation, and bibliographic coupling methods, community detection for identifying research clusters, and practical visualization with tools like Gephi, VOSviewer, and Python NetworkX.

## Data Collection and Preparation

### Source Databases

Citation network analysis requires structured bibliographic data with reference lists. The choice of database determines coverage and available metadata.

```
Database Comparison for Citation Analysis:

Web of Science (Clarivate):
  - Format: ISI/WoS plain text, BibTeX, CSV
  - Coverage: ~21,000 journals, back to 1900
  - Strengths: Cited reference data is most complete
  - Limits: 1,000 records per export, subscription required
  - Best for: High-quality citation network analysis

Scopus (Elsevier):
  - Format: CSV, BibTeX, RIS
  - Coverage: ~27,000 journals, back to 1970s for most
  - Strengths: Broader coverage than WoS, author IDs
  - Limits: 2,000 records per export, subscription required
  - Best for: Broader disciplinary coverage

OpenAlex (free):
  - Format: JSON via REST API
  - Coverage: ~250M works, all disciplines
  - Strengths: Free, open, comprehensive, API access
  - Limits: Reference linking less complete than WoS
  - Best for: Large-scale analysis, reproducible research

CrossRef (free):
  - Format: JSON via REST API
  - Coverage: ~150M DOIs across all publishers
  - Strengths: Free, authoritative DOI metadata, reference linking
  - Limits: No abstract text, citation counts may lag
  - Best for: Cross-publisher networks, DOI resolution
```

### Data Cleaning for Network Construction

```python
import pandas as pd

def clean_bibliographic_data(records):
    """
    Clean and deduplicate bibliographic records for network construction.

    Steps:
    1. Standardize DOIs (lowercase, strip prefixes)
    2. Deduplicate by DOI, then by title similarity
    3. Parse reference lists into structured format
    4. Filter records missing key fields
    """
    # Standardize DOIs
    records["doi"] = (
        records["doi"]
        .str.lower()
        .str.replace("https://doi.org/", "", regex=False)
        .str.replace("http://dx.doi.org/", "", regex=False)
        .str.strip()
    )

    # Remove duplicates by DOI
    records = records.drop_duplicates(subset="doi", keep="first")

    # Filter records without references (cannot build citation links)
    records = records[records["references"].notna()]
    records = records[records["references"].str.len() > 0]

    return records
```

## Network Construction Methods

### Direct Citation Network

The simplest approach: paper A cites paper B creates a directed edge from A to B.

```python
import networkx as nx

def build_direct_citation_network(records):
    """
    Build a directed citation network.
    Nodes = papers, Edges = citation relationships.

    Args:
        records: DataFrame with 'doi' and 'references' columns
                 where 'references' is a list of cited DOIs
    Returns:
        NetworkX DiGraph
    """
    G = nx.DiGraph()

    for _, row in records.iterrows():
        citing_doi = row["doi"]
        G.add_node(citing_doi, title=row.get("title", ""),
                   year=row.get("year", None))

        for ref_doi in row["references"]:
            G.add_edge(citing_doi, ref_doi)

    return G
```

### Co-Citation Network

Two papers are co-cited when a third paper cites both. Co-citation strength is the number of papers that cite both. This method identifies intellectual relationships between cited works.

```python
from itertools import combinations
from collections import Counter

def build_cocitation_network(records, min_cocitations=2):
    """
    Build an undirected co-citation network.
    Nodes = cited papers, Edges = co-citation frequency.
    """
    pair_counts = Counter()

    for _, row in records.iterrows():
        refs = sorted(set(row["references"]))
        for a, b in combinations(refs, 2):
            pair_counts[(a, b)] += 1

    G = nx.Graph()
    for (a, b), count in pair_counts.items():
        if count >= min_cocitations:
            G.add_edge(a, b, weight=count)

    return G
```

