The invention relates to the technical field of recommendation methods, in particular to a multi-target recommendation method based on auxiliary evolution, which comprises the following steps of: firstly, calculating a comprehensive
score of each individual, sorting the individuals, and storing the individuals in a candidate set; selecting individuals from the candidate set to form a main parent
population P and auxiliary parent populations P'and P '', and performing
crossover operation and
mutation operation on the parent populations P, P 'and P' 'to form populations P2, P' 2 and P '' 2; elite strategy is carried out on the populations P2, P '2 and P' '2, and redundant populations are eliminated; according to a sorting result and a calculation result, selecting a corresponding number of individuals and storing the individuals in the parent populations P3, P '3 and P' '3; using a parent
population P3; the individuals in the parent populations P, P 'and P' 'replace the individuals in the parent populations P, P 'and P' ', when the three parent populations evolve a certain algebra T1, knowledge migration is carried out among the three parent populations to form
advantage complementation, when the three parent populations evolve a certain algebra T2, migration learning is carried out among the users to reduce the situation of falling into
local optimum, loop iteration
processing is carried out according to the content, and the probability of falling into
local optimum is reduced. And after an iteration stopping condition is met, stopping iteration, and outputting a recommendation result. By screening the data, the
data processing precision can be effectively improved, and recommendation of two targets of accuracy and diversity is realized.