=Paper=
{{Paper
|id=Vol-1489/paper-05
|storemode=property
|title=Linking Accounts across Social Networks: the Case of StackOverflow, Github and Twitter
|pdfUrl=https://ceur-ws.org/Vol-1489/paper-05.pdf
|volume=Vol-1489
|dblpUrl=https://dblp.org/rec/conf/kdweb/SilvestriYBT15
}}
==Linking Accounts across Social Networks: the Case of StackOverflow, Github and Twitter==
https://ceur-ws.org/Vol-1489/paper-05.pdf
Linking Accounts across Social Networks: the
Case of StackOverflow, Github and Twitter
Giuseppe Silvestri1,2 , Jie Yang1 , Alessandro Bozzon1 , and Andrea Tagarelli2
1
Delft University of Technology, the Netherlands
giuseppe.silvestri.gios@gmail.com, {j.yang-3, a.bozzon}@tudelft.nl
2
University of Calabria, Italy
andrea.tagarelli@unical.it
Abstract. Social Web accommodates a wide spectrum of user activities,
including information sharing via social media networks (e.g., Twitter),
question answering in collaborative Q&A systems (e.g., StackOverflow),
and more profession-oriented activities such as social coding in code shar-
ing systems (e.g., Github). Social Web enables the distinctive opportu-
nity for understanding the interplay between multiple user activity types.
To enable such studies, a basic requirement, and a big challenge, is the
ability to link user profiles across multiple social networks.
By exploiting user attributes, platform-specific services, and different
matching strategies, this paper contributes a methodology for linking
user accounts across StackOverflow, Github and Twitter. We show how
tens of thousands of accounts in StackOverflow, Github, and Twitter
could be successfully linked. To showcase the type of research enabled
by datasets built with our methodology, we conduct a comparative study
of user interaction networks in the three platforms, and investigate cor-
relations between users interactions across the different networks.
1 Introduction
Social Web comprises a diversity of social networking platforms, which cover
a wide range of user activities. With the fact that a single user has multiple
accounts across different social networks, it has now become increasingly impor-
tant to link distributed user profiles belonging to the same user from multiple
sources, which can benefit various applications. For instance, it has been shown
that aggregating user profiles could improve personalized Web service such as
recommendation systems by solving the cold-start problem [1].
Linking user profiles across multiple social networks also provides an oppor-
tunity for better understanding the interplay between different types of people’s
activities. Let us take as an instance the domain of software programmers: they
share software related content in Twitter, seek or provide answers to software en-
gineering related questions in StackOverflow, and collaboratively code in Github.
These three different social networks (i.e., Twitter, StackOverflow and Github)
are used by programmers differently, in terms of their purposes and correspond-
ingly their activities. By aggregating the data sources from multiple networks,
we might explore at large scale the complete spectrum of programmers’ on-line
professional activities.
Linking users’ accounts across multiple social networks is considered a well-
known problem, thus attracting multiple techniques and solutions [1, 6–8, 4, 5].
Previous studies addressed the online activities of professional users, but investi-
gated a single type of activities in a single system [6, 7], or between two systems
from a single perspective. For instance, [8] analyzes how participation in Q&A
systems influences developers’ productivity. [2] also considered the influence that
each user has within and across two platforms, while exploiting features provided
by StackOverflow (Up Votes and Questions) and Github (popular users are en-
gaged more in commits, projects and issues). [3] focused on bridge users, in order
to recognize how these users can favor information exchange across networks.
To drive a deeper investigation over users’ professional activities, we are
motivated to construct a cross-system users’ accounts matching dataset from
Twitter, StackOverflow, and Github, to enable future studies of professional ac-
tivities from multiple perspectives. For instance, a dataset as such can help us
understand how different types of users (e.g., users with different expertise) are
engaged in different professional activities; it can also help in understanding how
different types of social interactions among users can influence the evolution of
communities of different professional activities. This paper contributes a method-
ology to link online users’ accounts across Twitter, StackOverflow and Github,
by exploiting different attributes of user profiles, platform specific API’s and
services, and a variety of accounts’ matching strategies. As a first trail of valu-
ing this dataset, we construct three social networks, including follower-followee
networks of Twitter and Github, and helper-helpee networks of StackOverflow.
