Figure 2 5.: Global Internet penetration and censorship levels.
Related Figures (102)
1.4.5. Actor space and document space To better differentiate between methods of impact analysis, the thesis defines However, since the rise of online social network sites (SNSs) in the last decade, detailed data of social networks and items which are shared via SNSs became available. Analysing and monitoring this data is of interest for individuals and companies alike, which use SNSs to market themselves or their brand. The necessity of monitoring this feedback led to a new field of research and a multitude of tools used by the industry, which fall under the umbrella term social media monitoring. Six months after this blog post YouTube switched to their current like /dislike privacy. Features discussed by the authors (news feed, privacy settings) differ Figure 2.11.: Unary views feedback on various SNSs purpose and can be omitted. Figure 2.15.: xkcd #1098 - Star Ratings discussed in the thesis: textual feedback. Figure 2.16.: Creating textual feedback on various SNSs of textual feedback an item has gathered a generic comment icon is used on Figure 2.17.: Steiner’s cartoon as published in The New Yorker in 1993. opportunity for important interpersonal interactions. Figure 2.18.: Moods used by the SNS deviantART. related to hosting audio files it might be important which parts of an audio Figure 2.20.: Heatmap of clicks on Google’s search results page where users click the most, blue areas indicate less activity. Since many SNSs use algorithms based on explicit feedback to determine in- teresting content, many users - and also companies - game the system to gain more exposure. Users are invited to express their opinion on a topic by in- teracting with the post via the site’s feedback mechanisms in an unintended manner. Figure shows two examples on Facebook, where users are asked to (a) like or share an item to vote for a brunette or blonde girl, (b) like or comment to show agreement or disagreement with an article. The line graph emphasizes the importance of single nodes that are able to block information flow. It shows that even though some actors may not have many ties to others, they still can be in a position to control information of a part o: the network completely. Such actors are called gatekeepers |Ere80) and, depen- dent on the part of the network they “control”, can achieve a high centrality using certain measures (e.g. actor betweenness centrality, see Section |3.2.3). details of the computation are not discussed here. Figure 3.3.: Directed star graph .4.2. Comparison of actor prestige indices LiveJournal, one of the most popular SNSs in Russia |Gor06], was created in 4.1.8. LiveJournal Various open-source SNSs, their implemented feedback types as well as their Evidently, the implemented feedback types vary from site to site. The one feedback type that is implemented on every investigated social network site is the comment, which is usually used to reply to existing posts. Other widely used feedback types are like (and the related favorite), which are mostly used to indicate that content was liked, as well as reshare, which is used to re-post content, thus expanding the reach of a published item. 4.1.10. Summary distribution of Likes by gender, countries, cities and languages of a Facebook Figure 4.13.: Graph showing a Page’s Reach. Each point represents the unique people reached in the 7-day period ending with that day. Along with Reach, Page Insights shows the frequency of views. The frequency duplicates by using either the account or the IP address of the user. Talking About This is the third category of Facebook’s Page Insights. It combines two metrics: “Talking About This’ and ‘Viral Reach’. Talking About This is defined as “the number of unique people who created a story about your Page”. Viral Reach is defined as “the number of unique people who saw a story published by a friend about your Page”. Figure [4.15]shows both metrics in graph form. in graph form. Figure 4.19.: Statistics about a Google+ post and reshares per hour. In contrast to Facebook’s impact analysis capabilites, the features of Google’s SNS are reshare centric. Detailed information about the origin of comments or +1s is not provided as of now - the focus lies on reshares, and the visualization of reshare sequences. For further details regarding the visualization techniques used by Google+ Ripples refer to |VWH*13]. Figure 4.24.: Detailed information about views Figure 4.25.: Absolute audience retention of a YouTube video from such features. that provide insights to its users. Figure 4.28.: Twitter Analytics: timeline activity dashboard Filter: Showing all 613 followers as of 6/12/2013 (days shown in Pacific time) Figure 4.29.: Twitter Analytics: graphical representations of followers over time over time. Figure 4.32.: Visualizations of various LinkedIn profiles visualizations of various LinkedIn profiles. InMaps is LinkedIn’s social network visualization tool. Access to the tool is restrictive, since your profile needs to have at least 50 connections and be 75% complete (in terms of profile information) to gain access. Figure shows visualizations of various LinkedIn profiles. friendship - is needed. Launched on August 14, 2009, the site is ranked 7“ in China (29 globally) according to the Alexa page rank. The service was launched in a reaction to the censoring of many popular non-chinese microblogging and social network sites like Twitter and Facebook [Wik13r]. Sina Weibo builds its social graph like Twitter does, using “following” relations. A user can follow any other user without his or her permission. No reciprocal action - such as confirmation of friendship - is needed. Figure 4.35.: Sina Weibo’s “Data”: Information about one’s fans Data shows detailed information about one’s