基于条件随机场的稳健口语理解研究  

Research on Robust Spoken Language Understanding based on Conditional Random Fields

作　　者：程露红 黄浩[1] 马平[1] 

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《新疆大学学报（自然科学版）》2016年第1期88-93,共6页Journal of Xinjiang University(Natural Science Edition)

基　　金：国家自然科学基金(61365005;60965002)

摘　　要：语音识别性能的不理想造成的识别错误以及不符合文法的口语输入往往造成口语理解性能下降.针对这个问题,提出一种改进口语理解稳健性的方法.该方法通过在训练数据集中人工加入错误文本噪声进行语料扩充,再进行条件随机场口语理解模型训练,用得到的模型对具有噪声的未知数据集进行口语理解.实验结果表明该方法能提高口语理解的稳健性,较未加入噪声训练得到的模型在准确率、召回率及F1值上都有显著的提高.The performance of natural language understanding is often degraded by undesirability speech recognition errors and ill-formed inputs in spoken language. A new method for robust spoken language understanding based on conditional random fields is proposed. Erroneous texts are artificially added in the training data for corpus expansion to train the model parameters of conditional random fields, the model is applied to the unknown data sets with noise for spoken language understanding. Experimental results show the proposed method can improve the robustness of spoken language understanding. Significant precision, recall and Fl-score improvements can be obtained compared with the model trained on clean spoken text database.

关 键 词：口语对话系统 口语理解 条件随机场 稳健性 

分 类 号：TP317.52[自动化与计算机技术—计算机软件与理论]