“This Meeting Should Have Been a Visualization!”

Exploring engagement patterns in multi-party meetings.

“Conversation is a social process where the words exchanged are only part of the story. What hides behind awkward silence may change the meaning of a discussion altogether.”

Due to the COVID-19 pandemic, the popularity of online meeting systems has become widespread, and recording meetings is as easy as clicking on a button. Technology now allows us to connect and collaborate easily. Engagement in multi-party meetups is a salient indicator of outcome. Poor attendee involvement can hinder progress and hurt team cohesion.

Over the last several months, I have been studying how people in a new team set up interact and what impacts their talk time. The below visualizations are generated based on a zoom transcript for a meeting of undergraduate students who met over zoom for the first time. The visualizations help us know who talked for what time, but we cannot draw insights about what they were speaking or any other intricate details.

Zoom transcript visualization shows that members of a newly formed team speak almost for the same amount of time.

Zoom transcript visualization shows that one member takes the lead, and the other members follow accordingly emphasizing that the talk time distribution is not uniform.

To fully understand the engagement in multi-party discussions, users must utilize tools that bring their attention to social cues that might be easy to miss, such as awkward silence or turn-taking behaviors. That is where Wang, Keck, and Vosough’s Discussion Flows come into the picture. Discussion Flows is a multi-level interactive visualization tool that could help analyze engagement in multi-party meetings. Research has shown that the level of engagement during meeting correlates with its outcomes. To enhance the methods for analyzing meeting engagements, we need tools that bring our attention to relevant social interaction. The Discussion Flow team lists the following main research findings:

• Participants are mainly interested in discussions related to their responsibilities. Hearing about an irrelevant project might be handy if they share similar work items.
• Context is important when interpreting decisions. Users are not happy with simply receiving a summary of the outcomes. Instead, they want to know how a decision was reached, such as who proposed a solution or how much deliberation happened.
• Participants with managing responsibilities emphasized the importance of understanding the participation of all group members across multiple meetings.

Visualization Design

Level 1: Top Level — The aim here is to give users the big picture and allow them to compare multiple meetings. Each meeting gets represented by a glyph (a vertical bar). Each glyph contains multiple color-coded stack bars. Each bar represents an item in the meeting agenda, and they are placed in the same order as they get discussed in the meeting. Each color segment in the stacked bar represents one speaker, and the height responds to the duration of that person’s involvement in the agenda. For example, the magnified glyph shows a meeting with five agenda items.

An overview of all meetings based on agendas using Discussion Flows

The Discussion Flows have two views that users can utilize to discover patterns based on different criteria:

1. Week View
2. Time View

Week View in which meetings are sorted into different columns representing days of the week

Level 2: Main Agenda — Once users find a specific meeting, they can click on a glyph to get level two of discussion flows divided into two techniques:

1. Flow Diagrams (to the right): They show a participant’s engagement agenda-wise. The vertical axis represents the duration of time spent per agenda, and the horizontal axis represents the order of agenda discussion. That is consistent with the logic of the glyph on the overview page. Another important insight gained from this visualization is to know if a participant was interrupted. A solid line means no or little interruption, a dashed line means some interruption, and a dotted line indicates a high level of interruptions.

An example of Discussion Flows Level 2 view visualizing the involvement of 6 participants into 10 agenda topics over time. In the overlaid rectangle, one can see the interruption patterns of non-native speakers after applying filters

2. Filter Controls (to the left): The left-most section of the visualization uses the parallel coordinates technique to add additional attributes for the participants, serving as filter controls. For example, gender and language in the above visualization.

Level 3: Sub-agenda — Most agenda items include more sub-level agendas. When necessary, users can zoom in and examine them at the next level.

Each topic is broken down into smaller sub-topics

Level 4: Words — Users can also navigate to the word level, which summarizes participant contributions during a customizable time interval instead of agendas. With a smaller interval, users can observe closeup interactions, such as overlapping agenda sub-items. Filter and highlight features work the same as previous levels.

When visualized at 5-minute intervals, we see the overall contribution from each speaker

This video shows how Discussion Flows work:

Also check out the interactive demo of Discussion Flows.

Conclusion

Discussion Flows can be used in real-life settings so that users with managing roles can focus on productivity and team-building aspects. I will surely be using Discussion Flows collectively to analyze the transcript for my use case and to look at the intricate details of the conversation. That would help me compare and contrast participant engagement in meetings based on my desired criteria.

Citation
Read the full paper here!

• Tao Wang, Mandy Keck, Zana Vosough. Discussion Flows: An Interactive Visualization for Analyzing Engagement in Multi-Party Meetings. Eurographics DL. https://diglib.eg.org/handle/10.2312/evs20211060