Knowledge-Based Information Systems in Research of Regional History

Representation of data on regional history by means of Web
sites is practised in many countries, because such Web sites
should perform – and more often than not they really perform
– three important functions. First, they further preservation
of historical information related to a certain region. Second,
Web sites on regional history promote and publicize historical
and cultural heritage of a region, hence they can serve as a
specifi c ‘visiting-card’ of a region. Third, they facilitate progress
in regional historical studies by means of involving researchers
in Web-community and providing them with primary data
and, at the same time, with appropriate research tools that
can be used for data retrieving and processing. However,
implementation of the above-mentioned functions calls for
different information technologies. For instance, preservation
of historical information requires structuring of information
related to a certain region, therefore databases technologies
should be used to perform this function. In order to draft and
update a ‘visiting-card’ of a region, a Web site can be designed on
the basis of Content Management System (CMS). Meanwhile,
coordination of research activities within the frameworks of
a certain Web-community calls for implementation of modern
technologies in order to create knowledge-based information
systems.
In this paper, a concept of a knowledge-based information
system for regional historical studies is framed and a pattern of
data representation, aggregation, and structuring is discussed.
This pattern can be described as a specialized semantic network
hereafter defi ned as Historical Semantic Network (HSN).
Up to now, the most popular pattern of representation
of regional historical information is a traditional Web site
that consists of static HTML-pages. A typical example is the
Web site Latgales Dati (created in 1994, see http://dau.lv/ld),
which provides information about the history and cultural
heritage of Eastern Latvia – Latgale (Soms 1995; Soms and
Ivanovs 2002). On the Web site Latgales Dati, historical data
are stored and processed according to the above-mentioned
traditional pattern. The shortcomings of this pattern are quite
obvious: it is impossible to revise the mode of processing and
representation of data on the Web site; it is rather diffi cult to
feed in additional information, to retrieve necessary data, and
to share information.
In the HEML project, a special language of representation
and linkage of information on historical events has been
developed (http://www.heml.org). Advantages of this language
are generally known: a modern approach to processing and
representation of historical information on the basis of XML,
a developed system of tools of visualization of information,
etc. However, the processing of information is subjected to
descriptions of historical events. It can be regarded as a drawback
of this language, since such descriptions do not form a solid
basis for historical data representation in databases. Actually,
any description of a historical event is an interpretation
of narratives and documentary records. For this reason,
databases should embrace diverse historical objects, namely,
documents and narratives, persons, historical monuments and
buildings, institutions, geographical sites, and so on, and so
forth. These objects can be described using different attributes
simultaneously, which specify characteristic features of the
objects. Sometimes, objects can be labelled with defi nite, unique
attributes only. The detailed, exhaustive information about
the objects can be represented by means of links between
objects, thus creating a specifi c semantic network. Within
this network, paramount importance is attached to special
temporal objects – the chronological ‘markers’. Connecting
links between a number of historical objects, including special
temporal objects, will form rather a solid basis for historical
discourse upon historical events, processes, and phenomena.
The fi rst characteristic feature of the semantic network, which
is being created on the basis of the Web site Latgales Dati,
is the principle role assigned to such historical objects as
documentary records and narratives. Thus, the connecting links
between the objects are constituted (actually, reconstructed) in
accordance with the evidences of historical sources. In order
to provide facilities for verifi cation of the links, the semantic
network should contain full texts of historical records as well
as scanned raster images of them. On the one hand, texts
and images can be directly connected with the objects of the
semantic network by means of SVG technology; on the other
hand, this connection can be established by means of mark-up
schemes used in XML technology (Ivanovs and Varfolomeyev
2005).
The second characteristic feature of HSN is fuzziness of almost
all aspects of information in the semantic network (Klir and
Yuan 1995). The source of this fuzziness is uncertainty of expert
evaluations and interpretations of historical data; moreover,
evidences of historical records can be either fragmentary and
contradictory, or doubtful and uncertain. Therefore, the level of
validity of information is ‘measured’ by variables, which can be
expressed by ordinary numbers (0–1). However, it seems that
they should be expressed by vector quantities as pairs (true,
false); the components true and false can assume values from
0 to 1. In this case, the pair (1,1) refers to very contradictory
and unreliable information, meanwhile the pair (0,0) means the absence of information. This approach was used in
construction of four-valued logic (Belnap 1977). The fuzziness
values defi ned for some objects can be propagated within a
semantic network by means of logical reasoning (Hähnle 1993).
This ‘measurement’ of validity of data is performed by users of
HSN; the results of ‘measurement’ should be changeable and
complementary, thus refl ecting cooperation of researchers
within the frameworks of Web-community.
