Recommendation System Based On Campus Card Data

With the advent of the era of big data,information overload problems have become increasingly prominent.How to find useful information from a large number of unstructured data by ourselves has become an urgent problem to solve,in this case,the recommendation system promised for us.This paper mainly introduces the collaborative filtering algorithm and graph-based algorithm on topN recommendation.We improved the algorithms based on our data set.This paper presents two new evaluation indicators that describe the position of goods in a recommendation list.For the collaborative filtering algorithm,we proposed an algorithm to reduce the complexity of computing similarities.For the graph model algorithm,we first excavated the potential friend relationship through the campus card details,and added this relation into the classical bipartite graph model.Further,a graph model of the relationship between goods and commodities was added according to whether or not the goods belong to the same canteen.For the graph model,we use the PageRank and PersonalRank algorithms,the Markov chain-based algorithm and the generalized inverse algorithms.We added principal component analysis to the graph model.Moreover,we carried out the related mathematical derivation.Especially we proved the properties of the distance of nodes in the graph by Markov theory which is a new method and is also applicable to the case where the adjacency matrix is asymmetric.At last,We combined the graph model and the collaborative filtering algorithms to give the recommendation.