Phenotypic characters are described in published works mostly using human languages ("natural languages"). They are valuable knowledge but not amenable to computational analyses. Text mining algorithms have been developed to extract useful information from the text, but the extracted information needs to be ontologized to ensure “apples are compared to applesâ€.  Highly trained biology researchers are taking on the role of biocurators to convert characters expressed in human languages into formal statements, for example, EQ (Entity-Quality) statements, using ontologies. During the process, terms encountered in the descriptions are added to ontologies. Grounding phenotypic data in the rich literature brings the knowledge accumulated in the past several hundred years to life. However, biologists in general are not involved in the ontology building process, and natural language descriptions are continuously being published. In this talk, we will discuss the issues encountered in building and using existing ontologies for phenotypic data curation and present our progress in creating ontology building tools biologists can use. We will discuss ontology building issues that contribute to inter-curator variations and ontology design patterns for phenotypic characters that have the potential of reducing the variations.
