Certain Investigation on Dynamic Clustering in Dynamic Datamining

Integrated Intelligent Research (IIR) International Journal of Data Mining Techniques and Applications
Volume: 02 Issue: 02 December 2013 Page No.71-75
ISSN: 2278-2419
71
Certain Investigation on Dynamic Clustering in
Dynamic Datamining
S.Angel Latha Mary1
,Dr.K.R.Shankar Kumar2
1
Associate Professor, CSE Department, Karpagam college of Engineering,Coimbatore, India
2
Professor, ECE Department, Ranganathan Engineering College
Email : xavierangellatha@gmail.com, shanwire@gmail.com
Abstract - Clustering is the process of grouping a set of objects
into classes of similar objects. Dynamic clustering comes in a
new research area that is concerned about dataset with dynamic
aspects. It requires updates of the clusters whenever new data
records are added to the dataset and may result in a change of
clustering over time. When there is a continuous update and
huge amount of dynamic data, rescan the database is not
possible in static data mining. But this is possible in Dynamic
data mining process. This dynamic data mining occurs when
the derived information is present for the purpose of analysis
and the environment is dynamic, i.e. many updates occur.
Since this has now been established by most researchers and
they will move into solving some of the problems and the
research is to concentrate on solving the problem of using data
mining dynamic databases. This paper gives some
investigation of existing work done in some papers related with
dynamic clustering and incremental data clustering.
Keywords- Clustering, incremental data clustering dynamic
clustering, dynamic data mining, Cluster evaluation, Cluster
validity Index.
I. INTRODUCTION
Clustering is the process of grouping a set of objects into
classes of similar objects. This data objects are similar to one
another within the same cluster and dissimilar to the objects in
the other clusters [1]. Clustering is a challenging problem in
scalability, ability to deal with different types of attributes,
discovery of cluster with arbitrary shape, domain knowledge to
determine input parameters, updating new dataset and
visualizing high-dimensional sparse data simultaneously [2].
Dynamic clustering comes in a new research area that is
concerned about dataset with dynamic aspects. It requires
updates of the clusters whenever new data records are added to
the dataset and may result in a change of clustering over time.
For example, the bank customer is interested in obtaining his
current account status. An economic analyst can receive a lot
of new articles every day and he would like to update the
relevant associations based on all current articles. Recent
developments of clustering systems uses dynamic data which
are concerned about the clustering process in dynamically
[3][4] .
II. DYNAMIC DATA MINING (DDM)
Data mining, the process of knowledge discovery in databases
(KDD), is concerned with finding patterns in the raw data and
finds useful information or to predict trends. Recently, the data
are growing with unpredictable rate. Discovering knowledge in
these data is a very expensive operation [5]. Running data
mining algorithms each time when there is a change in data is a
challenging problem. Therefore updating knowledge
dynamically will solve these problems. Dynamic data mining
is a shift from static analysis to dynamic analysis which
discoverers and updates knowledge along with new updated
data [3]. Dynamic data mining is very useful to obtain high
quality results in the field of time series analysis,
telecommunications, mobile networking, nanotechnology,
physics, chemistry, biology, health care, sociology and
economics [6]. When there is a continuous update and huge
amount of dynamic data, rescan the database is not possible in
static data mining. But this is possible in Dynamic data mining
process [7].
Dynamic data mining applies data mining algorithms on
dynamic database. It updates existing set of clusters
dynamically. A data warehouse is not updated immediately
when insertions and deletions takes place in the databases.
These updates are applied to the data warehouse periodically,
e.g. each night. This dynamic data mining occurs when the
derived information is present for the purpose of analysis and
the environment is dynamic, i.e. many updates occur [8][9].
The data mining tasks clustering, classification and
summarization use this dynamic data mining. Some examples
are some set of customers (clustering sales transactions),
Symptoms of distinguishing disease A from disease B
(classification in a medical data warehouse), Description of the
typical WWW access patterns (summarization in the data
warehouse of an internet provider).