By characterizing the networks features, we present our findings of how users
interact with others in different activities, and how different activities of the
same user correlate with each other.
The rest of the paper is organised as follows. Section 2 describes our method-
ology of matching users across StackOverflow, Github and Twitter, together with
the corresponding results of user matching. Based on these matched users, Sec-
tion 3 introduces our comparative study of user interactions between three user
interaction networks in StackOverflow, Github ad Twitter, and Section 4 con-
cludes our work.
2 Linking Accounts across Social Networks
This section describes our methodology of matching users across StackOverflow,
Github and Twitter. We first discuss the general settings of data retrieval for
the three social networking platforms, then present our user matching strategies
and workflows.
2.1 Retrieving data from multiple platforms
StackOverflow. We downloaded the most recently released data dump from
Internet archive3 . Due to privacy concern, since the end of 20144 StackOverflow
3
https://archive.org/details/stackexchange, accessed at April, 2015
4
http://meta.stackexchange.com/questions/221027/where-did-emailhash-go
data dump no longer contains hashed user emails. While not crucial, hashed
emails are a convenient and effective way to unambiguously match accounts. To
overcome this limitation, we extended the data from the data dump released on
September 2013 (which is the last released dump with hashed email addresses)
with the latest data contained in the 2015 data dump.
Github. The GHTorrent project5 has incrementally released Github data every
two months since March 2012. We parsed its data from the first release containing
user information (i.e., July, 2012) until the latest one on March 2015, and kept
all versions of user information in our database for account matching.
Twitter. Given an existing user name, the related account information (e.g.,
profile picture, website) and related posts in Twitter can be retrieved via Twitter
REST API6 . The Twitter.com Search7 functionality, on the other hand, allows
for fuzzy retrieval of users accounts, returning a candidate set of accounts having
screen names similar to the one provided as input. For our purposes, the latter
proved more useful than the former for fuzzy matching.
Fig. 1: Data collection workflow. SO, GH, TW are short for StackOverflow, Github,
and Twitter, respectively.
The main workflow of accounts’ linking across the three platforms is depicted
in Figure 1. Accounts from StackOverflow and Github were dumped and pro-
cessed first. We retrieved 4,132,407 and 4,288,132 accounts in StackOverflow and
Github, respectively. These sets of accounts were then matched to each other and
the resulting overlap further matched to a set of Twitter accounts. The latter
was retrieved using a strategy that we will discuss later in this section.
5
http://ghtorrent.org, accessed at April, 2015
6
https://dev.twitter.com/rest/public
7
https://twitter.com/search-home
2.2 Matching accounts across multiple platforms
We design three account matching strategies to find the same set of users in the
platforms under study:
– explicit matching, which aims at identifying the links explicitly provided
by users in one platform to their accounts in other platforms for user match-
ing.
– attribute-based matching, which leverages unique attributes of users’
accounts such as email to connect profiles across multiple platforms from
the same user.
– fuzzy matching, which exploits less accurate user attributes such as login
names and profile images to match user profiles.
Explicit matching is performed to link user accounts between StackOverflow and
Github, and further link them to Twitter. Attribute-based matching is performed
only between StackOverflow and Github, while fuzzy matching aims at linking
matched users in StackOverflow and Github to Twitter. We introduce as follows
the concrete steps we took for each of the matching strategies.
Explicit matching. Starting from our built dumps of StackOverflow and
Github, we perform explicit matching by analyzing user-provided links from the
user profiles in each of these platforms to the other platforms. We consider this
a very reliable method for account linking: matching information are provided
by users themselves, with strong incentives for truthful linking.
