The friend recommendation algorithm is called link prediction on social networks.
Content-based matching
We can recommend users who have similar content attributes to them as friends. The commonly used content attributes are given below.
- User demographic attributes, including age, gender, occupation, graduation school, and work unit.
- The user's interests, including the user's favorite items and published comments, etc.
- The user's location information, including the user's address, IP address and postal code, etc.
Friends recommendation based on common interests
- Calculate acquaintance based on what users like
- Extract the user's interest tags based on the user's speech in the social network to calculate the user's interest similarity
Friends recommendation based on social network graph
The simplest friend recommendation algorithm is to recommend a friend's friend to the user.
Three friend recommendation algorithms based on social networks will be introduced below.
expand
The most familiar conclusion about social networks is the six-degree principle. The principle of six degrees is that any two people in the society can know each other through the path of no more than six people. If it is converted into a graph term, the diameter of the social network graph is 6. The principle of six degrees has been perfectly proved on a uniform random graph. Readers interested in this can refer to the book Random Graph. Many researches on social networks are based on the theory of random graphs, so in-depth research on social networks must master the knowledge of random graph
theory.
There are two most famous problems in social network research. The first is how to measure the importance of people, that is, the centrality of social network vertices. The second problem is how to measure the relationship between people in social networks, that is, link prediction.