Genetic algorithm and novel similarity calculation strategy-based collaborative filtering recommendation algorithm

Abstract

The invention relates to a genetic algorithm and novel similarity calculation strategy-based collaborative filtering recommendation algorithm. According to the algorithm, the individual score average value and variance of each user are sufficiently utilized in an object similarity calculation scheme, and are used in an object recommendation-based collaborative filtering algorithm. Moreover, a weighted value in a similarity calculation formula is automatically adjusted by utilizing a genetic algorithm, and a mean absolute deviation value in a modeling process of a recommendation algorithm is controlled through the genetic algorithm. Experimental contrasts and result analysis indicate that the genetic algorithm and novel similarity calculation strategy-based collaborative filtering recommendation algorithm disclosed by the invention has the advantage that the indexes such as recommendation correctness, recall rate and the like are significantly improved.

Description

Technical field [0001] The invention relates to a personalized recommendation system technology, in particular to a collaborative filtering recommendation algorithm based on a genetic algorithm and a novel similarity calculation strategy. Background technique [0002] With the rapid development of the Internet and wireless communication technologies, massive amounts of information have been brought to people's daily lives. How to obtain useful information in the era of information explosion has become an urgent problem for scientific and technological workers. Personalized recommendation is an important means to solve information explosion and information overload. It is based on the user's interest characteristics and purchasing behavior to recommend information and products that users may be interested in. Recommendation algorithm is an important part of personalized recommendation. It uses knowledge such as mathematics and computer algorithms to infer what users might like. [...

AI Technical Summary

Problems solved by technology

The recommendation performance obtained by the similarity calculation scheme that does not consider the factor of user's personalized scoring habits is not ideal

On the other hand, the similarity calculation scheme will involve the assignment of multiple weight factors. Traditional methods use empirical values ​​or set the value of weight factors through manual debugging. These methods are time-consuming and laborious and have poor results.

Images