基于集成主动学习策略的需求依赖关系自动提取

Automatic Requirement Dependency Extraction Based on Integrated Active Learning Strategies

由于需求依赖提取是一项认知困难且易出错的任务,因此本文提出了基于集成主动学习策略的需求依赖关系自动提取方法.基于集成主动学习策略,为不确定性概率、文本相似差异度、主动学习变体预测分歧度这3个不同角度的影响因子采用变异系数法确定相应的权重,给出衡量依赖对样本信息价值量的计算公式,从而选取综合评价值最高的前K个依赖对样本作为最优样本.随着最优样本不断加入到初始训练集中,主动学习模型提取需求依赖关系的性能快速提升,从而与其他主动学习策略相比,实现了同等采样数量下更高的需求依赖提取评估度量.最后,通过对比分析了多种方法的实验结果,证明了本方法的可行性.

Since requirement dependency extraction is a cognitively difficult and error-prone task,this paper proposes an automatic requirement dependency extraction method based on integrated active learning strategy.Based on the integration of active learning strategy,the coefficient of variation method was used to determine the corresponding weights of the impact factors from three different angles:uncertainty probability,text similarity difference degree and active learning variant prediction divergence degree.The calculation formula to measure the information value of dependency pairs is given,and the top K dependency pairs with the highest comprehensive evaluation value are selected as the optimal samples.As the optimal samples are continuously added into the initial training set,the performance of the active learning model for requirement dependency extraction is rapidly improved,thus achieves a higher evaluation measure of requirement dependency extraction with the same number of samples compared with other active learning strategies.Finally,the feasibility of the proposed method is proved by comparing and analyzing the experimental results of various methods.