Research on the Matthews Correlation Coefficients Metrics of Personalized Recommendation Algorithm Evaluation

摘要: The personalized recommendation systems could better improve the service for network user and alleviate problem of information overload in Internet. As we all know, key point being a successful system is performance algorithm. When scholars put forward some new algorithms, they claim that algorithms have been improved respects, than previous So need evaluation metrics to evaluate algorithm performance. Due scholar didn’t fully understand mechanism algorithms. They mainly emphasized specific like Accuracy, Diversity. What’s more, academia did not establish complete unified assessment which credibility do work evaluation. how this objective reasonable still challengeable task. In article, discussed present with its respective advantages disadvantages. Then, use Matthews Correlation Coefficient algorithm’s All based on an open source projects called mahout provides rich set components construct classic results experiments show applicability correlation coefficient relative