Structured and Unstructured Information Extraction Using Text Mining and Natural Language Processing Techniques

International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 5 Issue: 11 32 – 43
_______________________________________________________________________________________________
32
IJRITCC | November 2017, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
Structured and Unstructured Information Extraction Using Text Mining and
Natural Language Processing Techniques
S. Nagarajan
Department of Computer Applications,
School of Information Technology,
Madurai Kamaraj University,
Madurai, Tamilnadu,India
nagasethu2000@yahoo.com
Dr. K. Perumal
Department of Computer Applications,
School of Information Technology,
Madurai Kamaraj University,
Madurai, Tamilnadu,India
perumalmkucs@gmail.com
Abstract—- Information on web is increasing at infinitum. Thus, web has become an unstructured global area where information even if
available, cannot be directly used for desired applications. One is often faced with an information overload and demands for some automated
help. Information extraction (IE) is the task of automatically extracting structured information from unstructured and/or semi-structured
machine-readable documents by means of Text Mining and Natural Language Processing (NLP) techniques. Extracted structured information
can be used for variety of enterprise or personal level task of varying complexity. The Information Extraction (IE) in also a set of knowledge in
order to answer to user consultations using natural language. The system is based on a Fuzzy Logic engine, which takes advantage of its
flexibility for managing sets of accumulated knowledge. These sets may be built in hierarchic levels by a tree structure. Information extraction is
structured data or knowledge from unstructured text by identifying references to named entities as well as stated relationships between such
entities. Data mining research assumes that the information to be ―mined‖ is already in the form of a relational database. IE can serve an
important technology for text mining. The knowledge discovered is expressed directly in the documents to be mined, then IE alone can serve as
an effective approach to text mining. However, if the documents contain concrete data in unstructured form rather than abstract knowledge, it
may be useful to first use IE to transform the unstructured data in the document corpus into a structured database, and then use traditional data
mining tools to identify abstract patterns in this extracted data. We propose a novel method for text mining with natural language processing
techniques to extract the information from data base with efficient way, where the extraction time and accuracy is measured and plotted with
simulation. Where the attributes of entities and relationship entities from structured and semi structured information .Results are compared with
conventional methods.
Keywords: Information Extraction (IE), Unstructured, semi structured, Data Mining, Natural Language Processing (NLP), Text mining ( TM)
__________________________________________________*****_________________________________________________
I. Introduction
The huge amount of documents on the web (or
specifically, the web pages) by searching through a search
engine or browsing through hyperlinks existed within web
pages. Users which have no specific target often choose
browsing web pages to achieve their final goal. However,
many users have difficulty of getting start from a page
which will eventually lead to their goals. Hence many portal
sites emerge to provide such starting points. These sites
often provide some sorts of navigating structure such as web
directories or web hierarchies. Users can achieve a thematic
navigation through such structures. However, these
structures are generally constructed by human experts by
hands, causing them lack of coverage and hard to maintain.
Data stored in most text databases are semi
structured data in that they are neither completely
unstructured nor completely structured. For example, a
document may contain a few structured fields, such as title,
authors, publication date, length, category, and, so on, but
also contain some largely unstructured text components,
such as abstract and contents. There have been a great deal
of studies on the modeling and implementation of semi
structured data in recent database research.[6] Information
Retrieval techniques, such as text indexing, have been
developed to handle unstructured documents. But,
traditional Information Retrieval techniques become
inadequate for the increasingly vast amounts of text data.
Typically, only a small fraction of the many available
documents will be relevant to a given individual or user.
Without knowing what could be in the documents, it is
difficult to formulate effective queries for analyzing and
extracting useful information from the data. Users need tools
to compare different documents, rank the importance and
relevance of the documents, or find patterns and trends
across multiple documents. Thus, Text Mining has become
an increasingly popular and essential theme in Data Mining.