From StackOverflow to Github, Twitter. We analyze StackOverflow user pro-
files to find explicit links to GitHub and Twitter users. For StackOverflow users
that provide links to their Github link, we parse the direct links, which are in
the form of https://github.com/GitHubLoginName and obtain their Github
login names, i.e., GithubLoginName. For StackOverflow users that provide di-
rect links to Twitter, which is usually in the form of http://www.twitter.com/
TwitterScreenName, we parse the Twitter screen name, i.e., TwitterScreenName.
Both GitHub login name and Twitter screen name uniquely identifies one user
in GitHub and Twitter, respectively.
From Github to StackOverflow, Twitter. We analyze Github user profiles sim-
ilarly to match user profiles in StackOverflow and Twitter. For StackOver-
flow, we adopt an additional strategy to obtain a cross-reference to the same
user: since some Github users provide their StackOverflow Careers profile8 ,
which is a CV-like page of senior StackOverflow users, we parse the HTML
code of the corresponding pages in order to retrieve the direct link (in the
form http://stackoverflow.com/users/id) to their real StackOverflow profile
pages.
The result of explicit matching is reported in Table 1. As it can be observed,
we are able to match thousands of users between the three platforms.
8
http://careers.stackoverflow.com/, StackOverflow Careers
From To #Matched Users
Github 4,536
StackOverflow
Twitter 10,068
StackOverflow 433
Github
Twitter 7,012
Table 1: Explicit matching.
Attribute-based matching. StackOverflow and Github provide users with
the option of registering their emails, which are encrypted into MD5 hashes in
the data dumps. This technique is known from literature [8, 2] to be a reliable
way to match users by their email reference.
There are in total 2,185,162 (≈52.9%) StackOverflow users and 510,523
(≈11.9%) Github users with email hash. Note that email hashes were previ-
ously considered for matching users between StackOverflow and Github in [8].
Besides using the email hashes explicitly provided by users, we exploit Gravatar9
to increase the number of available hashes in both platforms. We find that many
users use Gravatar to have a unique profile image across StackOverflow and
Github. By making HTTP request for a Gravatar profile image, we obtain a
user’s MD5 email hash 10 . We identified 2,897,175 (≈67.6%) Github users, and
430,860 (≈10.4%) StackOverflow users with Gravatar email hash available.
query =((StackOverf lowU sers[emailhash] ∩ GithubU sers[emailhash])
∪(StackOverf lowU sers[gravatarid] ∩ GithubU sers[gravatarid])
(1)
∪(StackOverf lowU sers[emailhash] ∩ GithubU sers[gravatarid])
∪(StackOverf lowU sers[gravatarid] ∩ GithubU sers[emailhash]))
Combing email hashes explicitly provided by users, and implicitly revealed
from their Gravatar Id, we use Query 1 for StackOverflow-Github user matching,
which encodes all meaningful joins between MD5 email hash and Gravatar Id
attributes across the two platforms. The result of attribute-based matching is
shown in Table 2. We finally obtained more than 600k exactly matched users
between StackOverflow and Github.
Fuzzy matching. Matching accounts from StackOverflow and Github with
Twitter accounts is intrinsically more difficult, since Twitter profiles need to be
obtained via Twitter API services.
Lookup and search. Two types of query requests are here considered, namely
Twitter REST API and Twitter.com Search, hereinafter referred to as Lookup
9
https://en.gravatar.com/ Gravatar, a globally recognized avatar.
10
https://en.gravatar.com/site/implement/images/ Gravatar: Image Request
Type #Matched Users
SO emailhash - GH gravatarid 580,979
SO emailhash - GH emailhash 107,572
SO gravatarid - GH emailhash 1,224
SO gravatarid - GH gravatarid 4,752
Union all above types 604,083
Table 2: Attribute-based matching between StackOverflow and Github.
and Search, respectively. The former method returns the full profile information
of the user corresponding to a given user screen name. Using Twitter REST
API, each request can process up to 100 inputs. By contrast, Twitter.com Search
permits to process only a single input for each request. While being less efficient,
Twitter.com Search is however more flexible in terms of the input — it accepts
any textual input.
