Page 175 - Journal of Library Science in China 2020 Vol.46

174
174 Journal of Library Science in China, Vol.12, 2020

to have a comprehensive and global perception of the intelligence environment but also have to
deeply understand the intelligence environment issues by merging background knowledge and tacit
knowledge of related issues.
Secondly, the article summarizes the existing knowledge fusion research and practice in the
field of knowledge science, and divides it into three development stages according to the main
research content and characteristics: agent-based knowledge fusion, pattern-based knowledge
fusion, and machine learning-based knowledge fusion. Agent-based knowledge fusion is an early
representative of knowledge fusion research. Its core task is to solve the problem of knowledge
sharing, reuse and transformation in a distributed information environment, and to implement
knowledge search and extraction of many knowledge resources through middleware technology.
Pattern-based knowledge fusion mainly realizes the fusion of knowledge by changing and
reorganizing the internal and external structure and attributes of knowledge driven by multi-source
ontology. Its main goal is to solve the problem of contextual situation awareness and decision-
making in complex scenarios. Machine learning-based knowledge fusion mainly uses automatic
methods such as machine learning to achieve automatic knowledge extraction and learning of open
data, the establishment of knowledge links, and the unification of knowledge. Finally, automatic
construction of large-scale knowledge bases and automatic organization of knowledge are
achieved.
Finally, combined with the intelligence practice in recent years, this paper proposes a general
framework for knowledge fusion research, including thought domain, theory domain, technology
domain, and application domain. The thought domain is our set of guiding thought for dealing with
complex, changeable, and deeply uncertain intelligence environments. It is used to guide and lead
intelligence research and intelligence work ideologically and theoretically. The thought domain of
knowledge fusion in the field of intelligence is composed of fusion thinking, intelligence thinking,
and computational thinking, and the three kinds of thinking complement each other. The main
basic or source disciplines of these three kinds of thinking are cognitive science, information
science, and computational science, respectively. The theory domain of knowledge fusion mainly
includes four aspects: knowledge fusion theory, knowledge fusion framework, knowledge fusion
model, and knowledge fusion method. Technology domain refers to different types of intelligence
and cognitive activities. Knowledge fusion models, methods, frameworks, software and other
technical elements are different. In this way, a set of technical elements oriented to a specific task
or solving a type of problem forms different technologies field. The current application fields
of knowledge fusion include financial intelligence, public safety, business analysis, military
intelligence, scientific forecasting, and public opinion management. We believe that with the
continuous development of technology, the application space of knowledge fusion will become
wider and wider.