_______________________________________________________________________________________________
33
_______________________________________________________________________________________
I. Table Show the comparison between structured and unstructured data
Parameter Unstructured Data Semi Structured Data
Technology Character and binary data Relational database tables
Transaction Management
No transaction management
and no concurrency
Matured transaction
management, various
concurrency techniques
Version Management Versioned as a whole
Versioning over tuples,
rows, tables etc.
Flexibility
Very flexible, absence of
schema
Schema dependent,
rigorous schema
Scalability Very scalable
Scaling DB Schema is
different
The above table shows that comparison between
unstructured and semi structured data with respect to various
parameters like technology, transaction management,
version management, flexibility and scalability. In
Technology basis, semi structure data having relational data
base tables shown in results. In flexibility basis, rigorous
schema possible than unstructured data.
 Natural Language Processing (NLP)
NLP is one of the oldest and most challenging
problems in the field of artificial intelligence. The role of
NLP in text mining is to deliver the system in the
information extraction phase as an input[15].
 Information Extraction (IE)
Information Extraction[13] is the task of
automatically extracting structured information from
unstructured and/or semi-structured machine-readable
documents. In most of the cases this activity includes
processing human language texts by means of natural
language processing (NLP)[14]. The recent activities in
multimedia document processing like automatic annotation
and mining information out of text images/audio/video
could be seen as information extraction and the best
practical and live example of IE is Google Search Engine.
Tasks performed by IE systems include:
Term analysis, which identifies the terms
appearing in a document. This is especially useful for
documents that contain many complex multi-word terms,
such as scientific research papers.
Named-entity recognition, which identifies the
names appearing in a document, such as names of people or
organizations. Some systems are also able to recognize dates
and expressions of time, quantities and associated units,
percentages, and so on.
Fact extraction, which identifies and extracts
complex facts from documents. Such facts could be
relationships between entities or events.
Information Extraction=Segmentation + Classification + Clustering + Association
The information extraction process follow is described in this example
For years, Microsoft Corporation CEO Bill Gates railed against the Economic philosophy of open-source
software with Orwellian fervor, denouncing its communal licensing as a "cancer" that stifled technological
innovation.
Today, Microsoft claims to "love" the open- source concept, by which software code is made public to
encourage improvement and development by outside programmers. Gates himself says Microsoft will gladly
disclose its crown jewels--the coveted code behind the Windows operating system--to select customers.
"We can be open source. We love the concept of shared source," said Bill Veghte, a Microsoft VP. "That's a
super-important shift for us in terms of code access.― Richard Stallman, founder of the Free Software
Foundation, countered saying...

_______________________________________________________________________________________________
34
_______________________________________________________________________________________
The named entity recognition is first part of information
extraction. It is called as segmentation.
Microsoft Corporation
CEO
Bill Gates
Microsoft
Gates
Microsoft
Bill Veghte
Microsoft
VP
Richard Stallman
Founder
Free Software Foundation
II.TABLE RESULT OF INFORMATION EXTRACTION ( example)
NAME TITLE ORGANIZATION
Bill Gates CEO Microsoft
Bill Veghte VP Microsoft
Richard
Stallman
Founder Free Software Foundation
 Text
 Mining
Text mining is the analysis of data contained in
natural language text[16]. The application of text mining
techniques to solve business problems is called text
analytics. Text mining can help an organization derive
potentially valuable business insights from text-based
content such as word documents, email and postings on
social media streams like Face book, Twitter and LinkedIn.
Mining unstructured data with natural language processing
(NLP), statistical modeling and machine learning techniques
can be challenging, however, because natural language text
is often inconsistent. It contains ambiguities caused by
inconsistent syntax and semantics, including slang, language
specific to vertical industries and age groups, double
entendres and sarcasm. The figure 1 is overview about the
information extraction based text mining structure.