We consider the following options of input for the Search method:
– login names, and names of users’ StackOverflow and Github accounts;
– URLs of user’s StackOverflow and Github accounts;
– users’ website URLs identified from their StackOverflow and Github profiles.
To find the best input for the Search method, we analyzed how many accounts
can be matched by using different user attributes. Matching is performed in two
steps: (1) given a user attribute, retrieve candidate users via Twitter.com Search;
(2) try matching the website URL of the Twitter candidates and the website
URL of the user StackOverflow (Github) profile. Results have shown that using
Github login name provides better matching of Twitter profiles than the URLs
of their accounts in StackOverflow or Github, as well as their website URLs.
We therefore chose to take Github login name as an input for Search to retrieve
candidate Twitter profiles for matching.
Accuracy of Lookup and Search methods. To assess the performance of
the Lookup and Search matching methods, we first categorized the Github login
names into the following categories: 1) the login contains only lower-case char-
acters, 2) it contains at least one upper-case character, 3) it contains numbers,
and 4) it contains special characters. Figure 2a shows the distribution of Github
login names according to the categorization above, from which we observe that
the majority of them are in the “lower-case” category.
To understand how different categories differ in the probability that at least
one candidate can be returned by Lookup and Search, in Figure 2b we analyzed
the percentages of Github login names that have at least one candidate returned
by Lookup and Search. High values indicate higher probability that the user can
be matched. We observe that the Search method performs better than Lookup
in all categories except in the “Number” category.
0.8
1
loginnames
with
>=
1
candidates
returned
Percent
of
#GH
0.8
loginnames
0.6
Percent
of
GH
0.6
0.4
0.4
Lookup
0.2
0.2
Search
0
0
Special
Number
Upper
Lower
Special
Number
Upper
Lower
Category
Category
(a) (b)
Fig. 2: Distributions of Github login names in selected categories (a) and number of
candidates returned by Lookup and Search for each category (b).
Category Lookup Search Gain
Lower .29 .39 +.10
Upper .28 .33 +.05
Number .24 .27 +.03
Special .00 .19 +.19
Table 3: Accuracy of Twitter user matching using Lookup and Search for different
categories of Github login name.
For each category, we randomly selected 100 Github login names, took them
as input for both Lookup and Search methods, then manually checked the matched
accounts. A user is considered to be matched with a Twitter account if there
is explicit Twitter information (e.g., personal website, profile description) that
can identify the user with high confidence. Table 3 shows that Search performs
better than Lookup, especially for Github login name that belong to the “lower-
case” and “special characters” categories. The least gain of Search over Lookup
corresponds to the category “Number” (less than 5%). Considering Figure 2b
and the higher efficiency of Lookup method, we chose to use Lookup for Github
login names in the “Number” category, and Search for the other categories.
Workflow of fuzzy matching. Figure 3 depicts the workflow of Lookup and
Search methods. Given a user Github login name, it first determines whether to
use Lookup or Search, then checks Twitter profiles for account matching. In the
step of ”Twitter Profile Check”, a user is matched to a Twitter account if s/he
satisfies the following criteria:
1. the website attribute of the user’s Twitter profile is exactly the same as the
website of his/her StackOverflow (Github) profile;
2. otherwise, the Twitter profile picture needs to be highly similar (e.g., ≥ 90%)
to her/his profile picture in StackOverflow (Github).
In criterion 1 we ignored ambiguous websites such as http://facebook.
com, which can bring to have False Positive for website matching, while for
Fig. 3: Twitter Lookup and Search workflows.
criterion 2 we performed image similarity via image hashing 11 . We manually
checked 100 users matched by website and profile picture, respectively. As before,
user profiles were considered as matched if the provided information gives high
confidence that they belong to the same user. The true positive rate of website
and profile pictures are 100% and 98%, respectively, which indicate that users
can be regarded as exactly matched.
Search method #Users analyzed #Users matched Matching %
Lookup 176,508 9,316 5.28%
Search 240,000 37,449 18.43%
Total 416,508 46,765 11.23%
Table 4: Twitter user matching results.
