3rd SMP2017ECDT评测报告_李大宇等1

has become the focus of academic and industrial research. Domain classification of user intention

is one of the most important research tasks, and its performance has a direct impact on the quality

of human-computer dialogue in specific domains. Aiming at the task of domain classification of

user intention, this paper discusses the influence of multiple features and multiple classifiers on

classification performance. On this basis, a multi-feature domain classification of user intention

system is designed, and achieved good results in The Evaluation of Chinese Human-Computer

Dialogue Technology (ECDT) of SMP2017.

Key words: Human-Computer Dialogue Technology, Domain classification of user intention,

multi-feature, classifier, SMP2017

Processing）举办了中文人机对话技术评测（ECDT）。本次中文人机对话技术评测分为两个

触发人机对话系统中的多个领域（domain），包括任务型垂直领域（如查询机票、酒店、公

的输入分类到相应的领域（domain）中，从而返回正确的回复结果。

我们山西大学文本挖掘团队参加了本次 ECDT 评测的任务一：用户意图领域分类。用