Fig. 1 Overview of IE based Text Mining Framework
 Structured And Unstructured Text Mining Data
Data mining have focused on structured data, such
as relational, transactional, and data warehouse. However, a
substantial portion of the available information is stored in
text databases, which consist of large collections of
documents from various sources, such as news articles,
research papers, books, digital libraries, email messages, and
Web pages. Text data base are rapidly growing due to the
increasing amount of information available in electronic
form, such as electronic publications, various kinds of
electronic documents, e-mail, and the World Wide Web
[20]. Nowadays most of the information in government,
industry, business, and other institutions are stored
electronically, in the form of text databases. Data stored in
most text databases are semi structured data in that they are
neither completely unstructured nor completely structured
[3]. For example, a document may contain a few structured
fields, such as title, authors, publication date, and category,
and so on.
Moreover, information retrieval techniques, such as
text indexing methods, have been developed to handle
unstructured documents. Traditional information retrieval
techniques [21, 8] become inadequate for the increasingly
vast amounts of text data. Typically, only a small fraction of
the many available documents will be relevant to a given
individual user. Without knowing what could be in the
documents, it is difficult to formulate effective queries for
analyzing and extracting useful information from the data.
Users need tools to compare different documents, rank the
importance and relevance of the documents, or find patterns
and trends across multiple documents. Thus, text mining has
become an increasingly popular and essential theme in data
mining.

_______________________________________________________________________________________________
35
_______________________________________________________________________________________
Fig. 2 Example for Unstructured data
Unstructured data [Fig. 2] refers to the information
that no identical structure within this kind of data is offered.
It's described as data, which cannot be kept in rows and
columns in a relational database. Example for unstructured
data is documented that is archived in folder, video and
images. Structured Data [23] [Figure 2] refers predefined
schema. The schema is instance information that conforms
to this specification. The Example of structured data is a
relational database system. Figure 3 illustrates an Entity
Relationship diagram (ER diagram), its concrete on tables
within an RDBMS [relational database management
system].
Fig. 3 Example for Structured data
A
B C
D
E F GF
ABC
Corp
123
Industry
Way
Goods
Inc
Busines
s Circle
H H
3
I ThingyI Widget
4
What do I do
with it?
What is it? Document Management
System/ERP/ECM/Share
Point/Database
Email
Electronic
documents
Paper
documents
Fax
PDF
Office
documents

_______________________________________________________________________________________________
36
_______________________________________________________________________________________
 Related Work
The navigational structure in this work is a set of
automatically identified category themes. Approaches on
automatically generate the category themes are similar in
context with research on topic identification or theme
generation of text documents. Salton and Singhal [19]
generated a text relationship map among text excerpts and
recognized all groups of three mutually related text excerpts.
A merging process is applied iteratively to these groups to
finally obtain the theme (or a set of themes) of all text
excerpts. Clifton and Cooley [7] used traditional data mining
techniques to identify topic in a text corpus. They used a
hyper graph partitioning scheme to cluster frequent item
sets. The topic is represented as a set of named entities of
the corresponding cluster. Ponte and Croft [18] applied
dynamic programming techniques to segment text into
relatively small segments. These segments can then be used
for topic identification. Lin [12] used a knowledge based
concept counting paradigm to identify topic through the
Word Net hierarchy. Hearst and Plaunt [9] argued that the
advent of full-length documents should be accompanied by
the need for subtopic identification. They developed
techniques for detecting subtopics and performed
experiments using sequences of locally concentrated
discussions rather than full-length documents. All these
work, in some extent, may identify topics of documents that
can be used as category themes for text categorization.
However, they either rely on predefined category hierarchy
(e.g. [12]) or do not reveal the hierarchy at all.
The application of intelligent agents in two aspects:
the Knowledge Acquisition (KA) approach and the
Information Extraction (IE) approach. Additionally, we
produce a brief review of the existing studies related with
knowledge extraction from natural language text documents.
The techniques used in these studies involve knowledge
acquisition agents, fuzzy logic intelligent agents, knowledge
management, term weighting, databases knowledge
discovery and biological text mining. The proposed work is
mainly related to two areas of research: knowledge
extraction from natural language text documents and
knowledge modeling using intelligent agents.
