Network applications in 2021: misinformation spread and pandemic modeling
In the contemporary world where people are strongly connected, the applications of graph analysis have been greatly expanded by Network scientists to solve more problems we are facing. The article above discusses several potential and accomplished applications shared by top network science researchers at the 2021 Graph Exploitation Symposium, hosted by MIT Lincoln Laboratory. These topics, including the spread of misinformation, pandemic modeling and drug design, became especially crucial today under the Covid-19 pandemic.
In the Social networks section of the article, the author posits the role of network science in analyzing influence operations (IO) that spread misinformation online, such as spreading false Covid-19 treatments to vulnerable populations. The way researchers develop tools to detect these IO accounts can be related to the course material, which addresses that local bridges are normally very rare in giant social networks, and if an account connects to a lot of other accounts that are completely irrelevant to each other (thus connects a lot of local bridges), it is very possible to be a spam account. Similarly, according to the article, the researchers trained an algorithm to label accounts based on the number of interactions with foreign news accounts, the number of links tweeted, and the number of languages used. Then, they used a statistical approach to score an account’s level of influence in spreading the narrative within that network. In that case, if an account interacted with users from many different nations in many different languages, it is highly likely to be an IO account.
Another topic of the article talks about how researchers used network science to design effective control strategies that models the virus’ spread. They developed a planning algorithm and applied it to 3 counties in Florida, Massachusetts, and California virtually. Taking into account the characteristics of a specific geographic center, such as the number of susceptible individuals and the number of infections there, the researchers instituted different strategies in those communities throughout the outbreak duration. As a result, the researchers were able to eradicate disease with targeted interventions specific to each area. This approach is highly related to several concepts discussed in INFO 2040 class. For example, the algorithm could calculate the Clustering Coefficient of the susceptible individuals and determine possible influences they can have on other people connected to them (e.g. if they have a high Clustering Coefficient, that means a lot of their friends are also connected, thus there will be a higher chance of wide virus spread). Also, the algorithm could refer to the Triadic Closure concept to find out individuals that are not connected to the susceptible ones but are still vulnerable to indirect transmissions due to common friends.
The concepts and potential applications of network science discussed in the article made the research area more practical and connected to the current world, and many nations are already starting to use network science and Graph Theory to track close contacts of infected individuals and control the spread of the virus. It is highly possible that network science will influence the world even more in the near future when more things become virtual and more information is available online.
Source: Lincoln Laboratory convenes top network scientists for Graph Exploitation Symposium