Walls-in-one: usage and temporal patterns in a social media aggregator

Nowadays people have at their disposal a wide selection of online social sites, each having its own peculiarity. This way, the adoption of multiple social platforms by the same person is becoming increasingly spread. Now people can exploit and combine their favorite social media according to their needs. For example, a tourist could share his/her position by Foursquare, meanwhile s/he uses Twitter to communicate his/her mood and shares on Instagram a selfie with the “Gioconda”.

At the same time, people have recently shown a growing interest in tools managing their online life in a centralized way, i.e. social media aggregators. Social media aggregators allow users to combine their own identities into a single profile by gathering their online activities from different social platforms, such as Twitter, YouTube, LinkedIn, Facebook, and many others. The aggregation is enabled by APIs provided by social networks, so that people may have control on the data the aggregation platform can access.

Most of these services offer a private space to the users, where they can share a content simultaneously on multiple media, but profiles with public contents and public API are not available. The media aggregator Alternion ³ stands out because, unlike most of similar services, allows users to decide which information about their social sites - including profile information and contents - can be made public. Furthermore, the service retrieves data from more than 200 social sites and manages social relationships among Alternion profiles.

Data collection methodology

Since the service does not expose a public API, we developed a crawler to retrieve the Alternion profiles and their public updates. An example of an Alternion profile page is shown in Fig. 1. The page exposes the social sites associated to the Alternion identity. Each icon links to the relative profile and shows the username chosen in the target social site. The "Updates" tab reports the public posts grouped by social sites. The content of each post highly depends on the information released by APIs.

We collected registered users by exploiting a search function which returns a different set of 40 random users every two minutes. Data collection started on October 2014 and resulted in 19.680 distinct profiles from different countries and in different languages. From a user profile we extract the set of social sites a user associates to the profile and all the public contents s/he had published since the registration date. For each content we gather information about the social media used to post it, the content itself according to the format returned by the APIs and the publication date. Furthermore, not to overload the service and to respect politeness, we limit the number of contents for each user to 10.000 at most. In practice, the dataset contains, for each individual, a list of her/is favorite social sites and a multidimensional time series of posts, where each dimension corresponds to a specific social site. All the analysis, but username patterns reported in the last section, will be performed on this dataset D1. In addition, each profile reports the degree of the user on the social networks that make this information available. However, we are not able to get the node neighborhoods since Alternion does not return the neighbors’ usernames. Consequently, we are not able to build any network from the information provided by the social media aggregator. Since we may be able to build different types of relation by exploiting implicit links such as mention, we model our dataset by a directed multigraph \(\mathcal {D}=(V,E,D)\). Each user (identity) is an element of V, while E⊆V×V×D, with D the set of social media in Alternion, is the set of directed multi-edges. Finally, we associate to each user u∈V a sequence of timestamped events (t,p e _d ), where pe represents a post published on the social media d∈D.

For the study of username patterns only, we collected a different dataset, D2, whose characteristics are specifically designed to this aim. We adopted a more targeted sampling approach by collecting 15.000 profiles with an English first name ⁴ to make the text analysis of usernames more affordable. This way we maintain the alphabet accordance between the username (typically ASCII character) and the information about the identity (first and last name are usually written in mother tongue). Finally, the dataset D2 contains for each individual a list of her/is usernames.

Dataset characteristics and representativity

The dataset D1⁵ contains the profile information of 19680 users, each reporting the platform Alternion among her/is social sites. Specifically the number of social platforms amounts to 152, from the most famous Facebook, Twitter to the less common Discus or Zazzle. In Fig. 2 we report the number of users who share their profile information of a social site, while in Fig. 3 we report the number of users for the top ten social platforms. As expected, and ignoring Alternion, Facebook is the most used social site and the top 10 correspond to the most popular social media.

In Fig. 3 we also report the number of active users per social site. Active user denotes a person who has published at least one content on the profile page. About half of the profiles (9829) are active and do prefer Twitter as publishing media. Although Twitter result is expected due to the Twitter intrinsic nature (interest and media broadcasting network), LinkedIn and Google+ are more favorite than Facebook when we consider active users, even if the continuous releases of new versions of the Facebook API may influence the available information. Anywise we generally note that the users who actively adopt social platforms always represent a fraction of the registered users.

