1. Recommendation system
Hook
If you use Amazon or other famous services,
you might be controlled by this technology…
Is it really your free will or not?
Key details
More details
・The recommendation system is one of AI technologies. This
technology analyzes your history, then recommend what you want
from a huge amount of items. This technology is used in so many
fields such as marketing, e-commerce and so on.[1] There is a lot of
successful experiences.[2]
・There are two types of the recommendation systems. The first type
is content-based filtering(CBF) which is based on the compatibility.
The second one is collaborative filtering(CF) which learns user’s
preferences from their history. In general, the accuracy of CF is higher
than CBF.
・There are several types of the recommendation systems and they
have their own strong and weak points[3]. The most suitable type for
you all depends on what you want to do in your service.[4]
[1] For example, Amazon and ZOZO in e-commerce field, YouTube and Netflix in video and movie
service, salesforce and AWS in web access and marketing field and so on.
[2] 35% of the Amazon’s sales is generated by the recommendation system. In other words, if they
lose their recommendation system, 35% of their sales(135 billions of dollars) will decrease soon.
75% of the Netflix’s users watch today is generated by the original recommendation system
called Netflix Prize.
In those examples, we can realize high user experiences and sales up by using this technology. Big and
famous B to C services in the world such as Amazon or Netflix definitely use this technology and the
market of this technology is growing up. According to some researches, the size of this market in
2028 will be five times bigger than now.
[3] CBF and CF has own strong and weak points. If we don’t have user’s big data, we can’t use CF
because this type must learn user’s history and then, they generate suggestions. In this case, we can
use CBF because this type doesn’t need user’s big data. The only thing we must do by using CBF is let
user’s register what they like at first. It means that if we just start our services, we can’t use CF but
CBF. As we gather big data from user’s in our services and then, we can finally use CF.
[4] As I mentioned before, CBF and CF have own characteristics. Thus, if we implement this
technology, we must think of which type is suitable for our services. Maybe we need the hybrid of
two types or we don’t need the recommendation system or we need other types. The cost of this
technology is expensive to implement. If we mistake the type of this technology, we can’t improve
user experiences and users could be bored by our services. Just take deliberate action. Just think
different.