Organizing a Virtual Networking Event — a Data-Scientists Approach

Photo by Taylor Vick on Unsplash

tl;dr: I used wonder.me as a collaborative platform and built a graph from hashtags the participants made during the session using pyvis. I then used this as a background image to let those participants create small groups on topics they like.

The Networking Challenge

I am currently studying Data Science & Business Analytics at the Digital Business University of Applied Science in Berlin. In this course, we follow a hybrid approach where the content is presented online, and we meet each other about once a month on the campus. In addition, we use messengers and collaborative tools like Teams to stay in touch between those on-site events. This, of course, came to a halt because of COVID-19. While the university is very well prepared for that, and all the study modules are working completely fine in a 100% remote situation, I needed more social interaction.

I decided to organize a virtual networking event and bring students together. However, by planning this, I faced two core challenges:

• Students don’t know each other very well, because they meet only occasionally. They also don’t know which other students may be interested in the same topics.
• Suppose you try to use regular collaboration software like Zoom or Teams. In that case, it’s hard to split, let’s say 20 people, into smaller chunks organically. Hence, everybody has the chance to contribute to discussions on spontaneous topics.

I had to find a way to let participants find their peers and build up small casual groups to discuss topics they love and get connected through it.

Building Organic Meetings

To overcome the challenge of too static discussions, I choose wonder.me as a platform. The significant advantage is the user interface in which you move over a 2D plane. Suppose you enter a specific area or move near enough to another participant. In that case, you enter a video call either with the particular person or with all the participants in that area. So you can decide for yourself with whom you come into a ring. It’s also for free — that’s probably also a big plus.

Figure 1: Using an Area in wonder.me will bring all the participants in a single call.

Building a Hashtag Graph

Networking is all about finding common themes and topics and getting in touch with people with similar expertise and interests. There are several ways to push the finding of interest groups. One way could be to use Icebreaker questions or provide common themes and force everyone to say something about them. We all know that this sometimes ends in awkward silence. The host then has to get the conversation going with another team-building concept.

My idea was to let the group decide themself what to talk about and help them find a connection between the topics they want to talk about and discuss. I tried to use a graph built on hashtags as nodes and create edges to relate them. I then wanted to use this graph as a background image for wonder.me so participants can move to the nodes they are interested in and start their conversations. Suppose this is not the case for some participants because no one else stands near their node. In that case, they can follow the edges to the next populated node and join a conversation there.

I used a list of “starter hashtags” to start things off and ensure that the topics also fit together. Everyone should use at least one of them and connect them to a topic they are interested in. The study programs inspired this list. In our case, we had students from data science, digital business management, and a cyber security program. So our list for this event could be something like:

• #datascience
• #businessanalytics
• #infosec
• #socialmedia
• …

The participant’s task was to build small graphs with the help of a simple syntax (Figure 2) and write those simple lines to the chat.

Figure 2: Simple syntax for building a graph structure.

One must write them in a line to connect two hashtags, divided by a semicolon like #datascience;#socialmedia. While the participants reported their hashtags in the chat, I copy-pasted them into a CSV file (called hashes.csv in this example). To visualize the graph, I used pyvis and built a small script (Snipped 1), which outputs an HTML page, where I further can facilitate the result like in Figure 3.

Snipped 1: This snippet turns hashtag combinations into a graph representation.

With the help of pyvis, it’s possible to reorganize the nodes using different settings for their physical behavior. While the distance between the nodes and the aspect ratio should be kept in mind, it’s also vital not to overengineer here because the event is still happening. It’s more critical getting it going than optimize the graph layout.

Figure 3: pyvis provides an interface to manipulate the graph.

After the graph fits nicely as a background image, the guests were asked to go to the location on the wonder.me plane which they find the most interesting. I then stopped the starting area, and the participants organically found themselves together in smaller groups to start their conversations.

Figure 4: Using the graph as a background image helps participants find peers with the same interests.

Learnings and Thoughts

There are some key takeaways from the event. I want to share some of them and invite you as a reader to give me some tips on making the networking event even better in the future. For example, I thought that adding weights to the edges would become thicker when more people used them to build their graphs. This probably would animate participants to focus more on existing nodes and edges instead of introducing new ones.

Use starter hashtags. This will nearly guarantee that everybody’s interests are connected to the leading network.

Don’t use too many hashtags. With a growing number of participants, the chances are that the number of hashtags and the complexity of the net will grow. If that happens, ask your guests to look at which connections were posted in the chat already. Then, you could animate them to build relationships between those existing nodes instead of introducing new ones.

Providing a visual to help understand the syntax of the graph. For example, using words to explain syntax to build up nodes and edges is okay. Still, with a visual like Figure 2, it’s much better to understand.

Keep an eye on the dynamics. People sometimes want to join groups that are already full or standing on a topic no one else wants to talk about. So you, as a host, should take action. Resize the groups, ask others to go around, and invite others to discuss different topics. Then, of course, be active and talk to people if they can’t find a group.

Use those hashtags for social media. If you organized this event out of your social network, you know have a pretty good list of topics your audience is interested in. You can use it to optimize your content or get in touch with people having similar interests.

I hope this blog post inspires you to make a digital networking event more organic. I will love to read your experience in the comments if you try it out.