From Language Archives to Digital Museums Synergizing Linguistic Databases

Chu-Ren Huang, Feng-Ju Lo, Hui-Jun Hsiao, Chiu-Jung Lu, and Ching-chun Hsieh

Language archives contains rich information beyond the explicitly encoded
linguistic level. However, since these linguistic databases are designed
with specific purpose(s), it is usually not easy to obtain implicit
information from them. This paper shows how several language archives can
be combined together to generate new information as well as to make
information more accessible. These methodologies are experimented in two
digital museum sites: WenGuo http://www.sinica.edu.tw/wen and SouWenJieZi
http://www.dmpo.sinica.edu.tw/~words. We will refer to this process of
knowledge collocation and generation as `synergizing' of digital archives
and will discuss the approaches with examples from the two sites.
First, archive synergizing must be anchored with Lexical KnowledgeBase
(LKB). The natural information units that play a role in all types of
linguistic archives are lexical items (or words). Hence, lexical items
provide a basis for information manipulation across these archives. There
are two aspects of anchoring with the LKB. On one hand, the archives must
be tokenized in a way that is compatible with the LKB. Lexical resources
are tokenized by nature while textual resources must be tokenized based on
the LKB anchor. On the other hand, the LKB anchor itself can have a rich
information structure that can be projected when synergizing. Take WenGuo
for example, the WenGuo LKB is based on three different elementary school
Chinese textbooks. The explicit marking of the first introduction of each
lexical item provides a natural index of linguistic ability.
Second, once all archives are linked together, lexical tokens as
information units can be projected to achieve instantiation (e.g. actual
realization in various texts) or elaboration (e.g. entries in other
special lexicons). The knowledge that can be inferred from the
instantiation and/or elaboration depends on the user's design and
interpretation. For instance, SouWenJieZi give instantiations in the four
Chinese classics (among other texts), Lunyu, Menzi, Laozi, Zhuangzi. They
happen to be the great books of Confucianism and Taoism respectively. So
the (non-)instantiation of a certain concept (as words) has implications
for the two main schools of thoughts.
	Third, redundant information of language archives can also
provided by the LKB anchor. For instance, the Chinese texts do not
explicitly represent pronunciation. This information, as well as
conversion among different romanization systems, can be projected from the
LKB.
	In conclusion, the synergizing procedure can be extended to all
language-based archives, such as all textual databases. Such applications
would have great implications for how knowledge is acquired in digital
archives.