Citation Network Dataset: DBLP+Citation, ACM Citation network

10,268,730 papers and 104,607,995 citation relationships (2026-09-10) Overview

The data set is designed for research purpose only. The citation data is extracted from DBLP, ACM, MAG (Microsoft Academic Graph), and other sources. The first version contains 629,814 papers and 632,752 citations. Each paper is associated with abstract, authors, year, venue, and title.

The data set can be used for clustering with network and side information, studying influence in the citation network, finding the most influential papers, topic modeling analysis, etc.

The significant decrease in the number of citations in the ACM-Citation-network V12 and DBLP-Citation-network V18 is due to the restriction of paper citations to within the dataset itself.

Citation-network V1:  629,814 papers and >632,752 citation relationships (2010-05-15).

Citation-network V2:  1,397,240 papers and >3,021,489 citation relationships (2010-09-13).

DBLP-Citation-network V3:  1,632,442 papers and >2,327,450 citation relationships (2010-10-22).

DBLP-Citation-network V4:  1,511,035 papers and 2,084,019 citation relationships (2011-01-08).

DBLP-Citation-network V5:  [download from mirror site] 1,572,277 papers and 2,084,019 citation relationships (2011-01-08).

DBLP-Citation-network V6:  2,084,055 papers and 2,244,018 citation relationships (2013-09-29).

DBLP-Citation-network V7:  [download from mirror site]  2,244,021 papers and 4,354,534 citation relationships (2014-05-25).

DBLP-Citation-network V8:  3,272,991 papers and 8,466,859 citation relationships (2016-07-14)

ACM-Citation-network V8:   2,381,688 papers and 10,476,564 citation relationships (2016-04-02)

ACM-Citation-network V9:    2,385,022 papers and 9,671,893 citation relationships (2017-01-20)

DBLP-Citation-network V9: 3,680,007 papers and 1,876,067 citation relationships (2017-07-03)

DBLP-Citation-network V10:  3,079,007 papers and 25,166,994 citation relationships (2017-10-27)

DBLP-Citation-network V11:  4,107,340 papers and 36,624,464 citation relationships (2019-05-05)

DBLP-Citation-network V12:  4,894,081 papers and 45,564,149 citation relationships (2020-04-09)

DBLP-Citation-network V13:  5,354,309 papers and 48,227,950 citation relationships (2021-05-14)

DBLP-Citation-network V14:  5,259,858 papers and 36,630,661 citation relationships (2023-01-31)

ACM-Citation-network V10:   2,579,904 papers and 17,254,368 citation relationships (2023-11-22)

DBLP-Citation-network V15:  6,275,271 papers and 42,104,228 citation relationships (2023-11-22)

ACM-Citation-network V11:   3,101,332 papers and 18,630,025 citation relationships (2024-02-08)

DBLP-Citation-network V16:  6,404,473 papers and 62,492,559 citation relationships (2024-02-08)

ACM-Citation-network V12:   2,952,025 papers and 1,911,278 citation relationships (2025-02-13)

DBLP-Citation-network V17:  6,614,726 papers and 5,103,745 citation relationships (2025-02-13)

DBLP-Citation-network V18:  6,733,427 papers and 73,702,758 citation relationships (2025-07-14)

ACM-Citation-network V13:   2,979,553 papers and 19,593,436 citation relationships (2026-09-10)

DBLP-Citation-network V19:  7,289,177 papers and 85,014,559 citation relationships (2026-09-10)

Table 1: Dataset statistics

The smaller dataset is organized into ~600,000 blocks, each for a paper. [Download]
The larger one is organized into ~2,000,000 blocks, each also for a paper. [Download]
The DBLP version is organized into 1,511,035 blocks, each also for a paper. [Download]

For each block, each line starting with a specific prefix indicates an attribute of the paper. More specifically,

For V1-V4, V7, V8, V9

#* --- paperTitle
#@ --- Authors
#t ---- Year
#c  --- publication venue
#index 00---- index id of this paper
#% ---- the id of references of this paper (there are multiple lines, with each indicating a reference)
#! --- Abstract

The following is an example:

