EVENT
AITCE Sponsored Three Hosted by:
Day Seminar on Department of Computer
Frontiers in Science and
Classification engineering
Algorithms for Data Aditya Institute of
Mining Technology &
Management
Dec 26-28, 2008 Tekkali – 532201 A.P.
1
Title of the Talk
Speaker:
Intelligent Data Dr. Mukesh Kumar
Mining using Rohil
Assistant Professor
LPA Win- Computer Science &
Prolog and Information Systems
Group
Associated Birla Institute of
Data Mining Technology and
Science
and Other Pilani – 333031
(Rajasthan)
Toolkits 2
Acknowledgements
1. Brian D Steel and Clive Spenser
Logic Programming Associates Limited
London (For providing LPA-WinProlog
and Associated Tools for the presentation)
2. Dr. N B Venkateswarlu, Professor,
AITAM, Tekkali (for his constant support
and encouragement)
3
Data Mining
• Data Mining (DM): the extraction
of interesting nontrivial, implicit,
previously unknown and
potentially useful information or
patterns from data in large
databases.
4
Data Mining: Confluence of Multiple
Disciplines
• DatabaseTechnology
• Probability and Statistics
• Machine Learning
• Information Science
• Visualization
• Multivariate Analysis
• Other Disciplines
5
Basic Primitives of DM technology
1. Data Preprocessing
2. DM Techniques:
– Cluster Analysis – Unsupervised Learning
– Classification – Supervised Learning
– Association Rules
6
What is not Data Mining
• (Deductive) query • Expert systems or
processing. small Machine
Learning/statistical
programs
7
Intelligent Data Mining .. 1
• Data Mining methods * These non-traditional
which make use of methods must be able
to summarize the high-
Integration of volume results into
“Knowledge Acquisition more manageable
from data” and chunks of knowledge.
“Knowledge acquisition * These methods include
from Experts” efficient and scalable
algorithms for extracting
• The knowledge “new” kinds of
acquisition from experts is knowledge focused on
bottleneck. the problem at hand.
8
Intelligent Data Mining .. 2
These methods • The use of fuzzy logic
to deal with imprecision
lead to families of and uncertainty in the
domain-specific data data mining results and
mining tools in case to improve the
we want to foster interpretability of data
mining results.
the adoption of data • Alternative techniques
mining in actual for the representation
real-world problems. and exploration of data
mining results.
9
Motivation
Data explosion problem
Automated data collection tools
and mature database technology Data warehousing
lead to tremendous amounts of and
data stored in databases, data data mining
warehouses and other
information repositories
Data warehousing and
online analytical processing
+ Extraction of interesting
Knowledge Starvation
= knowledge (rules,
regularities, patterns,
constraints) from
The need to see through and
data in large databases
interpret all this “useless” data.
10
How LPA Prolog fits into the
solution to Problem?
• To answer this we need to dicuss overview of:
• Prolog Programming Language
• Superior Features of LPA Win-Prolog
• Toolkits associated with LPA Win-Prolog
11
LPA-WinProlog (Introduction)
…1
• Prolog is a high-level • Win-Prolog is a well-
fifth generation established industrial-
‘declarative’ strength prolog-
programming compiler system.
language based on
Logic.
12
LPA-WinProlog (Introduction)
…2
• WIN-PROLOG is the leading Prolog
compiler system for Windows-based
PCs. Prolog is an established and
powerful AI language which provides
a high-level and productive
environment based on logical
inference.
13
LPA-WinProlog (Superior
Features) … 1
• Searching: Prolog has • Rule-based System:
a built-in search Built-in support for
engine with pattern rule-based
matching capabilities. programming.
• So, more easy to build
solutions for complex • So, easy to use as
problems such as knowledge
knowledge discovery, management tool.
configuration, routing
and scheduling etc.
14
LPA-WinProlog (Superior
Features) …2
• Dynamic Reasoning: • Adaptive programs:
Prolog has type-free Prolog is highly
variables and support dynamic and lets one
for metaprogramming. to add new rules and
• So, it allows one to facts on fly.
generate and execute • So, applications adapt
arbitrary functions at their behavior in
run-time e.g. a system response to changes in
that can learn how to the operating
do new tasks. environment.