Intelligent agent in KA and IE Zhiping, Tianwei,
& Yu, (2010) produced a formal model of agent-based
knowledge management in intelligent tutoring systems. It
consists of three agents working together to construct,
distribute, and maintain knowledge. The first one is a
knowledge acquisition agent which is responsible for the
construction of user model and domain knowledge base. The
second is a knowledge distribution agent which is
responsible for producing personalized teaching web pages
to students dynamically. Finally, a knowledge maintaining
agent which is responsible for the refinement of student
models and domain knowledge[19].
Ropero, Gomes, Carrasco, & Leon, (2012)
proposed a novel method for information extraction (IE) in a
set of knowledge in order to answer to user consultations
using natural language based on fuzzy logic engine. The sets
of accumulated knowledge may be built in hierarchic levels
by a tree structure. The aim of this system is to design and
implementation a fuzzy logic intelligent agent to manage
any set of knowledge where information is abundant,
ambiguous, or imprecise. This novel method was applied to
the case of university of Seville web portal which contain
vast amount of information, they also proposed a novel
method for Term Weighting (TW) based on fuzzy logic
instead of using traditional term weighting scheme (TF-
IDF)[11].
Knowledge extraction from text documents
Valencia-Garcia, Ruiz-Sanchez, Vivancos-Vicente,
Fernandez-Breis, & Martinez-Bejar, (2004) produced an
incremental approach for discovering medical knowledge
from texts. The system has been used to extract clinical
knowledge from texts concerning oncology. The authors
started from notion of there are huge amounts of medical
knowledge reside within text documents, so that the
automatic extraction of that knowledge would certainly be
beneficial for clinical activities. A user-centered approach
for the incremental extraction of knowledge from text which
is based on both knowledge technologies and natural
language processing techniques is presented in this work. In
same time, ontology is used to provide a formal, structured,
reusable and shared knowledge representation [22].
 Problem Definition
To extract the rules, the IE task takes the set of
tagged documents and produces a template representation
for every document. This can be easily converted into rule-
like form. For this purpose, a set of domain-independent
extraction patterns are written so that we could match them
against the input documents each extraction pattern
constructs an output representation that involves two levels
of linguistic knowledge: the rhetorical information
expressed in the abstract and the semantic information
contained in it, which we later convert into a predicate-like
form. The left-hand expression states the pattern to be
identified and the right hand side (following the colon)
states the corresponding semantic action to be produced. A
possible solution is to design a new IE methodology,
subsequently referred to as TEMsIE1, which is based on
various statistical and machine learning methods and
techniques, selected to meet the requirements of the
different IE phases. Although IE and data mining are similar
in their general processing steps, they achieve different but
complementary benefits. Consequently, augmenting IE with
data mining creates a synergy that exploits benefits of both
in order to improve the entire methodology significantly,
especially its quality and performance. Time period is one
most challenges in data extraction from data base. Here we
give solution of time response of mining system with
dimensional attributes.

_______________________________________________________________________________________________
37
_______________________________________________________________________________________
 Problem Solution
Data mining research assumes that the information
to be ―mined‖ is already in the form of a relational database.
IE can serve an important technology for text mining. The
knowledge discovered is expressed directly in the
documents to be mined, then IE alone can serve as an
effective approach to text mining. However, if the
documents contain concrete data in unstructured form rather
than abstract knowledge, it may be useful to first use IE to
transform the unstructured data in the document corpus into
a structured database, and then use traditional data mining
tools to identify abstract patterns in this extracted data. We
propose a text mining with natural language processing
technique e for extract the information from data base. We
give novel model for extract the information from data base
with dimensional attributes and retrieve with less timing for
data extraction. Also the function of an expert mining
technique is to convert unstructured human expertise to
structured knowledge in a knowledge base to deal with
easily. Intelligent agents or intelligent software agents use
artificial intelligence in the pursuit of our objectives. We
implemented Support Vector Machine (SVM) to check the
text grammar in expertise extraction process.
 Proposed Implementation
The development of an intelligent agent that is
capable to interview with a domain expert using question
and answer in natural language, extract relevant knowledge
from those answers, and convert these knowledge to a set of
antecedence-consequence rules. At same time this agent also
extracting a set of patterns or linguistic expressions and
stores them into a conceptual database which is used later.
