企业竞争合作关系的判断基础之一在于专利。在大数据时代背景下，专利数量激增，企业潜在关系预测既要合理运用方法缩小检索范围，又要结合时序概念进行层次推荐。以钙钛矿型太阳能电池为例，基于word2vec词向量模型与LDA模型对专利文本进行数据挖掘与语料库扩充，在技术生命周期理论基础上采用协同过滤推荐算法预测企业间竞合关系，构建梯度系数进行竞合强度判断。研究结果表明：在钙钛矿型太阳能电池领域，基于时间维度上的协同过滤思想适用于企业间潜在关系判断，算法的预测结果也通过了实证检验；同时，基于专利视角，企业间的竞合关系分为潜在强合作型、潜在弱合作型、潜在强竞争型和潜在弱竞合型。研究厘清企业间存在的潜在竞合关系，为今后企业检索潜在关系对象提供新的方法。

One of the bases for judging the competitive and cooperative relationship of enterprises is patent. In the context of big data era, the number of patents has soared, and the prediction of potential relationship between enterprises should not only use reasonable methods to narrow the search scope, but also combine the concept of time series to make hierarchical recommendation. Taking perovskite solar cells as an example, based on word2vec word vector model and LDA model, data mining and corpus expansion of patent text were carried out. Based on the theory of technology life cycle, collaborative filtering recommendation algorithm was used to predict the competition and cooperation relationship between enterprises, and a gradient coefficient was constructed to judge the competition and cooperation intensity. The research results show that in the field of perovskite-type solar cells, the idea of collaborative filtering based on the time dimension is applicable to the judgment of potential relationships between enterprises, and the prediction results of the algorithm have also passed the empirical test; At the same time, based on the patent perspective, the competition and cooperation between enterprises can be divided into potential strong cooperation, potential weak cooperation, potential strong competition and potential weak competition and cooperation. Research and clarify the potential competition and cooperation relationship between enterprises, and provide new methods for enterprises to retrieve potential relationship objects in the future.

本文系山西省哲学社会科学规划课题“山西省中小企业间数据融合安全共享机制研究”（2022YY097）；中国教育技术协会重点项目“财经专业背景下大数据核心技术产学研用多维一体的创新性研究”（XJJ202205014）；中国高等教育学会专项课题“大数据技术与应用产教融合实践与创新”（21CJZD06）；山西省高等学校教学改革创新项目“大数据技术与应用产学协同改革研究与实践”（J2021294）研究成果之一。