Firstly, employing the context-sensitive attribute grammar of HPSG, FAFS establishes the S-structures of UDCs, distinguishes between strong and weak UDCs based on HPSG syntactic attributed structure trees, and extracts the corresponding fillers for the traces. 首先,我们采用了HPSG的上下文有关属性文法。FAFS建立UDC句子的S-结构,区分基于HPSG句子属性结构树的强弱UDC句子,提取与语迹相对应的填充词。