We obtain a deeper understanding if we combine the above results with those presented in Fig. 4, where we report the number of posts grouped by social sites. Here we note that Facebook active users are more productive than Google+ users. In fact, although they are fewer, Facebook users publish more or less the same amount of posts as Google+’s. Finally, results in Fig. 4 confirm the predominant role of Twitter in the production of contents and posts, i.e. more than 6 million posts over an overall amount of more than 8.5 million posting events.

As regards the posting activities of the active users, in Fig. 5 we show the distribution of the number of published contents per user. The post sampling covers the time interval from 10 July 2005 to 10 November 2014. The distribution seems to obey to a heavy tail with exponential cut-off. During the examined period, we observe that users have been quite productive: on average, users have published about 800 different contents and half of the users have produced more than 300 updates.

The above results highlight that Alternion effectively captures a representative sample of today’s users of social media sites, in terms of favorite social sites and publishing media as compared with public available data¹ on social media usage. Indeed, the usage proportion across the social sites are almost equal to those collected by aggregating the official statistics released by the different social platforms. Despite data show an alignment between the Alternion users and the users of the different social media, the representativeness issue of the dataset still persists. For instance, users’ demography is unbalanced towards western and English-speaking countries, the contents posted on the Alternion profile depend on the API provided by the social media. Lastly, the adoption of social media aggregators is more common and useful to an audience active on social media. Nevertheless, it represents the most recent dataset able to capture how users behave across multiple sites and an easy tool for studying the on-going phenomenon of the multiple adoption of social platforms.

In a survey conducted in September 2014¹ it has been highlighted that the adoption of multiple social platforms is on the rise. In fact, 52 % of online users now use at least two social sites representing a significant increase from 2013, when the figures were close to 42 %. On the one hand the Alternion dataset allows us to quantify this phenomenon without requiring a periodical survey; on the other, additional information about posts could be extracted to verify whether the same observations hold for users’ engagement. Indeed, being registered to a service does not always imply using it.

To answer Q1 - How common is the use of multiple social sites? Does the same hold for active users? - we analyze how users are distributed across sites. The main goal of a social media aggregator is to group different social sites into a single access point. This characteristic allows us to compute how many different social websites people are able to manage. How users distribute their membership across sites has been studied in Zafarani and Liu (2014) only, where authors showed that it follows a power-law distribution from a dataset gathered in 2008. Our goal is to extend that result, by confirming it on a more recent and larger dataset, which relies on the current social platforms. In Fig. 6 we report the probability distribution function (PDF) of the number of sites joined by a given person. On average a user is simultaneously registered on 5 social sites, while more than 95 % of users have joined 17 platforms at most. Alternion results are consistent with the aforementioned survey; indeed about 56 % of Alternion users use at least three different social platforms⁶.

The behavior of active users gives a more direct and clear evidence of the adoption of multiple sites. In fact the publication of contents on a specific social media confirms its adoption in practice and it is strictly related to the user’s engagement. This way, we compute the distribution of the number of social platforms being simultaneously used by each active user. As shown in Fig. 6, the observations about the multiple usage get stronger: 73 % of active users publish on at least two social networks. In general, the propensity to adopt multiple social sites is becoming increasingly ingrained amongst online users, especially if we consider the effective publishing on different platforms.

The above analysis on the people’s propensity to be active on multiple social media does not take into account the amount of contents published on each platform. To this aim, we analyze the productivity of the users who join more the one site simultaneously. Let p ₁,p ₂,…,p _n indicate the number of posts of a user on the n sites where s/he is active and \(\bar {p}=\sum p_{i}/n\) the average number of posts per site. For each active user we compute \(\bar {p}\) and group them by the number of multiple sites they have joined; then we calculate the mean and the standard deviation of \(\bar {p}\) for each group. In Fig. 7 we report the above quantities as a function of the number of sites. The figure indicates that i) the more sites a user joins, less posts for site on average s/he publishes, and ii) users who post on few sites are more heterogeneous in their activity since the standard deviation decreases as the number of sites increases. We suppose that the first remark is a consequence of the limited human resources in terms of usable time and ability of handling multiple tasks at the same time Miritello et al. (2013).