Creating a semantic network, a number of principal
operations should be performed: reconstruction, linkage, and
representation. Reconstruction is either manual or automatic
generation of interrelations between objects that are not
directly refl ected in historical records. In this stage, new
historical objects can emerge; these objects are reconstructed
in accordance with the system of their mutual relations.
Linkage can be defi ned as detection of similar (in fuzzy sense)
objects, which within a semantic network (or within a number
of interrelated networks) form a defi nite unit. Representation is
a number of operations of retrieving of data from the semantic
network including representation of information by means of
tables, graphs, timeline, geographic visualization, etc.
HSN as a basic pattern of representation of historical
information in Internet provides researchers with proper
tools, which can easily transform data retrieved from such
Web sites. However, some problems should be solved in order
to ‘substitute’ the traditional Web site Latgales Dati for HSN.
There are different approaches to representation of semantic
networks in Internet. A generally accepted approach is division
of network links into subjects, predicates, and objects followed
by their representation by means of RDF. However, the RDF
pattern (actually, directed graph in terms of mathematics) is not
adequate to the hyper-graph pattern accepted in HSN, since
in HSN one and the same link can connect different objects
simultaneously. In order to represent hyper-graphs, different
semantic networks – such as WordNet (Fellbaum 1998),
MultiNet (Helbig 2006), topic maps (Dong and Li 2004) or even
simple Wiki-texts interconnected by so-called ‘categories’ –
should be used. Unfortunately, the above-mentioned semantic
networks do not provide proper tools to perform the tasks
set by HSN. For this purpose, the experience acquired by
designers of universal knowledge processing systems (e.g.
SNePS, see Shapiro 2007, or Cyc, see Panton 2006) should be
taken into consideration.
One more important problem is linkage of different HSN,
including Latgales Dati. Such linkage can be conducive to
cooperation of researchers, since it ensures remote access
to historical data and exchange of information and results of
research work. It seems that Web-services technology can
carry out this task. In this case, HSN should be registered in
a certain UDDI-server. A user’s query activates interaction of
a Web-server with numerous related Web-servers, which are
supplied with HSN-module. As a result, a semantic network
is being generated on the basis of fragments of different
networks relevant to the initial query.
The principles of creation of HSN substantiated above are
used in designing a new version of the knowledge-based
information system Latgales Dati. As software environment
a specialized Web application created at Petrozavodsk State
University is applied (http://mf.karelia.ru/hsn).
References
Belnap, N. A Useful Four-Valued Logic. In J.M. Dunn and G.
Epstein, ed. Modern Uses of Multiple-Valued Logic. Dordreht,
1977, pp. 8-37.
Fellbaum, C., ed. WordNet: An Electronic Lexical Database.
Cambridge, Mass., 1998.
Hähnle, R. Automated Deduction in Multiple-Valued Logics.
Oxford, 1993.
Helbig, H. Knowledge Representation and the Semantics of
Natural Language. Berlin; Heidelberg; New York, 2006.
Ivanovs, A., Varfolomeyev, A. Editing and Exploratory Analysis
of Medieval Documents by Means of XML Technologies. In
Humanities, Computers and Cultural Heritage. Amsterdam, 2005,
pp. 155-60.
Klir, G.R., Yuan, B. Fuzzy Sets and Fuzzy Logic: Theory and
Applications. Englewood Cliffs, NJ, 1995.
Dong, Y., Li, M. HyO-XTM: A Set of Hyper-Graph Operations
on XML Topic Map toward Knowledge Management. Future
Generation Computer Systems, 20, 1 (January 2004): 81-100.
Panton, K. et al. Common Sense Reasoning – From Cyc to
Intelligent Assistant. In Yang Cai and Julio Abascal, eds. Ambient
Intelligence in Everyday Life, LNAI 3864. Berlin; Heidelberg;
New York, 2006, pp. 1-31.
Shapiro, S.C. et al. SNePS 2.7 User’s Manual. Buffalo, NY, 2007
(http://www.cse.buffalo.edu/sneps/Manuals/manual27.pdf).
Soms, H. Systematisierung der Kulturgeschichtlichen
Materialien Latgales mit Hilfe des Computers. In The First
Conference on Baltic Studies in Europe: History. Riga, 1995, pp.
34-5.
Soms, H., Ivanovs, A. Historical Peculiarities of Eastern Latvia
(Latgale): Their Origin and Investigation. Humanities and Social
Sciences. Latvia, 2002, 3: 5-21.