Some factors for dynamic data mining.
 Whenever a database may have frequent updates.
 Database modified for every insertion or deletion.
 For every modification (insertion and deletion) the
database requires rescanning again. So the time
complexity is high.
 Hence the incremental clustering algorithm includes the
logic for insertion and deletion of the databases as separate
dynamic operation.
 After insertions and deletions to the database, the existing
clusters have to be updated.
 When the data is inserted or deleted the rescanning of the
whole database increases the time complexity. The system
with dynamic concept will not rescan the database, but it
updates the new data existing data. So it is more efficient
and suitable to use in a large multidimensional dynamic
database.
III. DYNAMIC CLUSTERING

ISSN: 2278-2419
72
Clustering and visualizing high dimensional dynamic data is a
challenging problem in the data mining. Most of the existing
clustering algorithms are based on the static statistical
relationship among data. In the clustering process, there are no
predefined classes and no examples that would show what kind
of desirable relations should be valid among the data [2]. The
databases will dynamically change due to frequent insertions
and deletions which changes clustering structure over time.
Completely reapplying the clustering algorithm to detect the
changes in the clustering structure and update the uncovered
data patterns following such deletions and insertions is very
expensive for large high dimensional fast changing dataset.
Dynamic clustering is a mechanism to adopt and discover
clusters where a data set is updated periodically through
insertions and deletions. There are many applications such as
incremental data mining in data warehousing applications,
sensor network which relies on dynamic data clustering
algorithms.
IV. EVALUATING THE QUALITY OF THE
RESULT CLUSTERS
The choice of number of the clusters is an important sub
problem of clustering. Since it needs a priori knowledge in
general and the vector dimensions are often higher than two,
which do not have visually apparent clusters. The solution of
this problem directly affects the quality of the result. If the
number of clusters is too small, different objects in data will
not be separated. Moreover, if this estimated number is too
large, relatively regions may be separated into a number of
smaller regions [10]. Both of these situations are to be avoided.
This problem is known as the cluster validation problem. The
aim is to estimate the number of clusters during the clustering
process. The basic idea is the evaluation of a clustering
structure by generating several clustering for various number
of clusters and compare them against with some evaluation
criteria. The procedure of evaluating the results of a clustering
algorithm is known as cluster validity. In general terms there
are three approaches to investigate cluster validity. The first is
based on external criteria. This implies that the results of a
clustering algorithm is evaluated based on a pre-specified
structure, which is imposed on a data set and reflects user
intuition about the clustering structure of the data set [11]. The
indices Rand Statistic, Jaccard Coefficient, Fowlkes–Mallows
index, Hubert’s statistic are used to measure the degree of
similarity based on external criteria.
The second approach is based on internal criteria. In this case
the clustering results are evaluated in terms of quantities that
involve the vectors of the data set themselves (e.g. proximity
matrix). The following methods can be used to assess the
quality clustering algorithms based on internal criterion:
Davies–Bouldin index, Dunn index and F-measure. The third
approach of clustering validity is based on relative criteria.
Here the basic idea is the evaluation of a clustering structure by
comparing it to other clustering schemes resulting by the same
algorithm but with different input parameter values. The two
first approaches are based on statistical tests and their major
drawback is their high computational cost. The third approach
aims at finding the best clustering scheme that a clustering
algorithm can define under certain assumptions and parameters
[11].
V. WORKS RELATED WITH DYNAMIC
CLUSTERING
The most perspective direction is based on the attempts to
model the work of human brain, which is highly complex,
nonlinear and parallel information processing system. Complex
cortex structure is modeled and formed by artificial neuron
lattices, which are joined by great amount of interlinks. This
global link of simple neurons provides their collective
behavior. Each neuron carries out the role of a processor.
