Why do bridging nodes matter? They connect ideas
Many real-world systems can be described as complex networks. Take the neurons in your brain, for example. They're connected by an intricate system of synapses, which send signals to one another and form a biological neural network.
A Quid network is structured as a set of nodes and connecting links. The links represent similarity in language, allowing the software to analyze text-based data such as news articles, company profiles, product reviews, and surveys.
A distinct property of networks is the formation of clusters, or dense groups of nodes in a network. Often during network analysis, people focus on nodes that are centrally located within their respective cluster, or among the network as a whole, overlooking both outlying nodes and those that may lie in between.
In this post, we move the spotlight of network analysis from central nodes to “bridging” nodes, which connect dense regions.
Analyzing news articles, reviews, and surveys
Let’s first look at bridging nodes within a network of news articles, where each node represents an article. Their links represent similarity in language. Colored clusters in a Quid network quickly provide a bird’s-eye view of a broad topic. A node bridging two clusters can indicate an article at the intersection of two narratives, and reveal the author and publication making that connection.
On the left is a network of thousands of news articles on global warming and climate change published in the U.S. in 2017, with the larger subtopic clusters labeled. Isolating the clusters focused on the effects on health and agriculture reveals a few articles linking the two subtopics.
While the articles in the dense regions focus on their respective subtopics tied to climate change and global warming, these articles make the connection about how changes in agriculture and food production caused by climate change are impacting human health around the globe.
Gaps in existing markets are where unmet needs reveal innovation opportunities. These gaps, also called whitespace, are easy to identify in networks. When analyzing a Quid network of companies in a particular market, the companies in a bridging position are combining products and technologies from distinct market sectors in innovative ways. Below is an example of how bridging nodes in the retail space reveal unique startups filling in the opportunity gaps.
On the left is a network of more than 1200 private companies tied to retail that received funding in the second half of 2016. Zooming in on the whitespace between the machine learning cluster and its neighbors, we see several machine learning startups leveraging sensors, Big Data, and cloud technologies mostly focused on marketing and advertising efforts. ChatKit (fka Shoppe AI) is developing a shopping assistant platform for retailers that allows their customers to converse with them to discover and purchase products. Sweden-based MOD.CAM builds sensor-based devices for computer vision processing that enable stores to determine the profile of anyone passing a certain spot in the store.
Another fascinating aspect of analyzing whitespace in a network of companies: it can hint at future innovation. In April 2014, we gave a presentation on the “The Future of Hardware” (password: babel), analyzing more than 1500 private companies developing advanced hardware. The analysis included a segment analyzing whitespace (30:55-32:43) between clusters focused on autonomous vehicles & UAVs, telematics, and marine technologies. We suggested companies such as Accipiter Radar, DeTect, and Robin Radar Systems developing radar technology tracking aircrafts and birds would lead to applying the same technologies to manage UAV fleets. Sure enough, the prediction came true.
In 2014, Robin Radar Systems’ offerings included ‘bird strike prevention in aviation’ and ‘bird monitoring in wind farms’. Two years later, their website reveals a third offering - ‘drone detection for security’.
At Quid, we embrace networks over spreadsheets and pie charts when analyzing text-based datasets. As opposed to simple word counts, networks offer context and hidden relationships in your data. Just like in your social network where a co-worker connects you to a chess club, bridging nodes will reveal the connecting concepts in your data.