#*Information geometry of U-Boost and Bregman divergence
#@Noboru Murata,Takashi Takenouchi,Takafumi Kanamori,Shinto Eguchi
#t2004
#cNeural Computation
#index436405
#%94584
#%282290
#%605546
#%620759
#%564877
#%564235
#%594837
#%479177
#%586607
#!We aim at an extension of AdaBoost to U-Boost, in the paradigm to build a stronger classification machine from a set of weak learning machines. A geometric understanding of the Bregman divergence defined by a generic convex function U leads to the U-Boost method in the framework of information geometry extended to the space of the finite measures over a label set. We propose two versions of U-Boost learning algorithms by taking account of whether the domain is restricted to the space of probability functions. In the sequential step, we observe that the two adjacent and the initial classifiers are associated with a right triangle in the scale via the Bregman divergence, called the Pythagorean relation. This leads to a mild convergence property of the U-Boost algorithm as seen in the expectation-maximization algorithm. Statistical discussions for consistency and robustness elucidate the properties of the U-Boost methods based on a stochastic assumption for training data.

The dataset can be downloaded [V1 (smaller): here] [V2 (larger): here]. [V3 (DBLP):  here] [V4 (DBLP + ACM citation): tar.bz2] [V7 (AMiner citation): zip]

#* --- paperTitle
#@ --- Authors
#year ---- Year
#conf --- publication venue
#citation --- citation number (both -1 and 0 means none)
#index ---- index id of this paper
#arnetid ---- pid in arnetminer database
#% ---- the id of references of this paper (there are multiple lines, with each indicating a reference)
#! --- Abstract

The following is an example:

#*Spatial Data Structures.
#@Hanan Samet
#year1995
#confModern Database Systems
#citation2743
#index25
#arnetid27
#%165
#!An overview is presented of the use of spatial data structures in spatial data
bases. The focus is on hierarchical data structures, including a number of varia
nts of quadtrees, which sort the data with respect to the space occupied by it.
Such techniques are known as spatial indexing methods. Hierarchical data structu
res are based on the principle of recursive decomposition. They are attractive b
ecause they are compact and depending on the nature of the data they save space
as well as time and also facilitate operations such as search. Examples are give
n of the use of these data structures in the representation of different data ty
pes such as regions, points, rectangles, lines, and volumes.

[V5 (DBLP + ACM citation): tar.bz2]   [download from mirror site]

[V6 (DBLP + ACM citation): rar]

If the file format is ".tar.bz2", please use "tar -jxvf" to extract the file.

[V7 (DBLP + ACM citation / separated by domains: zip]   [download from mirror site]

We provide a set of citation sub-networks separated by domains in the "domains" folder.

In each text file, each line represents a paper, which is in JSON schema. The data schema is:

                                                           Table 2: data schema (v10)

Below is an example:

{
 "authors": [
   "Leon A. Sakkal",
   "Kyle Z. Rajkowski",
   "Roger S. Armen"
 ],
 "n_citation": 0,
 "references": [
   "4f4f200c-0764-4fef-9718-b8bccf303dba",
   "aa699fbf-fabe-40e4-bd68-46eaf333f7b1"
 ],
 "title": "Prediction of consensus binding mode geometries for related chemical series of positive allosteric modulators of adenosine and muscarinic acetylcholine receptors",
 "venue": "Journal of Computational Chemistry",
 "year": 2017,
 "id": "013ea675-bb58-42f8-a423-f5534546b2b1"
}

For V11

In each text file, each line represents a paper, which is in JSON schema. The data schema is shown in Table 3 below.

For V12

The dataset is organized in a JSON file. You can load it as a whole in JSON format, or you can load it line by line (except the first and last line, each line is a paper dictionary).

The data schema is shown in Table 3 below. This version includes the following attributes: id, title, authors.name, authors.org, authors.id, venue.id, venue.raw, year, fos.name, fos.w, references, n_citation, page_start, page_end, doc_type, publisher, volume, issue, doi, indexed_abstract. Id involved in V12 (paper id, author id, venue id) is represented as Long type (instead of string).

For V13

Backed to v11 schema, where id and references are in String form.

                                                        Table 3: data schema (v11+)

Note: The paper fields of study are extracted from MAG papers. The data schema can be found in reference data schema and FAQs.