### Bibliographic Coupling Network

Two papers are bibliographically coupled when they share one or more references. This method groups papers with similar theoretical or methodological foundations.

```python
def build_bibliographic_coupling_network(records, min_shared=3):
    """
    Build an undirected bibliographic coupling network.
    Nodes = citing papers, Edges = number of shared references.
    """
    ref_sets = {}
    for _, row in records.iterrows():
        ref_sets[row["doi"]] = set(row["references"])

    G = nx.Graph()
    dois = list(ref_sets.keys())
    for i in range(len(dois)):
        for j in range(i + 1, len(dois)):
            shared = len(ref_sets[dois[i]] & ref_sets[dois[j]])
            if shared >= min_shared:
                G.add_edge(dois[i], dois[j], weight=shared)

    return G
```

## Network Analysis

### Key Metrics

```
Node-level metrics:
  - In-degree (direct citation): number of times a paper is cited
    -> identifies influential papers
  - Betweenness centrality: how often a node lies on shortest paths
    -> identifies bridging papers connecting subfields
  - PageRank: iterative importance score based on who cites the paper
    -> identifies papers cited by other influential papers

Network-level metrics:
  - Density: proportion of possible edges that exist
  - Clustering coefficient: tendency of nodes to form triangles
  - Average path length: mean shortest path between node pairs
  - Number of connected components: isolated clusters
```

### Community Detection

Community detection algorithms identify clusters of densely connected papers, corresponding to research subfields or intellectual traditions.

```python
import community as community_louvain

def detect_communities(G):
    """
    Detect communities using the Louvain algorithm.
    Returns a dictionary mapping node -> community_id.
    """
    partition = community_louvain.best_partition(G, weight="weight")

    # Summarize communities
    communities = {}
    for node, comm_id in partition.items():
        communities.setdefault(comm_id, []).append(node)

    for comm_id, members in sorted(communities.items()):
        print(f"Community {comm_id}: {len(members)} papers")

    return partition
```

## Visualization

### Tool Recommendations

```
Gephi (desktop application):
  - Best for: Interactive exploration of medium networks (1k-50k nodes)
  - Layout algorithms: ForceAtlas2, Fruchterman-Reingold
  - Export: SVG, PDF, PNG
  - Workflow: Import GEXF/GraphML -> layout -> partition by community
              -> adjust sizes by centrality -> export

VOSviewer (desktop application):
  - Best for: Bibliometric networks specifically
  - Direct import from WoS/Scopus export files
  - Built-in clustering and overlay visualizations
  - Limitation: less customizable than Gephi

Python (matplotlib, pyvis):
  - Best for: Reproducible, scriptable visualizations
  - Use pyvis for interactive HTML network graphs
  - Use matplotlib for static publication-quality figures
```

Citation network analysis provides a quantitative lens on the structure of scientific knowledge, revealing invisible colleges, emerging research fronts, and foundational works that shape entire disciplines.

