Second Version of SoSeVi (2016)

The second version of the Social Set Visualizer has been presented in Publica- tion III [Flesch et al.2016] of this dissertation. Moving towards the second version, the split of Before, During, and After periods is made more apparent in the user interface. Furthermore, the design settles on an UpSet-inspired visualization of sets, which is very suitable for the visualization of large-scale set intersections. The UpSet- inspired set visualization in the second version of the Social Set Visualizer is detailed in section 3.4.6. Application performance is significantly improved by optimizing the database layout and storage backend, as well as the utilization of advanced caching techniques based on a Redis in-memory database.

Figure 4.2 shows the second version of the Social Set Visualizer as an inter- active tool for large-scale set intersection calculations. This version of the Social Set Visualizer depicts the first visual analytics tool to visualize migration flows in Big Social Data through means of clever set intersections along the two Social Set Analysis dimensions of time and space. Version 2 of the Social Set Visualizer project

62 Chapter 4. Development as presented in this figure provides a significantly better interface for Social Set Analysis tasks of Big Social Data originating from Facebook. It depicts the first Visual Analytics tool to visualize migration flows between set intersections in Big Social Data. Furthermore, it allows an interactive selection of a relevant time period for analysis and calculation of intersections for Before, During and After periods over a big data-style dataset fetched from Facebook. Set intersections are dynamically encoded through a combination matrix. The migration of actors between time pe- riods and set intersections is showcased through cardinality changes on the right hand side. User-provided event markers signify important real-world events with relevance for the underlying research domain of the investigating analyst.

The DashboardView (Figure 4.2) depicts the main view of the web application. It contains the main visualizations and is initially shown to the user. It consists of an overall activity visualization [F]. The researcher moves the time period selection tool [S] to navigate the data, and toggle data sets from different Facebook walls depending on the analysis tasks at hand. Based on the user-selected time period, which is labeled as the During period, Before and After time periods can be deducted

B D A

S

F In C M

Figure 4.2: Version 2 of the Social Set Visualizer (SoSeVi) dashboard as of May 2016, showcasing 8M Facebook interactions from the Volkswagen pages adapted to the Social Set Analysis approach: [F] main activity chart zoomed in on the user-selected time period using the [S] selection tool. Underneath alphabetical word clouds for the time periods [B] before, [D] during, and [A] after are shown. On the right side, all set [In] intersections encoded as combination matrix, where [C] displays the cardinality of each intersection. On the very right, a visualization of period-over-period actor [M] migration between set intersections is displayed.

4.4. Iterations on the IT Artifact 63 by looking at the beginning (earliest event) and the end (most recent event) of the underlying data. Three alphabetical word clouds underneath the main activity chart [F] illustrate the most important conversation topics in the [B]efore, [D]uring and [A]fter periods. This showcases the pluggable architecture of the Social Set Visualizer, which facilitates set-based content analysis tasks on the underlying dataset through use of the user-selected time frame.

To the right of the main activity visualization, the set intersection visualizations [In] are dynamically generated based on the user selection of the time period. Set intersections are encoded in a combination matrix. Each data source uses a distinct color. For each set intersection, up to three bar graphs are displayed in [C]. In case the set has a cardinality greater than zero, bar graphs for the Before, During and

After periods are created. The bars are horizontally stacked, with the topmost bar

signifying the Before period, the center bar During, and the lowest After. Right next to [C], a visualization of actor migration between periods and set intersections is displayed [M]. The Social Set Visualizer calculates all possible set intersections for each set with all sets of the following period based on the user-selected time period from the selection tool [S]. Therefore, the set intersections of all Before and During periods, but also of all During and After periods are calculated and shown. Migration flows can be individually analyzed by the user through interactive controls.

More information on [C] and [M] is detailed in Figure 4.3. Colored lines originat- ing from the bar graphs in [C] are displayed in case the Social Set Visualizer has found set intersections with subsequent time periods. Thereby, the scale in [M] gives information on the cardinality of the migration between sets, and the curved line points to the destination set. Figure 4.4 illustrates how the researcher can zoom in on a single migration path and gather detailed information about sources and des- tinations of migrations. The cardinality numbers in [M] depict the actual migration volume, whereas the pointed arrows indicate the destination set of the migration.

64 Chapter 4. Development

Figure 4.4: Visualization of actor migration originating from the Calvin Klein Face- book wall Before time period, showcasing strength and destinations of migration to set intersections in the During time period

This functionality implements the detail on demand principle which was introduced earlier.

RawdataView presents a detailed search interface for the underlying Facebook

activity data. It is accessible to the user through various means by interacting with the visualizations of the DashboardView. It is a handy set of tools for analysis of actor mobility and cross-postings between different time periods and Facebook walls. ActorsView presents a dedicated interface for analysis tasks related to Actor Mobility across time and space of companies’ facebook walls. The visualizations of actor mobility in DashboardView refer to ActorsView in order to provide the user with further details when requested.

In line with Action Design Research methodology, the main learnings from cre- ation of version 2 of the Social Set Visualizer consist of the novel visualization of social media migration patterns and utilization of selection-matrix inspired set vi- sualizations, an imitation of the Upset approach. Database and APIs have matured, but during development of the migration visualization and the optional on-mouseover detail-on-demand features, it became apparent that a domain-specific query lan- guage which allows for on-the-fly set definitions is very much needed. Hence, Social Set Query Language was introduced in the next version of the Social Set Visualizer for streamlining both development of the set-based visualizations and actual use in reproducible Social Set Analysis case studies.