The other intelligent agent is proposed in this work is to
present a technique for extracting knowledge from natural
language text documents. This agent tries to categorize input
text documents into relevant or non-relevant documents with
reference to a particular domain by calculating a threshold
value based on term frequency-inverse document frequency
(TF-IDF) for each input text document, Term Frequency
(TF) is the frequency of occurrence of a term in a document
and Inverse Document Frequency (IDF) varies inversely
with the number of documents to which the term is assigned.
TMIA tried to calculate the weight of term frequency for
each feature (term) in text documents. Typically, the
weights of term frequency for the phrase level are higher
than for the word level because each element of the phrase
level set consists of more than one word while the elements
of the word level set consist of only one word. Eq(1) is the
classical formula with some modification of TF-IDF used
for term weighting:
where count(ti,dj) refers to the frequency of term ti
in document dj, |corpus| refers to the number of documents
in the corpus, count-doc(ti,corpus) is the number of
documents in the corpus that contain the term ti, ri is the
number of words in term ti which matched words in text
documents. Eq(2) is the formula which calculates the
average of all term weights calculated in eq(1):
Where TF-IDF(ti,dj) the weight of term frequency
for the ith term, N is the number of features (terms), S is the
total number of words that compose N. Now the TMIA
should makea decision to accept or reject those text
documents by compare the Av-TF value which is calculated
in eq(2) with the threshold value as:
IF Av-TF >= Threshold value THEN accept OTHERWISE
reject (3)
To accept means that, the produced text documents
for processing are relevant to a particular problem diagnosis
domain. In other words, it contains some relevant
knowledge about that domain which can be extracted by
other components of the proposed TMIA. On the other hand,
to reject means that they are non-relevant to that domain and
TMIA will ask the user to enter a new text documents file.
 Hybrid Intelligent Mining Agent (HIMA)
Expert mining is a process of extracting useful
knowledge or meaningful patterns from human domain
experts directly without interference of knowledge
engineers, or it is a knowledge detection and resolution
process of human domain experts. So, expert mining
knowledge discovery in human brains which is looking for
patterns of expertise looks like text mining which searches
for patterns in text. Expert mining is a process of analyzing
human expertise to extract knowledge (facts and rules)
which is useful to solve particular problems in a specific
domain. Expertise is unstructured or semi-structured,
unorganized, and complicated to deal with; in other words,
the function of an expert mining technique is to convert
unstructured human expertise to structured knowledge in a
knowledge base to deal with easily. Intelligent agents or
intelligent software agents use artificial intelligence in the
pursuit of their objectives.
The most important design considerations of
intelligent agent systems is the agent interface as expert
friendly as possible and hide the complexity of other
components of the proposed agent. The intelligent agent is
determined by the nature of its agent interface, as it is the
part of the intelligent agent that interacts with the domain

_______________________________________________________________________________________________
38
_______________________________________________________________________________________
expert using questions and answers in natural language and
menu driven techniques to manage the interview between
the domain expert and the HIMA to achieve the main goal
of that agent. The fig.4 represents the intelligent agents
important in optimize knowledge management. The data
mining agents perform various functions of data mining. It
is increasingly significant to develop methods and
techniques. The knowledge discovery processes can be
assisted by agents in order to increase the quality of
knowledge and to processes the patterns. The agent plays
main role in making decision processes in knowledge and
text mining process.
Fig. 4 Hybrid Mining with Agent Interface System
 Knowledge Extraction
The knowledge extraction process involves trying to
extract knowledge from the interview with the domain
expert by taking the expert's answers and putting them in a
pre-formatted template which is prepared for that purpose.
This template consists of three parts as
1. Situation: it includes a main situation and a set of
symptoms in the case of a production rule description or
only a main situation in the case of a fact description.
2. Description: it includes all real descriptions for each
situation in the first component as a sentence.
