Creating an Annotated Corpus for the Analysis of Causal Relations

In this paper, we report the results of our investigation of the characteristics of in-text causal relations. First, we designed causal relation tags. With our designed tag set, three annotators annotated 750 newspaper articles. Then, using the annotated corpus, we investigated the causal relation instances from three viewpoints: (1) cue phrase markers, (2) part-of-speech information, and (3) position in sentences. Our quantitative study shows that causal relation instances are represented in the several types of linguistic expressions.