Finally, in Fig. 8, we show the number of users who have joined different social sites by considering two classes: social media and social networks. The former denotes social sites where relationships are implicit, the latter indicates social sites where relationships are explicitly defined. These figures show that more than 90 % of users have registered to at most 5 sites and some of them share information on more than 10 social media. In general we observe that it is more likely to adopt social media than social networks.

To answer Q2 - Is a person’s centrality uniform, i.e. more or less equally distributed across the social sites in use? - we treat the user degree on different social sites as an index of centrality in the network. This information, reported on the Alternion profile, is not available for every site since just a few APIs allow to retrieve this information. So, we limit our analysis on the degree to Facebook, Twitter (in/out), LinkedIn and YouTube (in/out).

Q2 corresponds to verify if statistically significant correlations between the degrees of the same group of users in different sites exist. The presence of these correlations measures whether or not popular users in one site maintain their centrality across media. If we denote \({k_{u}^{s}}\) as the degree of the node u in the site s, we can evaluate the degree of correlation between pairwise social sites by adopting different methods Boccaletti et al. (2014). First we compute, for each pair of social sites, the joint distribution P(k ^s1,k ^s2) to obtain a characterization of the relations between the degree sequences. For example, in Fig. 9, where we show the joint distribution for Facebook and LinkedIn, we observe a slightly positive correlation, especially in the left bottom part of the distribution (dark blue regions). In particular people with about 200 friends in Facebook and about 20/25 relationships in LinkedIn are more likely than others.

By analyzing the joint distributions only, it is difficult to compare the relations between a social site against the other ones. To this aim, we compute the average degree of a node in a social site s ₁ conditioned on the degree in the site s ₂. The resulting computation for LinkedIn conditioned on Facebook has been shown in Fig. 10. In this case we observe an initial increase of the average degree in LinkedIn as a function of the Facebook degree up to about 500 friends. Then we observe a more unstable trend also due to a low number of data points. In the Fig. 11 we report a more pronounced lack of correlation involving the out-degree in Twitter and YouTube. Here we cannot establish an increasing relation between the out-degrees, so people how follow many users in Twitter would not follow the same amount of channels in YouTube. The specificity of the two social platforms, i.e. news broadcasting and video sharing, could be the cause of the weak correlation.

Keeping Facebook as conditioning social site, we observe that the average degree increases in some media like LinkedIn and Twitter, while the in/out degrees in YouTube are uncorrelated with the Facebook degree; as shown in Fig. 12. The comparison between the degree and the in/out degree is dictated by the inability to extract mutual links (more similar to a friendship link in Facebook or LinkedIn) since we can only retrieve the counting of the followers/followees in Twitter and YouTube.

Finally, to get an overall picture of the pairwise degree correlations, we apply a rank correlation analysis on the different pairwise sequences. Rank correlation analysis allows us to test if the ranking induced by the different degrees is similar or not. As a rank correlation method, we compute the Kendall’s rank correlation coefficient τ _b ⁷ on the ranking induced by the degrees. In Fig. 13 we visualize the rank correlation matrix, where each row (column) corresponds to a different social site. A strong positive correlation does not exist, rather the scenario is multifaceted. In most pairs, there is only a limited positive correlation (0.1−0.23) between degree centralities. This means that users may have a very different centrality across the services, i.e. a single user might be an hub on one system and loose part of its hubbiness on the other. One reason may rely on the different goals of the services; whereas LinkedIn is business-oriented or Twitter is an interest network, Facebook incorporates all the previous features. So, for instance, LinkedIn may only capture a part of the Facebook friends.

