When users interact with internet content it is important, specifically for recommendation systems, to get their feedback. There are mainly 3 ways that are commonly used to gather users feedback on the Internet.
In the our case the content scored by the users could be any web entity such as news articles, products, vendors,…
1- The 5 star system.
The 5 stars system is one of the most flexible ways to gather user feedback. in this system the users score the content by giving a given number of stars. 4 or 5 stars is generally considered as a good rating (positive feedback) while 1 or 2 stars are considered bad rating (negative feedback).
Amazon is known for using this system in their website as well as most of the e-commerce ecosystem.
There are other variations of the 5 star system, but they are essentially an application of the same principles:
- variations on the number of stars: 3, 4 stars or more that 5 stars. Some websites may go until 10 Stars.
- Sliders, or continuous rating system. This allow a full range of scores and could be useful if more precise ratings are needed.
The 5 star system is usually combined with users comments. Users have then the opportunity to give more details about why they liked or disliked the product or the content.
2- The Like/dislike system
In this approach only 2 values are allowed, Users click on like, or thumb up if they appreciate the content or on dislike/thumb down if they don’t appreciate the content. this approach is less precise but have the advantage of forcing the users make a decision on the quality of the content.
The big social media sites use the type of rating, because this binary system easily convey the idea of agreeing/disagreeing with the content without necessarily attaching a good/bad label.
3- The behaviour-based rating system
This is a family of implicit ratings systems based on what the users do with content rather on what they actually say about it. This system functions as a proxy for the previous ratings or a as a complement. The different measures that can be used are:
- Purchase or repeat purchase behaviour: The fact that the user actually purchased or purchased a second time the product can be considered as an endorsement and positive rating for the product.
- Views and time spent with the material: When dealing with content on the social media or news sites the time spent by the user in the content page or the number of time the user viewed the content can play the same role.
Generally any measure that track the actual consumption of the users is a good measure. The tricky part is that sometimes it is difficult to use measures like “time spend” and to control for the user’s actual time spending on the content page. The users, for example could reach the page and and be diverted by some off-line activity for example.