Page 105 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 42

104 Journal of Library Science in China, Vol. 8, 2016

The top classification is used to classify concepts from different source thesauri based on their
disciplines, specialties or subjects. The top ontology can be used to classify concepts from different
source thesauri based on their intrinsic attributes. The top classification and ontology assemble
and integrate concepts in different ways. Both of them can facilitate the semantic integration of
concepts from multiple thesauri, and can also be used to navigate concepts in a thesaurus or cross-
thesauri.
A top classification must adopt a universal classification which is authoritative, strongly
influential, widely accepted and used in nearly all disciplines, as well as has good compatibility
with other systems. The classification system in the new CT (2014) is developed based on China
th
Document Classification (the 4 edition), and draw useful information from Chinese Library
Classification (the 5th edition). It inherits and innovates the two classifications, and responds the
demands of classifying information, terms and concepts. Clearly, at present, the classification of
the new CT is the best choice for the NTW semantic integration system.
The structure of thesaurus concept scheme is similar to ontology, but attribute space of a
thesaurus concept is not accurately defined, so the lateral association relationships between
concepts are unclear. Introducing top ontology is helpful for a thesaurus to evolve into ontology.
The top ontology’s classes assemble concepts with similar attributes by classifying thesaurus
concepts into related classes. Concepts in same class inherit the attribute space of the class,
which makes it possible to articulate lateral association relationships. To construct top ontology,
sufficient investigation should be performed focusing on features of hierarchical structures in the
source thesauri. And particular analysis should be undertaken to identify features of hierarchical
structures’ arrangement in various disciplines, to determinate facets, basic semantic types and
hierarchical framework. Attribute space of a class should be designed gradually and openly,
meanwhile discipline features and application demands should be considered.

2.2.2 Concept database and basic lexicon
The existence and development of NTW depend on the web. Therefore, the description of NTW
should follow semantic web specifications. In NTW, a knowledge description model is designed
based on SKOS+XL (see Section 4.1). Concepts and terms are separated, and terms are labels of
concepts. Semantic attributes are described on the concept layer, whereas non-semantic attributes
are described on the term layer. The member thesauri of NTW are logically independent, and
physically stored in consistent format. Concepts and their semantic attributes are assembled to
build concept database. Terms and their non-semantic attributes are deposited in basic lexicon.
Concepts in the concept database are linked to terms in the basic lexicon through labeling
attributes.
The basic lexicon is a database of terms and their attributes acquired not only from NTW
member thesauri but also from other sources, such as document keywords, user search
keywords, special dictionary terms, standardized nouns released by China National Terms