The Research Of Literature Recommendation Based On Multiple Interests

With the arrival of the era of big data,how to quickly find the target information from a huge database is an urgent problem that researchers need to solve.Association rule mining algorithm is one of the effective methods to solve such academic research problems.The mining of association rules is mainly studied in three aspects: the dimension of data,the level of abstraction of data,and the type of processing variables in rules.In terms of rules,three aspects are studied: positive association rules,negative association rules and rare association rules.At present,there are two limitations in the mining of association rules:(1)The measurement methods such as support,confidence,and promotion are excessively dependent on expert knowledge or complicated adjustment process in terms of value.(2)It is often difficult to interpret rare association rules.Based on the existing researches,this paper proposes a Markov logic network framework model for association rules based on the above deficiencies.The main contributions are as follows:(1)Integrate positive association rules,negative association rules and rare rules mining algorithms in a unified framework model.The Markov logic network model does not need to set these metrics to be able to mine the frequently occurring set of binary variables in the transaction database.(2)Stochastic gradient descent method is used to learn the parameters in the Markov logic network model of association rules.the rules obtained by the Markov logic network framework model of association rules can well illustrate positive association rules,negative association rules and rare association rules.This paper makes use of two data sets with different sizes to perform the rule-precision accuracy test on the Markov logic network framework model algorithm and traditional association rule algorithm of association rules.The results show that compared with the traditional association rules algorithm,the rules obtained by the Markov logic network framework model of association rules have higher prediction accuracy.