Page 118 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2015 Vol. 41
P. 118
Ying WANG, Zhixiong ZHANG, Hui SUN & Feng LEI / A method of semantic representation and 117
organization of the historic knowledge on contemporary China
the contents about “The Third Plenary Session of Eleventh Central Committee”. The sentence
“The Third Plenary Session of Eleventh Central Committee, which held in Beijing on December
18 to 22, 1978, has profound significance in the history of communist party of China since the
establishment” implies some facts: the holding time of “The Third Plenary Session of Eleventh
Central Committee” is “December 18-22, 1978”, the location is “Beijing”, and so on. According to
the datatype and object properties defined in the historical ontology, we collected many predicate
verbs, such as “hold”, “convene” , “take place”, etc, and developed extraction rules of “conference-
time”, “conference-location” and so on. With syntactic analysis and extraction rules, semantic
relations hidden in sentences were extracted from text items, forming a series of fact triples as the
basis of constructing the properties and relations of instances.
While automatic processing can help to find some potential knowledge, because of the
complexity of natural language, the accuracy of the results from text mining still cannot be
guaranteed and thus cannot be directly added into the ontology. Instead, these need to be identified,
complemented and revised by human experts based on their own domain knowledge and with
reference to original text items.
2.4 Build knowledge graph based on the associations of knowledge objects
Through the above process, historical reference books were divided into text items, which are
regard as the knowledge units of knowledge content, and knowledge objects and their relations
were extracted from text items. This forms the “book-item-fact-object” procedure of the
“Mining down” method. The extracted knowledge objects and facts enrich the historical ontology
by forming the knowledge object layer and the fact layer. At the same time, the links between
knowledge objects, facts and text items build a complex network with the associations on the three
layers.
As shown in Figure 4, a text item “The Third Plenary Session of Eleventh Central Committee
of the Communist Party of China” in “Dictionary of the history of the Chinese Communist Party”
shows the facts about holding time, location, attending members, and related events of The
Third Plenary Session of the Communist Party of China. The text items with the same title in
“Encyclopedia of the National History of the People’s Republic of China” and “Chronicle of the
People’s Republic of China” not only contain the above facts but also reveal its related concepts
of “Emancipate the Mind” and “Seek Truth from the Facts”. In the text item “The great historical
turning point” of “Conspectus of Chinese modern history”, the occurrence time, place, related
conference, and related event are displayed, in addition to the facts such as related persons and
conferences of “The 11th National Congress of the Communist Party of China”, related persons
of “The movement to criticize the ‘Gang of four’”, and so on. With the use of text mining
technologies and domain knowledge of historical experts, the internal knowledge was discovered