To account for limitations with the Twitter APIs, at the time of this writing
we were able to analysis a subset of linked accounts from StackOverflow and
Github. We ordered accounts according to their popularity (measured by #fol-
11
http://hzqtc.github.io/2013/04/image-duplication-detection.html, Image
Duplication Detection
Graph # Nodes # Edges Density
GSO 6672 18995 4.267e-04
GGH 13160 106792 6.167e-04
GT W 16070 829846 2.213e-03
Table 5: Characteristics of the user networks in Twitter, StackOverflow, and Github.
lowees) in Github, and matched them to Twitter accordingly. Table 4 reports
the user matching results. We analyzed 416k accounts, specifically 240k by using
Search and 176k by using Lookup. The number of accounts matched are 37k and
9k, respectively, with a total of 46k accounts matched to Twitter.
3 User Interaction across Networks
To showcase the type of research that is enabled by a dataset built with our
methodology, we designed a study aimed at providing an answer to the follow-
ing two research questions: RQ1: how do users connect with each other in dif-
ferent social networks? RQ2: does the relative importance of users vary across
social networks? To this end, we first inferred the interaction networks over the
same set of users in the three platforms, then analyzed network features and
correlations of user centrality in the three networks.
Building user interaction networks. We constructed two directed graphs
GT W , GGH that encode following relationships of users in Twitter and Github,
respectively, i.e., a directed edge e = u → v indicates that user u follows user
v. While being absent of explicit following-followee relationship, StackOverflow
provides an implicit ”help network” among users according to who answers to
whom. Therefore, we built a directed graph GSO such that an edge e = u → v
indicates that user u is helped by v, i.e., at least one question of u is answered
by v.
Due to the rate limit of Twitter REST API, we built the three user interaction
network graphs for the 20k most popular users among the 46k matched users
(Table 4). As before, popularity is defined according to #followees in Github.
RQ1: How do users connect with each other in different social net-
works? Table 5 reports basic statistics of the users’ networks in the considered
social networks. By comparing the #nodes in the three networks, we observe
that, in the same set of 20k users, more users are involved in both Github and
Twitter interaction networks than those involved in StackOverflow interaction
network. This indicates that users are more likely to be active in explicit inter-
action based on followship than in helping-based interaction.
Comparing the density of these networks, results show that users have similar
connection intensity in StackOverflow and Github, both of which are however
(a) StackOverflow in- and out- degree. (b) Github in- and out-degree.
(c) Twitter in- and out-degree.
Fig. 4: Degree distribution for Twitter, StackOverflow and Github networks.
10 times lower than user interaction in Twitter. This would imply that users are
more likely to connect with each other in general-purpose social networks like
Twitter than in profession-oriented networks like StackOverflow and Github.
Figure 4 shows the in-degree and out-degree distributions over the three net-
works. In StackOverflow, both distributions conforms to power-law, indicating
that most users follow (resp. are followed by) a small number of users, while
there is a small number of users that follow (resp. are followed by) many users.
In addition, in-degree distribution looks more skewed than out-degree distribu-
tion – in other words, users tend to follow the same set of users, who is followed
by many users. Similarly in Github and Twitter, in-degree distribution is more
skewed than out-degree distribution, indicating that a small number of users are
highly popular in the network. Comparing the three networks, StackOverflow
is the one that has most similar distributions of in-degree and out-degree. We
consider the fact that the StackOverflow helping-helpee network is built implic-
itly from question-answering activity between users, while the following-followee
relations in Github and Twitter are explicitly constructed by users. The result
suggests that explicit connection mechanisms result in a more skewed popularity
among the users of a platform.
RQ2: does the relative importance of users vary across social net-
works? To answer this question, we choose to correlate users’ centrality scores
(a) StackOveflow - Github (b) StackOveflow - Twitter
(c) Github - Twitter
Fig. 5: Pair-wise network centrality correlations.
in the different networks. A high cross-network correlation of user centrality
scores would indicate similar user importance in different settings; for instance,
a high correlation of user centrality in StackOverflow and Github networks will
suggest that a user who is helpful in answering to others’ questions in StackOver-
flow will be followed by many users in Github (and vice versa); on the contrary,
a low correlation would indicate that users’ activity in one platform is not in-
dicative of their activities in another platform, e.g., an influential user in Github
may not likely to answer questions in StackOverflow.