We compute the friend deviation between Facebook and Linkedin(\(\Delta k^{FL}_{u}\)), Facebook and Twitter(\(\Delta k^{FT}_{u}\)), Facebook and YouTube(\( \Delta k^{FY}_{u}\)), Twitter and LinkedIn(\(\Delta k^{TL}_{u}\)) and Twitter and YouTube(\(\Delta k^{TY}_{u}\)) for the users who have joined them. We report the trends of the friend deviations in Fig. 14 sorted in decreasing order. We observe that \(\Delta k^{FT}_{u}\), \(\Delta k^{FL}_{u}\), \(\Delta k^{FY}_{u}\), \(\Delta k^{TL}_{u}\) and \(\Delta k^{TY}_{u}\) are positive for 5360 users (out of 8527), 3313 users (out of 4361), 2944 users (out of 3177), 2518 users (out of 4148) and 2760 users (out of 3098), respectively. These results indicate that users in our dataset prefer to create friendships in Facebook rather than in LinkedIn, Twitter and YouTube. Moreover, users prefer Twitter rather than LinkedIn and YouTube to establish friendships. One remarkable result is that users preferring Twitter rather than Facebook have significantly more friends than those they have in Facebook.

In general maintaining the importance across social media is not a straightforward task and asks for a deeper understanding; for example it is not clear how user’s neighborhoods in different media overlap.

Finally we apply the above methodology i) to investigate how often users publish contents in different social sites, and ii) to asses if a form of correlation exists between the amount of posts and the number of friends. For point i), we are verifying whether users who post a lot and often on a social platform, are equally active in other platforms (see Q3). We measure the activity level of a user within a social media by means of the posting rate, measured in number of posts per week. By considering the posting rate rather than the post count, we mitigate the effects given by the adoption of social media in different periods. We report results about the analysis of the rank correlation matrix applied to pairwise posting rate sequences. In Fig. 15 we visualize the Kendall’s coefficient τ _b for the most used pairs of sites. Unlike the above discussion on centrality, there is a more evident positive correlation between the posting activity across social sites. The obtained values do not mean that users are equally active on both social media, however there is a positive tendency to be active in different social sites. In general, the maintenance of the posting activity across social media is not a straightforward task, like in the degree analysis.

By the second point we wonder if users with many friends in a OSN are more active and productive than people with fewer friends. To this aim, we consider only users whose degree and posting rate are available and combine these information. The analysis of the Kendall’s coefficient τ _b highlights a medium positive correlation between the two variables for all the online social networks (τ _b ∈[0.27,0.4]) except YouTube (τ _b =0.1). High degree people tend to post and publish more than those with few friends.

A case study on 4 social media

Up to now the analysis has focused on the pairs of social media. In this section we present a particular case study that involves 524 users who have joined Google+, Pinterest, LinkedIn and Twitter. We select this subset of social media because they have the highest number of users w.r.t. other subsets of 4 elements and their users are also very active. For instance in Twitter users published 1025 elements on average, followed by LinkedIn (139.32), Google+ (137.55) and Pinterest (134.71). In fact, we observe that less than 40 % of users in Google+, Pinterest and LinkedIn have more than 100 posts. While in Twitter almost 87 percent of the users produced more than 100 posts.

In the light of the results about the moderate correlation among the posting rates, we wonder if, in this group, people who actively post in a service necessarily produce many posts in the other services. To this aim we denote the top 5 % of users in each media as most active users. Only two users are the most active in all the services. 36 %, 30 % and 42 % of the most active users in Twitter are also the most active in Google+, Pinterest and LinkedIn respectively. 23 % and 14 % of the most active users in Google+ are also the most active in Pinterest and LinkedIn respectively. The above observations support the presence of users who are very active pairwise but, whereas the number of sites actively used increases, the number of posts and consequently the productivity reduces as observed in the previous section.

We quantify the diversity of the posting activity across social media by defining the posting deviation \(\Delta p^{s1s2}_{u}\) of a user u as:

$$ \Delta p^{s1s2}_{u}=\frac{|np^{s1}_{u}-np^{s1}_{u}|}{max(np^{s1}_{u},np^{s1}_{u})} $$

(2)

where \(np^{s1}_{u}(np^{s2}_{u})\) denotes the number of posts of user u on the social media s1 (s2). \(\Delta p^{s1s2}_{u}\) ranges from 0 to 1: if Δ p decreases towards 0, u tends to publish the same number of posts both in s1 and s2. As shown in Fig. 16, where we report the posting deviation Δ p between Google+ and the other social networks, this quantity decreases linearly. From the figure emerge that users’ behavior is more variable between Google+ and Twitter than the other social networks. Only 9 % of users show a post deviation less than 0.5. Most of the users prefer to post on a single service. So users who laboriously publish posts in one service do not publish with the same rate in the other ones.