That's why neuron network structure is the most appropriate
base for parallel computing. There is no need to prepare data
(in neural network input data is already parallelized). For
parallel computing to work correctly software should be able to
partition its work and data it operates on over hundreds of
processors. High speed and with the same time high quality
solution of most various complicated problems can be received
by means of microsystem's collective behavior property. The
main idea of self organization is in distributed character of data
processing, when one element dynamics means nothing, but at
the same time group dynamics define macroscopic unique state
of the whole system, that allows this system to reveal
capabilities for adaptation, learning, data mining and as one of
the results high computation effectiveness [5].Advances in
experimental brain science give evidence to the hypothesis that
cognition, memory, attention processes are the results of
cooperative chaotic dynamics of brain cortex elements
(neurons). Thus the design of artificial dynamic neural
networks on the base of neurobiological prototype seems to be
the right direction of the search for innovative clustering
techniques. Computer science development predetermined
promising possibilities of computer modeling. It became
possible to study complex nonlinear systems. Dynamics
exponential unpredictability of chaotic systems, their extreme
instability generates variety of system's possible states that can
help us to describe all the multiformity of our planet.
It is assumed to be very advantageous to obtain clustering
problem solution using effects produced by chaotic systems
interaction. This research tries to make next step in the
development of universal clustering technique. Dynamic data
mining combines modern data mining techniques with modern
time series analysis techniques.Saka and Nasraoui [2]
designing a simultaneous clustering and visualization
algorithm and the Dynamic FClust Algorithm which is based
on flocks of agents as a biological metaphor. This algorithm
falls within the swarm based clustering family, which is unique
compared to other approaches, because its model is an ongoing
swarm of agents that socially interact with each other and is
therefore inherently dynamic.Yucheng Kao and Szu-Yuan Lee
[12] presented a new dynamic data clustering algorithm based
on K-Means and combinatorial particle swarm optimization,
called KCPSO. Unlike the traditional K-Means method,
KCPSO does not need a specific number of clusters given
before performing the clustering process and is able to find the
optimal number of clusters during the clustering process. In
each iteration of KCPSO, a discrete PSO (Particle Swarm
Optimization) is used to optimize the number of clusters with
which the K-Means is used to find the best clustering result.
KCPSO has been developed into a software system and
evaluated by testing some datasets. Encouraging results show

ISSN: 2278-2419
73
that KCPSO is an effective algorithm for solving dynamic
clustering problems.
Elghazel Haytham et al [13] presented a dynamic version for
the b-coloring based clustering approach which relies only on
dissimilarity matrix and cluster dominating vertices in order to
cluster new data as they are added to the data collection or to
rearrange a partition when an existing data is removed. A real
advantage of this method is that it performs a dynamic
classification that correctly satisfies the b-coloring properties
and the clustering performances in terms of quality and
runtime, when the number of clusters is not pre-defined and
without any exception on the type of data. The results obtained
over three UCI data sets have illustrated the efficiency of the
algorithm to generate good results than Single-Pass and k-NN
(k-nearest neighbor) algorithms. There are many interesting
issues to pursue: (1) leading additional experiments on a larger
medical data set where a patient stay typology is required and
an inlet patient stay is regular and has to be incorporate to the
typology. (2) Extending the incremental concept to add or
remove simultaneously sets of instances and (3) to define some
operators which permit to combine easily different clustering’s
constructing on different data.Ester et al [14] presented an
incremental clustering algorithm based on the clustering
algorithm DBSCAN for mining in a data warehousing
environment which is applicable to any database containing
data from a metric space, e.g., to a spatial database or to a
WWW-log database. Due to the density based nature of
DBSCAN, the insertion or deletion of an object affects the
current clustering only in the neighborhood of that object. Thus
efficient algorithms could be given for incremental insertions
and deletions to an existing clustering. Based on the formal
definition of clusters, this incremental algorithm yields the
same result as DBSCAN. Incremental DBSCAN yields
significant speed-up factors over DBSCAN even for large
numbers of daily updates in a data warehouse. The authors
were assumed that the parameter values Eps and MinPts of
incremental DBSCAN did not change significantly when
inserting and deleting objects.
