Dgraph: large scale dynamic graph dataset

In recent days, , Yang Yang's scientific research group of Zhejiang University （yangy.org） Hexin also jointly released a large-scale dynamic graph data set DGraph, Aimed at service graph neural network 、 Graph mining 、 Social networks 、 Researchers in the direction of anomaly detection , Provide large-scale data of real scenes .DGraph On the one hand, it can be used as the standard data to verify the performance of the correlation graph model , On the other hand, it can also be used to carry out user portrait 、 Network analysis and other research work .

Dataset home page ：

https://dgraph.xinye.com/

Github:

https://github.com/DGraphXinye/

Related papers ：

DGraph: A Large-Scale Financial Dataset for Graph Anomaly Detection. Xuanwen Huang, Yang Yang*, Yang Wang, Chunping Wang, Zhisheng Zhang, Jiarong Xu, and Lei Chen. Preprint, 2022. (http://yangy.org/works/dgraph/dgraph_2022.pdf)

Data set description

DGraph The source data of is provided by Xinye Technology .DGraph It is a directed dynamic graph with no right , Contains more than 370 Ten thousand nodes and 430 Ten thousand dynamic edges . As shown in the figure below ,DGraph The node in represents the financial lending user of Xinye technology service , A directed edge indicates an urgent contact relationship , Each node contains the attribute characteristics after desensitization , And a label indicating whether it is a financial fraud user .

Data features

The scene is real

DGraph It comes from the real financial business scenario , Its construction logic is close to the industrial landing , It provides an opportunity for users of data sets to explore how to extend the graph model to the financial field . To be specific ,DGraph The proportion of abnormal and normal users in is about 1:100, Its “ The label is unbalanced ” The characteristics of the are in line with the real scene , Support exception detection 、 Research on classification of unbalanced nodes .

Structural dynamics

DGraph User relationships in are sampled from across 27 A business scenario for months , And the network structure will evolve over time , It provides data support for the current dynamic graph model and mining research .

Large scale

DGraph contain 370 Thousands of desensitized real financial lending users and 430 Ten thousand dynamic relationships , Its scale is about the largest dynamic graph data in the financial field Elliptic Of 17 times , Support the research and evaluation of large-scale graph models . Besides ,DGraph Contained in the 60% Of “ Background node ”, That is, it is not a classification or analysis object, but it actually exists 、 Nodes that have an indirect impact on business logic . These nodes play an important role in maintaining the connectivity of the network , Widely exists in industry . Reasonable processing of background nodes can effectively improve the storage space of data and the operation efficiency of the model in large-scale data scenarios .DGraph It contains more than 200 10000 background nodes , It can support researchers to explore the properties of background nodes .

Open source community maintenance

Ranking List

DGraph Users can submit at any time 、 Refreshed performance leaderboard （leaderboard）, To track the research progress of the latest graph model . The list provides a unified evaluation process , All results are open and transparent .

Research results

DGraph It has rich characteristics , Support graph research in multiple directions .

Algorithm contest

Xinye technology revolves around DGraph The seventh Xinye Technology Cup algorithm competition was held , Task and DGraph The fraud user identification in is consistent . The competition is open to the whole society , Colleges and universities at home and abroad 、 Scientific research institutes 、 Internet enterprises can sign up for the competition , The bonus pool is abundant , total 31 Thousands of yuan .

Welcome interested colleagues to patronize DGraph Public data website , Work together to provide rich application data for the field of artificial intelligence , Work together to build an open digital ecosystem .

Cooperation platform

Read more

# cast draft   through Avenue #

  Let your words be seen by more people  

How to make more high-quality content reach the reader group in a shorter path , How about reducing the cost of finding quality content for readers ？ The answer is ： People you don't know .

There are always people you don't know , Know what you want to know .PaperWeekly Maybe it could be a bridge , Push different backgrounds 、 Scholars and academic inspiration in different directions collide with each other , There are more possibilities . 

PaperWeekly Encourage university laboratories or individuals to , Share all kinds of quality content on our platform , It can be Interpretation of the latest paper , It can also be Analysis of academic hot spots 、 Scientific research experience or Competition experience explanation etc. . We have only one purpose , Let knowledge really flow .

  The basic requirements of the manuscript ：

• The article is really personal Original works , Not published in public channels , For example, articles published or to be published on other platforms , Please clearly mark  

• It is suggested that  markdown  Format writing , The pictures are sent as attachments , The picture should be clear , No copyright issues

• PaperWeekly Respect the right of authorship , And will be adopted for each original first manuscript , Provide Competitive remuneration in the industry , Specifically, according to the amount of reading and the quality of the article, the ladder system is used for settlement

  Contribution channel ：

• Send email ：[email protected] 

• Please note your immediate contact information （ WeChat ）, So that we can contact the author as soon as we choose the manuscript

• You can also directly add Xiaobian wechat （pwbot02） Quick contribution , remarks ： full name - contribute

△ Long press add PaperWeekly Small make up

Now? , stay 「 You know 」 We can also be found

Go to Zhihu home page and search 「PaperWeekly」

Click on 「 Focus on 」 Subscribe to our column

·

·