Table 4:  data schema (v14)

Table 5: data schema (v15)

If you use this data set for research, please must cite the following paper:

• Jie Tang, Jing Zhang, Limin Yao, Juanzi Li, Li Zhang, and Zhong Su. ArnetMiner: Extraction and Mining of Academic Social Networks. In Proceedings of the Fourteenth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (SIGKDD'2008). pp.990-998. [PDF] [Slides] [System] [API]

You please also consider referring to the following papers:

• Jie Tang, Limin Yao, Duo Zhang, and Jing Zhang. A Combination Approach to Web User Profiling. ACM Transactions on Knowledge Discovery from Data (TKDD), (vol. 5 no. 1), Article 2 (December 2010), 44 pages.  [PDF]
• Jie Tang, A.C.M. Fong, Bo Wang, and Jing Zhang. A Unified Probabilistic Framework for Name Disambiguation in Digital Library. IEEE Transaction on Knowledge and Data Engineering (TKDE) , Volume 24, Issue 6, 2012, Pages 975-987. [PDF][Code&Data&System]
• Jie Tang, Jing Zhang, Ruoming Jin, Zi Yang, Keke Cai, Li Zhang, and Zhong Su. Topic Level Expertise Search over Heterogeneous Networks. Machine Learning Journal, Volume 82, Issue 2 (2011), Pages 211-237. [PDF] [URL]
• Jie Tang, Duo Zhang, and Limin Yao. Social Network Extraction of Academic Researchers. In Proceedings of 2007 IEEE International Conference on Data Mining(ICDM'2007). pp. 292-301. [PDF] [Slides] [Data]
• Arnab Sinha, Zhihong Shen, Yang Song, Hao Ma, Darrin Eide, Bo-June (Paul) Hsu, and Kuansan Wang. 2015. An Overview of Microsoft Academic Service (MAS) and Applications. In Proceedings of the 24th International Conference on World Wide Web (WWW ’15 Companion). ACM, New York, NY, USA, 243-246. [PDF][System][API]  ***For V10, V11, V12***

@INPROCEEDINGS{Tang:08KDD,

   AUTHOR = "Jie Tang and Jing Zhang and Limin Yao and Juanzi Li and Li Zhang and Zhong Su",

   TITLE = "ArnetMiner: Extraction and Mining of Academic Social Networks",

   pages = "990-998",

   YEAR = {2008},

   BOOKTITLE = "KDD'08",

}

@article{Tang:10TKDD,

    author = {Jie Tang and Limin Yao and Duo Zhang and Jing Zhang},

    title = {A Combination Approach to Web User Profiling},

    journal = {ACM TKDD},

    year = {2010},

    volume = {5},

    number = {1},

   pages = {1--44},

}

@article{Tang:11ML,

    author = {Jie Tang and Jing Zhang and Ruoming Jin and Zi Yang and Keke Cai and Li Zhang and Zhong Su},

    title = {Topic Level Expertise Search over Heterogeneous Networks},

    year = {2011},

    volume = {82},

    number = {2},

    pages = {211--237},

    journal = {Machine Learning Journal},

}

@article{Tang:12TKDE,

   author = {Jie Tang and Alvis C.M. Fong and Bo Wang and Jing Zhang},

   title = {A Unified Probabilistic Framework for Name Disambiguation in Digital Library},

   journal ={IEEE Transactions on Knowledge and Data Engineering},

   volume = {24},

   number = {6},

   year = {2012},

   pages = {975-987},

}

@INPROCEEDINGS{Tang:07ICDM,

   AUTHOR = "Jie Tang and Duo Zhang and Limin Yao",

   TITLE = "Social Network Extraction of Academic Researchers",

   PAGES = "292-301",

   YEAR = {2007},

   BOOKTITLE = "ICDM'07",

}

@inproceedings{sinha2015overview,
 title={An overview of microsoft academic service (mas) and applications},
 author={Sinha, Arnab and Shen, Zhihong and Song, Yang and Ma, Hao and Eide, Darrin and Hsu, Bo-june Paul and Wang, Kuansan},
 booktitle={Proceedings of the 24th international conference on world wide web},
 pages={243--246},
 year={2015},
 organization={ACM}
}