fans: geographical distribution, Figure 4.38.: Pinterest Web Analytics: Impressions and Reach Figure 4.37.: Various metrics available in Pinterest Web Analytics Figure 4.40.: Feedback options are shown at the bottom of an image. Figure 4.44.: Academia Analytics: Traffic sources and top keywords shows geographical information regarding the origin of views. to total document views over a 30 day period. Figure 4.43.: Academia Analytics: Profile views and document views Eleven closed-source SNSs, their implemented feedback types as well as their Figure 4.45.: Academia Analytics: Viewer world map Figure 4.46.: Academia Analytics: Most viewed documents 4.2.12. Summary Figure 5.1.: Interdependencies of the dacodi components tions explain the components briefly. Feedback collection in dacodi is done in two steps: Table 5.2.: API rate limits of the platforms supported by dacodi Feedback objects belong to Remote objects, thus every Feedback object must Feedback objects belong to Remote objects, thus every Feedback object must have an associated Remote. As a result, Feedback objects have a slim im- plementation. Any additional information that might be needed, such as channel or authentication information, can be retrieved via the associated Re- mote. The object model of dacodi enables a straight-forward aggregation of feedback on the cascade that is the relation of Users > Publications > CommonWeaverModels > Remotes > Feedback. This object relationship effectively forms a tree for every user, with the user as the root and Feedback objects being leafs (Figure [5.5). Figure 5.6.: Feedback aggregation in the dacodi object model called, which in turn calls the feedbacks method of the CommonWeaverModel Table 6.1.: Data selection criteria applied to social network datasets in the Stanford Large Network Dataset Collection (SNAP) 52 Actor identity has to be given, i.e. actors must not be anonymized Table 6.2.: Absolute and relative number of nodes and edges of the datasets As apparent in both the frequency distribution, as well as the visualization, most actors have a low potential impact according to the actor degree centrality Figure 6.3.: Visualization of the degree centrality measure (ego-Twitter- partial). Actors are colored according to their degree centrality, with blue indicating low centrality, followed by green, yellow, or- ange and finally red. indicating highest centrality. Figure 6.4.: Frequency distribution of potential impact C, Figure 6.5.: Visualization of the closeness centrality measure (ego-Twitter- partial). Actors are colored according to their degree centrality, with blue indicating low centrality, followed by green, yellow, or- ange and finally red, indicating highest centrality. Figure 6.7.: Visualization of the eigenvector centrality measure (ego- Twitter- partial). Actors are colored according to their degree centrality, with blue indicating low centrality, followed by green, yellow, or- ange and finally red, indicating highest centrality. The following subsection examines the correlation of the various measures to determine potential impact and the actual impact as retrieved by dacodi. Figure 6.10.: Correlation of potential impact C’, and actual impact Figure 6.12.: Correlation of potential impact C}, and actual impact Figure 6.11.: Correlation of potential impact CG and actual impact Table 6.3.: Correlation coefficients of potential impact and actual impact Global actor degree centrality Figure A.1.: Social network sites world map, June 2012 Figure A.2.: Social network sites world map, December 2011 Figure A.3.: Social network sites world map, June 2011 Figure A.4.: Social network sites world map, December 2010 Figure A.5.: Social network sites world map, June 2010 Figure A.6.: Social network sites world map, December 2009 Figure A.7.: Social network sites world map, June 2009 Again, to normalize the results the sum of all actor information centrality indices of all nodes is required. Ne can now calculate the actor information centrality indices for all nodes Listing D.1: GET /feedbacks Figure E.3.: Frequency distribution of potential impact Ss in ego- Twitter- an ananaid
![LiveJournal, one of the most popular SNSs in Russia |Gor06], was created in 4.1.8. LiveJournal](https://figures.academia-assets.com/36188217/figure_021.jpg)
![Talking About This is the third category of Facebook’s Page Insights. It combines two metrics: “Talking About This’ and ‘Viral Reach’. Talking About This is defined as “the number of unique people who created a story about your Page”. Viral Reach is defined as “the number of unique people who saw a story published by a friend about your Page”. Figure [4.15]shows both metrics in graph form. in graph form.](https://figures.academia-assets.com/36188217/figure_025.jpg)
![Figure 4.19.: Statistics about a Google+ post and reshares per hour. In contrast to Facebook’s impact analysis capabilites, the features of Google’s SNS are reshare centric. Detailed information about the origin of comments or +1s is not provided as of now - the focus lies on reshares, and the visualization of reshare sequences. For further details regarding the visualization techniques used by Google+ Ripples refer to |VWH*13].](https://figures.academia-assets.com/36188217/figure_029.jpg)
![friendship - is needed. Launched on August 14, 2009, the site is ranked 7“ in China (29 globally) according to the Alexa page rank. The service was launched in a reaction to the censoring of many popular non-chinese microblogging and social network sites like Twitter and Facebook [Wik13r]. Sina Weibo builds its social graph like Twitter does, using “following” relations. A user can follow any other user without his or her permission. No reciprocal action - such as confirmation of friendship - is needed.](https://figures.academia-assets.com/36188217/figure_037.jpg)