15
LPA-WinProlog (Superior
Features) … 3
• Expressiveness • Incremental
Compilation
• Built-in Execution, Pattern
Matching and Search
• Rapid Prototyping
• Increased Modularity,
Portability and reusability • Automatic dynamic
memory allocation and
• Polymorphism: deallocation:
16
LPA-WinProlog (Superior
Features) … 4
• Performance: - Intelligent
– A Choice of compilers Components
– Efficient run-time - Distributed
systems Computing
– Platform compatibility
– Database integration
* A wide range of
– DLL Interface
additional toolkits
– DDE Interface
– OLE Support
17
LPA-WinProlog (Superior
Features) … 5
• Musical Instrument • Easy Customisation
Digital Interface • Powerful Metatools
(MIDI) • Extensive Cross-
• Built-in Winsock Platform
TCP/IP Support Compatibility
• Prolog in Full
Colour!
• A number of
• In-Depth Unicode
Toolkits
Support (subsequent
slides list these) 18
LPA-WinProlog (Toolkits) … 1
• Data Mining • DataMite support the
Toolkit discovery of rules and
(DataMite): is a patterns within
relational databases
collection of routines,
such as Access,
supplied in the form of
Oracle, SQL Server
an API.
etc.
19
LPA-WinProlog (Toolkits) … 2
• VisiRule: is a • Case Based
graphical tool for Reasoning (CBR)
developing and Toolkit: is a collection
delivering business of routines, supplied in the
form of an API, which
rules systems and
support the retrieval of
components simply by similar cases within
drawing the decision relational databases such
logic. as Access, Oracle, SQL
Server etc.
20
LPA-WinProlog (Toolkits) … 3
• Intelligence Server : • Prodata Database
allows to embed Interface: provides a
LPA-based intelligent tight coupling between
components in other the WIN-PROLOG
applications written environment and
using almost any various commercial
Windows database systems
programming based on ODBC.
language or visual
development system.
21
LPA-WinProlog (Toolkits) … 4
• Portable Dialog • Using Chimera, you
Manager: provides can write sophisticated
a platform- multi-agent
independent interface applications which can
between Prolog run on multiple
programs and the host machines across any
machine's graphics TCP/IP network, from
user interface (GUI) a small home setup to
dialog subsystem. the entire Internet.
22
LPA-WinProlog (Toolkits) … 5
• ProWeb Server : • It allows Web sites to
supports the use the powerful
development, testing reasoning capacity of
and deployment of Prolog completely in
intelligent, dynamic the background, with
server-based HTML, Java and other
applications on standard tools
intranets and the providing the user
Internet. interface.
23
LPA-WinProlog (Toolkits) … 6
• ScaffoldIT : enables • ScaffoldIT will
organizations to create analyse this master
personalised and document, determine
customised documents what information is
by means of a live required and ask the
web session. relevant questions by
means of an
interactive web
session.
24
LPA-WinProlog (Toolkits) … 7
• Flex expert system • Flex provides a
toolkit: is an comprehensive and
expressive and flexible versatile set of
rule-based facilities for both
development system programmers and non-
for building and programmers to
delivering scalable and construct reliable and
flexible expert systems maintainable
and business rules applications.
applications. 25
LPA-WinProlog (Toolkits) … 8
• Fuzzy Logic • Flint provides a
Toolkit (Flint): is a comprehensive and
powerful sub-system versatile set of
which augments the facilities for
decision-making programmers who
power of both Prolog wish to incorporate
and Flex. uncertainty within
their expert systems
and decision support
applications.
26
LPA-WinProlog (Toolkits) … 9
• Web-flex: for • Prolog++: is a full
availability of the flex object-oriented system
expert system toolkit integrated within a
on the World Wide Prolog framework.
Web.
27
Demonstration … 1
• A Graphical User • A GUI to view Mean
Interface (GUI) for Vector, Covariance
simulation of K- Matrix, Divergence
Means clustering matrix for a given
algorithm. multi-dimensional
data-set related to a
training site or
otherwise given as
input.
28
Demonstration … 2
• A GUI to display the • A GUI to generate a
results generated by near-optimal decision
LPA Data Mining tree from given
Toolkit (Datamite). examples which may
include both
qualitative and
quantitative data.
29
Demonstration … 3
• An illustration of
features and
functionalities relevant
to Intelligent Data
Mining of LPA Win-
Prolog and associated
data mining and other
toolkits.
30
Results and Future Research
Agenda
Results Future Research
• Use of LPA Win-
Demonstrated possible Prolog to combine
uses of LPA Win- Genetic Algorithms,
Prolog and associate Neural Networks, and
tools for Intelligent an advanced agent
Data Mining communication
framework to handle
dynamic and complex
problems. 31
Thanks
Thanking You All
32