<p>Human resources data usually contains diversified information involving different positions, industries, regions, and other aspects. It is uneven and complex, such as market fluctuations, and cannot extract adequate information, making human resources recommendations’ accuracy low. To improve recruitment efficiency, a non-uniform human resource balanced recommendation algorithm is proposed considering the needs of enterprise users. Obtain the enterprise user demand preference data according to the recruitment websites of domestic and foreign enterprises, clean and clear the redundant data in the collected data through the kettle tool, and form a dataset. Using the pre-processed enterprise user demand preference data, the collaborative filtering recommendation algorithm and the bipartite graph recommendation algorithm are effectively combined, and a bipartite graph recommendation algorithm showing preferences is proposed. Through this algorithm, the initial value of job seekers and the human resource allocation coefficient in the bipartite graph recommendation algorithm is adjusted to realize the uneven recommendation of human resources for enterprise users. Through experimental verification, it is known that this algorithm can achieve human resources recommendation, the recommendation results meet the needs of enterprise users, the recommendation effect is better, the generalization ability is strong, and it can more comprehensively solve the problem of enterprise talent supply.</p>