如今越来越多的乘客选择乘坐舒适快捷的飞机出行,中国航空运输需求因此逐年增长,航空公司在获得更多盈利空间的同时也面临激烈的竞争.对航空公司的用户评论进行意见信息抽取,不仅可用于航空公司改进服务质量和用户体验, 还可为用户选择满意的航空公司提供参考.文章首次以新浪微博平台上航空公司的用户评论为基础数据,利用条件随机场进行意见信息抽取.在有关研究中,专家学者大多凭借以往知识的了解对特征对象和特征词进行人工标注,鲜少分析用户在本评论语料中的关注点.因此,文章创新性地在人工标注前首先利用TF-IDF 算法进行关键词提取,找到本评论语料中用户的关注点,最后以超过93\% 的$F$平均值证明模型的有效性,为后续的研究提供了新方向.

Nowadays, the requirements of air transportation increased year after year, whether the business man or the common people are more and more willing to choose the comfortable and speedy plane as their transportation. The airline get the more profit, as well, the competition between the different airlines is intense. Not only does it can improve quality of service and user experience, but also provide more reference for passengers by using the comment of airline based on information extraction. In this paper, we use the conditional random field to extract information of comments based on the comments of airlines from Sina Weibo. In the relevant study, most experts and researchers utilize manually annotated to deal with feature object and feature words on the basis of past knowledge. Nobody has made a careful analysis of users' concerns of comments corpus. Therefore, we creatively use the TF-IDF algorithm to extract keywords before manually annotated and find the users' concerns in comment corpus. Finally, it proves the validity of the model by $F$-value which is more than 93\%, and it provides a new direction for the follow-up study.