Efficient Intrusion Detection with KNN Classification and DS Theory - Proceedings of All India Seminar on Biomedical Engineering 2012 (AISOBE-2012) - page 173

Statistical rules are used in general in the classification of textual information,
which include several tasks in Information Retrieval. It includes not only the
determination of good documents in terms of relevance attending to user needs,
but also the classification of documents into categories (topics) attending to prede
ned classes [ 3 ]. In the following, we include studies found in the literature about
both the retrieval and the categorization tasks.
The use of rules for categorization comes from a process of classification of
documents into different categories regarding their topics in order to optimize a
posteriori retrieval process. One of the most relevant works of categorization using
rules is the one of [ 4 ]. The general idea of this work is the discovery of classifi-
cation patterns automatically for document categorization. The aim of the induc-
tion process is for sets of decision rules to distinguish among different categories
which documents belong to. The attributes of the rules can be a word or a pair of
words constructing a dictionary where an elimination process of the less frequent
words is carried out. Finally, association rules have also been used for categori-
zation [ 5 ], where the authors propose a solution for text categorization based on
the application of the best generated association rules to build a classifier.
The Dempster-Shafer Theory
The Dempster-Shafer theory (DST) of evidence originated in the work of [ 6 , 7 ]on
theory of probabilities with upper and lower bounds. It has since been extended by
numerous authors and popularized, but only to a degree, in the literature on
Artificial Intelligence (AI) and expert systems, as a technique for modeling rea-
soning under uncertainty. In this respect it can be seen to offer numerous advan-
tages over the more ''traditional'' methods of Statistics and Bayesian decision
theory. Hajek [ 8 ] remarked that real, practical applications of DST methods have
been rare, but subsequent to these remarks there has been a marked increase in the
applications incorporating the use of DST. Although DST is not in widespread use,
it has been applied with some success to such topics as face recognition [ 9 ],
statistical classification [ 10 ], and target identification [ 11 ]. Additional applications
centered on multisource information, including medical diagnosis [ 12 ] and plan
recognition [ 13 ]. An exception is the paper by Cortes-Rello and Golshani [ 14 ],
which although written for a computing science-AI readership does deal with the
''knowledge domain'' of forecasting and Marketing Planning. For those with even
limited knowledge of these domains the paper appears rather naive. Referring for
example to rather venerable old editions of standard texts such as [ 15 ]. The aim of
this paper is to suggest that there is a good deal of potential in the DST approach,
which is as yet largely unexploited. The origins of the mathematical theory of
probability date back at least to the work of the eighteenth century scholar, The
Reversed Thomas [ 16 ], whose work was published posthumously in 1763.
It provides the foundations for the theory of statistical inference (involving both
estimation and testing of hypotheses) and for techniques of design making under