To obtain users’ centrality values, we used classic PageRank model. We then
calculated Pearson correlation of the centrality values for the same set of users
in every pair of graphs. Results are shown in Figure 5. For StackOveflow and
Github networks, we have a Pearson coefficient of -0.0185170 that reveals no
linear correlation between PageRank values of users on both platforms; this
means that, as shown in Figure 5a, most influential users on StackOverflow do
not have the same importance on Github and vice versa. Similar remark can be
made on StackOveflow versus Twitter, where Pearson correlation is -0.0014857.
By contrast, in the Github - Twitter case, we observe a Pearson coefficient of
0.7554060, which implies that user interactions of Github and Twitter networks
are correlated.
4 Conclusions and Future Works
We addressed the problem of user matching across StackOverflow, Github and
Twitter social networks. We proposed a methodology that combines different
matching strategies and makes use of different user attributes and platform-
specific services for linking user accounts. Many of the proposed linking strate-
gies can be generalized to other social networking platforms. For instance, most
social networking platforms provide REST API’s and search, for which the link-
ing techniques Lookup and Search can be applied. These methods together allow
us to obtain much better results than in literature. Our study of interaction net-
works based on the matched users in the three platforms has provided interesting
insights: 1) users in general-purpose social media networks like Twitter are more
connected than in profession-oriented social networks like StackOverflow and
Github; 2) social networking platforms that enable the functionality of explicit
user connection (Github and Twitter) will result in more skewed distribution of
user popularity, and more correlated user activities between them, than (with)
the one that only provides implicit user connection mechanisms (StackOverflow).
As part of future work, we plan to deepen our analysis of the user interaction
networks properties such as the formation and evolution of communities, and
the topics discussed by the users and communities across the three networks.
References
1. F. Abel, E. Herder, G.-J. Houben, N. Henze, and D. Krause. Cross-system user
modeling and personalization on the social web. User Modeling and User-Adapted
Interaction, 23(2-3):169–209, 2013.
2. A. S. Badashian, A. Esteki, A. Gholipour, A. Hindle, and E. Stroulia. Involvement,
contribution and influence in github and stack overflow. In CASCON ’14 Proceed-
ings of 24th Annual International Conference on Computer Science and Software
Engineering, pages 19–33. ACM, 2014.
3. F. Buccafurri, V. D. Foti, G. Lax, A. Nocera, and D. Ursino. Bridge analysis in a
social internetworking scenario. Inf. Sci., 224:1–18, 2013.
4. F. Buccafurri, G. Lax, A. Nocera, and D. Ursino. Discovering missing me edges
across social networks. Inf. Sci., 319:18–37, 2015.
5. P. Jain, P. Kumaraguru, and A. Joshi. @i seek ’fb.me’: Identifying users across
multiple online social networks. In Proceedings of the 22Nd International Confer-
ence on World Wide Web Companion, WWW ’13 Companion, pages 1259–1268,
Republic and Canton of Geneva, Switzerland, 2013. International World Wide Web
Conferences Steering Committee.
6. C. Treude, O. Barzilay, and M.-A. Storey. How do programmers ask and answer
questions on the web?: Nier track. In Software Engineering (ICSE), 2011 33rd
International Conference on, pages 804–807. IEEE, 2011.
7. J. Tsay, L. Dabbish, and J. Herbsleb. Influence of social and technical factors
for evaluating contribution in github. In Software Engineering (ICSE), 2014 36rd
International Conference on, pages 356–366. ACM, 2014.
8. B. Vasilescu, V. Filkov, and A. Serebrenik. Stackoverflow and github: associations
between software development and crowdsourced knowledge. In Social Computing
(SocialCom), 2013 International Conference on, pages 188–195. IEEE, 2013.