MIT
Technology Review: Online shopping sites make use of
various algorithms to suggest items for you to purchase based
on what you and other users have purchased in the past. One
effect of such recommendations is that some locales or products
suffer from the sudden influx of people directed to them by the
recommendation. To try to avoid this problem, a team of
researchers led by Stanislao Gualdi of the University of
Fribourg in Switzerland has applied a feature of particle
physics. At the atomic level, particles tend to occupy the most
energetically favorable states; but the number of particles
that can occupy any given state depends on the type of
particle. Gualdi drew a parallel between this concept and that
of commercial products, which can be shared by either many or
just a few people. The team developed a model that can limit
the number of users allowed for a given product. When testing
their model against empirical data of DVD rentals, they found
that limiting rentals ensures that a wider range of DVDs get
rented. And the more DVDs rented, the broader the range and
accuracy of the ensuing recommendations. The overall effect was
a healthier rental system. However, whether the model would
work for actual retailers, who focus on maximizing their
profits, is uncertain.
