机器学习研究中的一个重要课题是如何有效结合已标识数据和未标识数据去推断未标识点标识.本文利用相近原则和直推方法求解这个问题.源于直观事实的相近原则是指“在输入空间中相近的对象其输出也相近”.为求得满足这个原则的半指导学习问题的解,我们给出了作为一般方法的基于相近原则的半指导问题直推学习机.得到解的解析表达和迭代算法,公式形式简洁,计算实现容易.给出实例验证该方法在解决实际问题中的有效性,并用图例与支撑向量机和半指导支撑向量机的解作了对比. An important issue in machine learning research is how to effectively combine the identified data with the unidentified data to deduce the unidentified point identification.In this paper, we use the similar principle and direct method to solve this problem.The similar principle from intuitionistic fact means that “ Similar objects in the input space have similar outputs. ”To find a solution to the semi-coherent learning problem that satisfies this principle, we give a direct covariance based semi-coherent direct learning machine as a general method. Analytic expression and iterative algorithm, the formula is simple and the calculation is easy to implement.Examples are given to verify the effectiveness of the method in solving practical problems, and compared with the solutions of support vector machine and semi-guided support vector machine.