Elena and Sofya [5] described about centuries humans admire
animate nature and accessories applied by life creatures to
fulfill various functions. At first it was just formal resemblance
and mechanistic imitation, then along with sciences maturity
the focus shifted on inner construction of living systems.
However due to the complexity of a living system it is
reproduced partly. Separate subsystems embody limited set of
functions and principals. Just independently showed up
artificial neural networks (attempts to mimic neural system),
genetic algorithms (data transfer by means of inheritance),
artificial immune systems (partial reproduction of immune
system), evolutionary modeling (imitation of evolution
development principals). The idea of natural self-organization
within individuals is the basis for swarm and ant colony
technologies. It is important to note that nearly all mentioned
technologies deal with distributed parallel data processing
thanks to numerous simple processing units comprised into
self-organized networks that adapt to ever-changing
environment (input information).Of course there exist
substantial peculiarities in the types of local cooperation and
global behavior mechanisms predetermined by system's goal
(as it is well-known systems demonstrate not only
interconnectivity of elements but their ability to serve one
purpose). Evolution of society, new computer technologies
have in common idea of small worlds modeling. Communities
of various natures (interests clubs, computer clusters,
marketing networks, etc.) speak up for strong local linkage of
units and weak connectivity outward nearest neighbors (nodes
of the net).
Recent research on brain activities gives evidence for its
cluster organization. So generally the small world models
reflect both animate nature and abiocoen. Originally the notion
bio-inspired comprised problem solving approaches borrowed
from living systems but nowadays it is understood more
widely. Results in physics in the field of chaos theory and
nonlinear dynamics contribute greatly to bio-inspired
methodology as soon as nonlinear chaotic models find their
application in data mining - first and fore most bio-inspired
scientific area. It proposes to classify bio inspired methods on
different issues:
a. Structure and connection: neural networks (macro level)
and artificial immune systems (micro level);
b. Collective behavior: ant-based networks, swarm methods,
multi agent systems, small- world networks;
c. Evolution and selection: genetic algorithm, evolutionary
programming and evolutionary modeling, evolutionary
computations;
d. Linguistics: fuzzy logic.
To step forward with generalization one can note that nearly all
mentioned methods realize collective data processing through
adaptation to external environment. Exception is fuzzy logic
more relative to classical mathematics (interval logic reflects
the diversity of natural language descriptions) Though bio
inspired methods are applied to solve a wide set of problems it
focus on clustering problem as the most complex and resource
consuming. The division of input set of objects into subsets
(mainly non-overlapping) in most cases is interpreted as
optimization task with goal function determined by inter and
inner cluster distances. This approach obliges the user to give
them a priori information about priorities: what is of most
importance - compactness of clusters and their diversity in
feature space or inner cluster density and small number of
clusters. The formalization process of clustering problems in
terms of optimization procedures is one of the edge one in data
mining.
Recent modifications of bio inspired methods are developed as
heuristics. It desires to enlarge the abilities of intellectual
systems a separate knowledge domain within artificial
intelligence field revealed.Soft computing (SC) considers
various combinations of bio-inspired methods. As a result there
appeared such hybrid methods like: neural fuzzy methods,
genetic algorithms with elements of fuzzy logic (FL), hybrid
comprised by genetic algorithms (GA) and neural networks
(NN); fuzzy logic with genetic algorithm constituent, fuzzy
systems with neural network constituent, etc. One of the main
ideas of such combinations is to obtain flexible tool that allow
to solve complex problems and to compensate drawbacks of
one approach by means of cooperation with another.For
example, FL and NN combination provides learning abilities
and at the same time formalized knowledge can be represented
due to fuzzy logic element. Fuzzy logic is applied as soon as to

ISSN: 2278-2419
74
add some flexibility to a data mining technique. One of the
main drawbacks of all fuzzy systems are absence of learning
capabilities, absence of parallel distributing processing and
what is more critical the rely on expert's opinions when
membership functions are tuned. In advance to input
parameters sensitivity almost all methods suffer from
dimension curse and remain to be resource consuming. The
efficiency of these methods depends greatly on the parallel
processing hardware that simulate processing units: neurons of
neural networks, lymphocyte in artificial immune systems, ants
and swarms, agents in multi-agent systems, nodes in small-
world networks, chromosomes in genetic algorithms, genetic
programming and genetic modeling.
