The tenth chapter on-line consumption decision
This is mainly about consumer psychology, more close to psychology, rather than technical aspects.
1. The traditional decision-making model is that people's interests are consistent and will not change, but modern research shows that users ' preferences are not stable in the decision-making process 2. People will make decisions based on the display environment of the goods: the same goods, placed under different circumstances, the price is different 3. First/Recent effect: items at the top and end of the list are easier to remember 4. Framework Effect: Presentation Framework determines user decision 5. Start effect 6. 。。。
Personal evaluation, if you want to learn more about this piece of content, it is better to find a book of consumer psychology to see
==================================================
11th Chapter recommendation System and Next generation Internet
Prior to the recommendation method, the premise is that it is on a website, the user on the site of the items have browsing behavior, purchase behavior, and so on, the goal is to enhance the purchase of goods.
With the development of the Internet, more resources are available on the Internet. Such as: The advent of wikis, give a lot of ontology (words), with these words as a dimension, can not only describe the text content (such as news content), but also can portray user behavior (such as: User browsing, collection and other behaviors), then based on the ontology can be the news to the relevant users.
For example, many social networking sites, users have their own tags, how these tags reflect the interests of users? How can I use it in a referral system? How to make the label information more accurate to the user's recommendation when the user has an existing item consumption behavior (such as rating a movie) and has its own label. --1. When collaborating, you can use tags to calculate user similarity; 2. The discovery of the link between the label and the user's consumption behavior? Able to solve the non-consumption behavior of cold start users; 3. Consider the recommendation problem as a classification problem, and the label is one of the input features.
==================================================
12th. Recommendations in pervasive environments
The main problem is how to apply the recommendation system to the mobile Internet, for example: LBS. There is no substance to this one.
==================================================
13th Chapter Summary and Prospect
Nothing more than substance.
Finish.
"Book Notes" (Recommender systems An Introduction), chapter tenth to 13th chapters of the recommendation system
