Estimation of office intellectual activities for not-distractive interruption

为了缓解办公室工作效率因工作中断而降低的问题，分析了办公室工作中断性与动作和声音的关系。结果表明，大的运动和声压的数量与可中断性，而由于案头工作的小运动的数量呈负相关。通过多元线性回归分析开发了一种估计算法，并将其应用于10天的数据集。实验结果验证了办公室可中断性自动估计的可行性，并基于人的语音的谐波和频偏特征，设计了一个办公室可中断性自动估计系统。除了PC操作活动特征外，还将会话持续和会话结束两种会话状态应用于工人的可中断性估计。估计实验证明了提高可中断性估计精度的可行性。

To relief the office productivity degrade due to interruptions, the relationship between office interruptibility and motions and sounds has been analyzed. The results demonstrated that the number of large motions and sound pressure correlate with interruptibility, while the number of small motions due to desk work negatively correlates. An estimation algorithm has been developed through multiple linear regression analysis and applied to a 10-days data set. The experimental results demonstrated the feasibility of automatic office interruptibility estimation.Furthermore, an automatic estimation system has been prototyped based on harmonicity and frequency deviation features of human voice. The two conversational statuses, duaring conversation and conversation end, were applied to the worker's interruptibility estimation in addition to PC operation activity features. The estimation experiments demonstrated the feasibility to improve the interruptibility estimation accuracy.