The benefit from synergetic effects considers not only
collective dynamics but also physical and chemical nature of
construction elements - nonlinear oscillators with chaotic
dynamics. As it is shown in numerous investigations on
nonlinear dynamics: the more is the problem complexity. The
more complex should be the system dynamics. All over the
world investigations on molecular level take place to get new
materials, to find new medicine, to solve pattern recognition
problem etc. Most of them consume knowledge from adjacent
disciplines: biology, chemistry, math, informatics, nonlinear
dynamics and synergetics.
VI. RELATED WORKS
Due to the continuous, unbounded, and high speed
characteristics of dynamic data, there is a huge amount of data
and there is not enough time to rescan the whole database or
perform a rescan as in traditional data mining algorithms
whenever an update occurs [4]. Ganti et al [15] examine
mining of data streams. A block evolution model is introduced
where a data set is updated periodically through insertions and
deletions. In this model the data set consists of conceptually
infinite sequence of data blocks D1, D2, ... that arrive at times
1, 2, ... where each block has a set of records. The authors
highlight two challenges in mining evolving blocks of data:
change detection and data mining model maintenance. In
change detection, the differences between two data blocks are
determined. Next, a data mining model should be maintained
under the insertions and deletions of blocks of the data
according to a specified data span and block selection
sequence.Crespoa and Weberb [3] presented a methodology for
dynamic data mining using fuzzy clustering that assigns static
objects to dynamic classes. Changes that they have studied are
movement, creation and
elimination of classes and any of their combination. Once a
data mining system is installed and is being used in daily
operations, the user has to be concerned with the system’s
future performance because the extracted knowledge is based
on past behavior of the analyzed objects If future performance
is very similar to past performance (e.g. if company customers
files do not change their files over time) using the initial data
mining system could be justified. If, however, performance
changes over time (e.g. if hospital patients do not change their
files over time), the continued use of the early system could
lead to an unsuitable results and (as an effect) to an
unacceptable decisions based on these results. Here dynamic
data mining could be extremely helpful in making the right
decision in the right time and affects the efficiency of the
decision.
There are three strategies if a user is to keep applying his/her
data mining system in a changing environment.
 The user can neglect changes in the environment and keep
on applying the initial system without any further updates.
It has the advantage of being “computationally cheap”
since no update to data mining system is performed. Also
it does not require changes in subsequent processes. Its
disadvantage is that current updates could not be detected.
 Every certain period of time, depending on the application,
a new system is developed using all the available data. The
advantage in this case is the user has always a system “up-
to-date” due to the use of current data. Disadvantages of
this strategy are the computational costs of creating a new
system every time from scratch.
 Based on the initial system and “new data” an update of
data is performed. This will be shown to be available
method in this dissertation.
In the area of data mining various methods have been
developed in order to find useful information patterns within
data. Among the most important methods are association rules,
clustering and decision trees methods. For each of the above
data mining methods, updating has different aspects and some
updating approaches have been proposed:Decision trees:
Various techniques for incremental learning and tree
restructuring.Neural networks: Updating is often used in the
sense of re-learning or improving the net’s performance by
learning with new examples presented to the
network.Clustering: Chung and Mcleod describes in more
detailed approaches for dynamic data mining using clustering
techniques.Association rules :Raghavan et al developed a
system for dynamic data mining for association rules. Chung
and Mcleod proposed mining framework that supports the
identification of useful patterns based on incremental data
clustering, they focused their attention on news stream mining,
they presented a sophisticated incremental hierarchical
document clustering algorithm using a neighbourhood search.
