Abstract: Based on an empirical study of 37 university libraries’ 156 self-constructed databases, we summarized 8 existing characteristics nowadays: ① In terms of topics and content, thematic database is the main pattern of the database; ② In terms of project establishing, the main pattern is to establish the project by universities themselves; ③ In terms of software platform, libraries usually purchase such platforms from software vendors or database vendors while developing some features themselves; ④ In terms of personnel allocation, part-time is the mainstream model; ⑤ In terms of data sources and processing mode, libraries primarily process the data owned by universities by themselves; ⑥ In terms of the amount of data, the database is of mini-type; ⑦ In terms of retrieval method, single database retrieval is the mainstream mode; ⑧ In terms of users, the databases primarily serve the universities themselves. Based on the aforementioned analyses and problems, this paper put forward features of ideal databases to be constructed by university libraries.

Key words: Libraries of higher learning institutes, Self-constructed database, Model