Corpus is a kind of important resource for knowledge acquisition in the natural language processing （NLP）. However, up to now, in the biomedical domain comparatively fewer corpus focus on semantic association among all tokens in a sentence. We proposed an annotation scheme based on feature structure theory for enriching biomedical domain corpora with token semantic association （TSA）. There are 227 documents of the BioNLP GE ST training data annotated to form TSA corpus in which each annotated item shows a token semantic association that appears as a triple. The annotation of token semantic association has the potential to significantly advance biomedical text mining by providing rich token semantic information for NLP systems especially for the sophisticated IE systems, such as bio-event extraction.
In this paper we propose a novel model ＂recursive directed graph＂ based on feature structure, and apply it to represent the semantic relations of postpositive attributive structures in biomedical texts. The usages of postpositive attributive are complex and variable, especially three categories： present participle phrase, past participle phrase, and preposition phrase as postpositire attributive, which always bring the difficulties of automatic parsing. We summarize these categories and annotate the semantic information. Compared with dependency structure, feature structure, being recursive directed graph, enhances semantic information extraction in biomedical field. The annotation results show that recursive directed graph is more suitable to extract complex semantic relations for biomedical text mining.