Reigrotzki et al (2001) presented the application of several
process control-related methods applied in the context of
monitoring and controlling data quality in financial databases.
They showed that the quality control process can be considered
as a classical control loop which can be measured via
application of quality tests which exploit data redundancy
defined by Meta information or extracted from data by
statistical models. Appropriate processing and visualization of
the tests results enable Human or automatic detection and
diagnosis of data quality problems. Moreover, the model-based
methods give an insight into business-related information
contained in the data. The methods have been applied to the
data quality monitoring of a real financial database at a
customer site, delivering business benefits, such as
improvements of the modeling quality, a reduction in the
number of the modeling cycles, and better data understanding.
These benefits in turn lead to financial savings.In many
situations, new information is more important than old
information, such as in publication database, stock
transactions, grocery markets, or web-log records.
Consequently, a frequent itemset in the dynamic database is
also important even if it is infrequent in the updated database.

ISSN: 2278-2419
75
Incremental clustering is the process of updating an existing set
of clusters incrementally rather than mining them from the
scratch on each database update. A brief overview of work
done on incremental clustering algorithms is given next.
COBWEB was proposed by Fisher [16]. It is an incremental
clustering algorithm that builds taxonomy of clusters without
having a pre-defined number of clusters. Gennary et al [17]
proposed CLASSIT which associates normal distributions with
cluster nodes. The main drawback of both COBWEB and
CLASSIT is that they results in highly unbalanced trees.
Charikar et al [18] introduced new deterministic and
randomized incremental clustering algorithms while trying to
minimize the maximum diameters of the clusters. The diameter
of a cluster is its maximum distance among its points and is
used in the restructuring process of the clusters. When a new
point arrives, it is either assigned to one of the current clusters
or it initializes its own cluster while two existing clusters are
combined into one. Ester et al [14] presented Incremental
DBSCAN suitable for mining in a data warehousing
environment. Incremental DBSCAN is based on the DBSCAN
algorithm which is a density based clustering algorithm. It uses
R* Tree as an index structure to perform region queries. Due to
its density based qualities, in Incremental DBSCAN the effects
of inserting and deleting objects are limited only to the
neighborhood of these objects. Incremental DBSCAN requires
only a distance function and is applicable to any data set from a
metric space. However, the proposed method does not address
the problem of changing point densities over time, which
would require adapting the input parameters for Incremental
DBSCAN over time. Another limitation of the algorithm is that
it adds or deletes one data point at a time. An incremental
clustering algorithm based on SWARM intelligence is given in
Chen and Meng[19].
VII. CONCLUSION
The ability to solve complex clustering problems in terms of
oscillations clustering language in future research can be
extended by dynamic inputs or at the beginning of the road, as
there are many aspects of data mining that have not been
tested. Up to date most of the data mining projects have been
dealing with verifying the actual data mining concepts. Since
this has now been established most researchers will move into
solving some of the problems and in this case, the research is
to concentrate on solving the problem of using data mining
dynamic databases. This chapter gives existing work done in
some papers related with dynamic clustering and incremental
data clustering.
REFERENCES
[1] Han, J. and Kamber, M. “Data Mining: Concepts and Techniques”,
Morgan Kaufmann Publishers, 2006.
[2] Saka, E. and Nasraoui, O. “On dynamic data clustering and visualization
using swarm intelligence”, In Data Engineering Workshops (ICDEW), 26th
International Conference on IEEE, pp. 337-340, 2010.
[3] Crespoa, F. and Weber, R. “A methodology for dynamic data mining
based on fuzzy clustering”, Fuzzy Sets and Systems, Elsevier,
Vol. 150, pp. 267-284, 2005.
