基于机器学习方法的动物遗传与育种学科热点趋势预测

Prediction of Hot Trends in Animal Genetics and Breeding Based on Machine Learning

采用Relim算法对科技文献数据进行热点主题词挖掘,利用时间序列集成实现未来一段时间的热点主题词预测。从Web of Science数据库中采集2000—2017年动物遗传与育种领域相关的71990研究论文作为研究对象,运用频繁项集算法Relim,对该领域的热点研究主题进行识别研究,并利用时间序列集成方法对热点研究主题在未来一段时间内的演变趋势进行预测。结果表明,该方法能较好地对某一个领域的热点主题词进行预测,且集成后的预测模型对时间序列预测效果较好,可帮助科研人员和政策制定者了解特定学科领域的主题热点状况。

The Relim algorithm was used to mine hot topic keywords in scientific and technical literature data,and time series integration was used to predict hot topic keywords in the future.From the Web of Science database,71990 research papers related to the field of animal genetics and breeding from 2000 to 2017 were collected as research objects,and the frequent itemset algorithm Relim was used to identify hot topics in the field,and a time series integration method was used to predict the evolution of hot research topics in the future.The results showed that this method could better predict hot topic keywords in a certain field,and the integrated prediction model had a better prediction effect on time series,which could help researchers and policy makers to understand the hot topics in specific subject areas.