版本历史

  1. v2026.07 GitHub ↗
    原版更新说明

    ## Highlights - **First tagged release.** Every gate is green: `make check` (validate + 183 unit tests + eval-harness + numeric benchmarks), validate-catalog, quality-evals, and OpenSSF Scorecard on `main`. - **Rigor double loop at 13 method families.** Each family closes taxonomy tag → eval scenario (does the agent avoid the method's known traps?) → deterministic numeric benchmark (does it recover the true effect?). New this cycle: heterogeneous treatment effects (CATE) and quantile treatment effects (QTE). - **Paper-WorkFlow flagship at 29/29 executable rigor gates**, now merged to its `main` with a drift-gated README badge — the credibility artifact is checked by CI, so it can never silently go stale ([RIGOR.md](https://github.com/brycewang-stanford/Paper-WorkFlow/blob/main/RIGOR.md)). - **1,150 vendored skills / 69 collections**, all license-audited, security-scanned, tracked in a machine-readable catalog with CI-enforced freshness, and importable as a single whole-repo router skill. - **Hardened weekly upstream syncs**: synced trees must pass the full gate before merge; this cycle both sync PRs were verified, patched (StatsPAI's SkillOpt execution-gate card restored), and merged. - **First-party plugin marketplace** (Python / R / Stata empirical-analysis plugins + guide) shipped in the previous cycle, now under a tagged version. ## Known Follow-Ups - Ecosystem benchmark comparison (AERS vs econometrics agents on the shared benchmark suite) — [PLAN-2026-07 week 3](https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills/blob/main/docs/PLAN-2026-07.md). - Decomposition (Oaxaca–Blinder) and shift-share (Bartik) method families (weeks 3–4). - GitHub Pages catalog site, README demo screenshots, five-language README consistency script (week 4). ## Catalog - Top-level collections: **69** - `SKILL.md` files: **1150** - Cataloged agent tools: **335** ([`tools/CATALOG.md`](../tools/CATALOG.md)) ## Methodological rigor - Method families in the coverage map: **13** ([`RIGOR_COVERAGE.md`](RIGOR_COVERAGE.md)) - Eval scenarios: **26** (138 rubric items, 122 auto-checkable) ([`../eval-harness/`](../eval-harness/README.md)) - Benchmark tasks: **13** (70 gold checks, 58 required) ([`../benchmark/`](../benchmark/README.md)) ### Method families (skills tagged) - `iv`: 27 - `rdd`: 25 - `panel-fe`: 19 - `staggered-did`: 17 - `bayesian`: 13 - `did`: 10 - `event-study`: 10 - `matching`: 10 - `synthetic-control`: 10 - `cate`: 8 - `dml`: 8 - `survival`: 3 - `quantile`: 2 ### Eval scenarios by category - `causal-identification`: 14 - `citation-hygiene`: 1 - `reproducibility`: 2 - `research-integrity`: 1 - `runtime-safety`: 1 - `statistical-validity`: 2 - `writing-compliance`: 3 - `writing-style`: 2 ## License buckets - AGPL-3.0: 1 - Apache-2.0: 1 - CC-BY-NC-4.0 (non-commercial): 1 - CC-BY-SA-4.0: 1 - CC-BY-SA-4.0 (repository default): 4 - GPL-3.0: 1 - MIT: 35 - MIT Non-Commercial: 1 - Mixed (per-skill; see folder): 2 - UNKNOWN - check upstream: 22 ## Quality gates - `make check` — catalog/provenance/audit/eval freshness, repo hygiene, Paper-WorkFlow gates, stdlib unit tests, eval-harness lint + graded smoke, and the strict numeric benchmark (reference-gold drift fails the build). - External links and OpenSSF Scorecard run on their scheduled workflows.

当前收录版本

0.0.1

核对来源版本 ↗

质量评测

满分 100。前四项来自 DeepSkill 收录的质量报告。

场景价值92/100
评分依据

提供完整的学术引用网络构建与分析端到端解决方案

触发精准95/100
评分依据

触发词精准完整,覆盖 citation network、bibliometrics、graph analysis 等多个变体

结构规范93/100
评分依据

文档结构完整,Phase 划分清晰(数据收集、网络构建、分析、可视化)

执行可靠91/100
评分依据

工具来源明确(NetworkX, community_louvain, Gephi, VOSviewer)

证据可验 (前四项均值)93/100
计算方式

取前四项原始分值的算术平均,四舍五入取整。可通过收录报告核对各项评分与说明,运行记录需单独查证。

权限与使用前检查

根据当前主文档自动整理,命中内容可能是示例或防护说明;不是实际权限声明,也未覆盖附件、外部依赖或运行行为。

未从主文档识别到上述常见操作,不能据此判断无需权限或没有风险。

文档引用的域名或地址

这些地址可能用于文档链接或示例,不是已批准的联网清单。使用前核对实际目的地及发送的数据。

  • doi.org
  • dx.doi.org

本站展示提示不会授予权限。涉及文件改动、密钥、外部发送或费用时,仍由你使用的工具按需确认。

安全评级与访问范围

risk · 静态证据核验,不代表运行安全保证。

使用前请由宿主核对并授权实际文件读写、外部服务和联网目的地;本站的核验不会授予任何运行权限。

来源与使用声明 SkillPi 的部分技能资料来自公开来源或创作者提交。第三方内容的权利归各自权利人所有。安装或运行前,请查看原站的许可、权限要求和使用风险。

来源与使用

请按详情页的作者、原始来源和许可证核对使用条件,运行前检查技能需要的权限。目录收录不等于已验证安装或安全性。