[4] Hebah, H.O. Naseredd, “Stream Data Mining”, in International Journal of
Web Applications, Vol. 1, No. 4, pp. 183-190, 2009.
[5] Elena N. Benderskaya and Sofya V. Zhukova, “Dynamic Data Mining:
Synergy of Bio-Inspired Clustering Methods”, Knowledge-Oriented
Applications in Data Mining.
[6] Xu, R., Wunsh, D. II, “Survey of clustering algorithms”, IEEE Trans. on
Neural Networks, Vol. 15, pp. 645-678, 2005.
[7] Zhang, J., Tianrui Li, Da Ruan and Dun Liu, “Neighborhood Rough Sets
for Dynamic Data Mining”, International Journal of Intelligent Systems, Wiley
Periodicals, Inc., Vol. 27, pp. 317-342, 2012.
[8] Sanjay Chakraborty and Nagwani, N.K. “Analysis and Study of
Incremental DBSCAN Clustering Algorithm”, International Journal
of Enterprise Computing and Business Systems, Vol. 1, Issue 2,
pp. 2230-8849, 2011.
[9] Sanjay Chakraborty, Nagwani, N.K. and Lopamudra Dey, “Performance
Comparison of Incremental K-Means and Incremental DBSCAN Algorithms”,
International Journal of Computer Applications (0975 – 8887) , Vol. 27, No.
11, 2011.
[10] Svetlana Cherednichenko, Outlier Detection in Clustering, 2005.
[11] Halkidi, M., Batistakis, Y. and Vazirgiannis, M. “Cluster Validity
Methods: part I”, In Proceedings of the ACM SIGMOD International
Conference on Management of Data, Vol. 31, Issue 2, pp. 40-45, 2002.
[12] Yucheng Kao and Szu-Yuan Lee, “Combining K-Means and particle
swarm optimization for dynamic data clustering problems”, This paper appears
in: Intelligent Computing and Intelligent Systems, ICIS 2009, IEEE
International Conference on, Vol. 1, pp. 757-761, 2009.
[13] Elghazel Haytham, Hamamache Kheddouci, Véronique Deslandres and
Alain Dussauchoy, “A Partially Dynamic Clustering Algorithm for Data
Insertion and Removal”, Discovery Science ,Lecture Notes in Computer
Science, Vol. 4755, pp. 78-90, 2007.
[14] Ester, M. and Wittmann, R. “Incremental Generalization for Mining in a
Data Warehousing Environment”, Proc. 6th Int. Conf. on Extending Database
Technology, Valencia, Spain, 1998, in: Lecture Notes in Computer Science,
Springer, Vol. 1377, pp. 135-152, 1998.
[15] Ganti, V., Gehrke, J., Ramakrishnan, R. and Loh, W. “Mining data
streams under block evolution”, In ACM SIGKDD Explorations,
Vol. 3(2), pp. 1-10, 2002.
[16] Fisher, “Knowledge acquisition via incremental conceptual clustering”,
Machine Learning, Vol. 2, pp. 139-172, 1987.
[17] Gennary, J., Langley, P. and Fisher, D. “Model of Incremental Concept
Formation”, Artificial Intelligence Journal, Vol. 40, pp. 11-61, 1989.
[18] Charikar, M., Chekuri, C., Feder, T. and Motwani, R. “Incremental
clustering and dynamic information retrieval”, 29th Symposium on Theory of
Computing, pp. 626-635, 1997.
[19] Chen, Z. and Meng, Q.C. “An incremental clustering algorithm based on
SWARM intelligence theory”, Proc. of the 3rd Int. Conf. on Machine Learning
and Cybernetics, Shanghai, 26-29 August, 2004

Certain Investigation on Dynamic Clustering in Dynamic Datamining

More Related Content

Certain Investigation on Dynamic Clustering in Dynamic Datamining