Approaches to Constructing Metadata in Data Curation

Data curation, which attracts lots of attention in North America and Europe, can help researchers maintain, manage, reuse, and add value to data in its life cycle. The supplication of data curation depends on metadata, which allows researchers to retrieve, identify, and access data through describing data. The methods of constructing metadata in data curation are on its development; therefore the purpose of this thesis is to guide academic libraries by providing an outline about constructing metadata in data curation.
The research first discusses the meaning, fields, applications, and constructing steps of metadata in data curation, and then follows those steps to conduct metadata in practice. Firstly researcher integrates the fields of three kinds of metadata in data curation, including DataShare Profile, DataStaR minimum metadata, DataCite Metadata Schema from three foreign researches, resulting in 22 fields.Secondly researcher interviewed 12 professors in the National Taiwan University (NTU) for their opinions toward metadata, and described 12 kinds of data with those fields. Finally researcher revised the original fields with the catalogers in the NTU Library and ended up with 22 fields.
There are mainly four contributions of this research of metadata in data curation. They introduce the concept of metadata in data curation, provide the constructing steps and suggestions, and develop the fields based on foreign researches and opinions from professors and librarians in the NTU, and prompt researchers to view data from collection-level then making resource more easily discovered.
Some suggestions are found to be useful for constructing metadata in data curation. For data curation conductors, they may first train people to become data curation specialists, and invite domain experts to participate data curation. For metadata constructors, they should understand development of domain data repositories as well as researchers’ needs; they should also develop fields by starting from few subjects, as well as using semantic web and keeping attentions to the development of metadata in data curation. For those who describe data, they should consider how detailed each field should be, help researchers understand and accept collection-level metadata, and use semi-automatic description methods. In addition, for researchers in the future, they may focus on metadata for non-research data, develop fields based on actual applications, and construct fields concerning data life cycle.