To face the new challenge of giving an all-around picture of people’s online behavior, in this paper we perform an analysis on the same users across multiple social media. Our study relies on a new rich dataset gathered from the social media aggregator Alternion. It collects information about how users post their favorite contents, their centrality on different social media and the usernames they choose. The study of this dataset let emerge two main insights. On the one hand we confirmed that the on-going phenomenon of the adoption of multiple social site is spreading. Not just people sign up in many social media but they are active and exploit their services as well. On the other, the temporal information about how and when users post allowed us to investigate how people manage the opportunity of having different communication channels at their disposal. As far as we know this study represents the first attempt to deal with the people’s activities on multiple social media by using a large set of social platforms and users.

As regards the multiple adoption of social sites, the analysis of social media usage shows that Alternion data capture the typical trend in today’s users, despite the limitations discussed on the dataset section. The usage of multiple platforms is gaining momentum. In fact, half of the users signed up in at least three different social sites. This result stresses the importance of a multiplex approach when conducting studies which rely on online social networks. In fact people are expressing their identity and their behaviors through multiple communication media. This observation is further strengthened by the results about active users. The fact that 73 % of active users publish on at least two social sites means that people choose the right online channel to communicate and convey their contents. Also in this case the multiplex approach is fundamental in the extraction of people’s interests and preferences. To this aim we plan to exploit these first results to study how users build their social identities across their social platforms Kooti et al. (2012). In particular we will verify whether different social norms characterize the major social platforms and if users adapt their self-presentation to these norms, as the results about the choice of the username may suggest. Finally we will ask whether the identity of a single user expressed through the contents s/he publishes is persistent and coherent with the profile information.

The multiplex nature of the Alternion dataset allowed us to investigate the maintenance of users’ popularity and centrality across social sites. In fact, despite the plethora of communication media, individuals are not confident with each media in the same way. For instance, someone better expresses and communicates through video, others through texts or blogs. That may result into different levels of engagement with fans or friends and, consequently, in different degree centralities. To this we asked whether a person’s popularity is uniform across the social sites. The analysis led to not straightforward results and indicated that user’s popularity in a given social site barely corresponds or does not correspond at all to his/her popularity on another social platform. Nverethenless, a more evident positive correlation between the posting activities across social platforms exists. In the future we plan to analyze the reasons behind the weak correlations we observe. In particular we will verify whether the social norms of the online platforms and the services they provide impact on the centrality of the users.

The multidimensional and longitudinal nature of the dataset is fundamental in the understanding of the human dynamics in communication and online social networks since it offers the opportunity to study how people handle different media. The primary goal of our analysis was to characterize the time-series associated to the posting activity of the single users across multiple social platforms and unveil their statistical properties focusing on measures which describe their level of burstiness. The analysis of the burstiness moved in two directions: i) we classified the aggregated behavior of each single user; and ii) within each user we characterized the bursty behavior in each single social media. The results show that i) singularly each event sequence keeps an high level of burstiness, i.e. the bursty behavior of the aggregated interaction sequence is the union of bursty event sequences and; ii) a period of high activity in the aggregated sequence does not imply that the user is highly productive in each social media during the same period. These observations represent the basis for a model of temporal patterns in multilayered social network which combines the idea of attention allocation and novel models which reproduce the human bursty dynamics Perra et al. (2012). We also plan to investigate the interaction sequences through frequent pattern analysis in order to highlight whether users are characterized by specific usage subsequences, i.e. they have a predefined scheduling in the usage of their preferred social media. The above analysis will be combined to the content analysis of the posts to see if personal topics change along the observation period or users change their posting behavior in terms of covered subjects or whether some social platforms specialize on specific subjects.

¹ http://www.pewinternet.org/2015/01/09/social-media-update-2014/

² http://www.alternion.com

³ http://www.alternion.com

⁴ http://census.gov

⁵ The dataset is available by e-mail.

⁶ Alternion account is included by default, so counting starts from 1

⁷ τ _b takes into account ties.

⁸ δ is the Heaviside function