3. Pattern: it consists of two types of patterns; main pattern
which represent the head of the rule (consequence) or fact,
and sub-patterns which represent the body of the rule
(antecedence). The difference between them is the first one
has arguments and the other doesn't.
The card-pyramid parsing described in the previous
section requires classifiers for each of the entity and relation
productions. We use a Support Vector Machine (SVM) [4]
classifier for each of the entity productions in the grammar.
An entity classifier gets as input a sentence and a candidate
entity indicated by the range of the indices of its words. It
outputs the probability that the candidate entity is of the
respective entity type. Probabilities for the SVM outputs are
computed using the method by [5]. We use all possible word
subsequences of the candidate entity words as implicit
features using a word-subsequence kernel [2]. In addition,
we use the following standard entity extraction features: the
Part of Speech (POS) tag sequence of the candidate entity
words, two words before and after the candidate entity and
their POS tags, whether any or all candidate entity words are
capitalized whether any or all words are found in a list of
entity names, whether any word has ―suffixment‖ or ―ing‖,
and finally the alphanumeric pattern of characters [1] of the
last candidate entity word obtained by replacing each
character by its character type (lowercase, uppercase or
numeric) and collapsing any consecutive repetition (for
example, the alphanumeric pattern for CoNLL2010 will be
AaA0). The full kernel is computed by adding the word-
subsequence kernel and the dot-product of all these features,
exploiting the convolution property of kernels. We also use
an SVM classifier for each of the relation productions in the
grammar which outputs the probability that the relation
holds between the two entities. Parsing requires specifying a
grammar for the card-pyramid. The productions in the
User
Domain Expert Interface
Agent
Discovery
Agent
Discovery
Agent
Local
Database
Local
Database

_______________________________________________________________________________________________
39
_______________________________________________________________________________________
grammar are of two types. For leaf nodes, the productions
are of the form entity Label, which stands for candidate
entity, is the only terminal. The figure 5 [5]Show the
knowledge extraction with agent intelligent systems
connectivity. It provides the capability of dynamically
incorporating knowledge. This knowledge has been
extracted with the use of DM techniques.
Fig. 5 Agent Intelligent System with Knowledge Extraction
Hybrid Intelligent Mining Agent (HIMA)
Expert mining is a process of extracting useful knowledge or
meaningful patterns from human domain experts directly
without interference of knowledge engineers, or it is a
knowledge detection and resolution process of human
domain experts. So, expert mining knowledge discovery in
human brains which is looking for patterns of expertise,
looks like text mining which searches for patterns in text.
Expert mining is a process of analyzing human expertise to
extract knowledge (facts and rules) which is useful to solve
particular problems in a specific domain. Expertise is
unstructured or semi-structured, unorganized, and
complicated to deal with; in other words, the function of an
expert mining technique is to convert unstructured human
expertise to structured knowledge in a knowledge base to
deal with easily. Intelligent agents or intelligent software
agents use artificial intelligence in the pursuit of their
objectives. They are capable of performing autonomous
action in environments to achieve their goals . Wooldridge
describes agents as computing entities which have four
features: reactivity, autonomy, interaction, and initiative.
Reactivity means that a system maintains continuous
interaction with its environment if any change occurs in that
environment. Autonomy is the main characteristic of agents.
In other words, agents can cooperate autonomously to
achieve predefined goal. Another aspect of agents is their
ability to interact with other agents or humans using agent-
communication language. Agents cooperating with each
other can contribute to achieve goals because some goals
can only be performed through cooperative work. Finally,
the initiative feature of agents means they generate and
attempt to perform goals by taking initiative instead of only
being based on external environment events. These features
of an intelligent agent clearly appear in the proposed HIMA.
The reactivity feature is visible when the HIMA can work in
different diagnosis domain; In other words, HIMA continues
to work even if human the domain experts change. HIMA is
capable a working independently to fulfill goals of the
proposed agent and it also can cooperate, interact and
contribute with other agents (TMIA) or humans (domain
experts) to achieve the main goals of the multi-intelligent
Data/Problem Model Knowledge Extraction
Knowledge
Model 1
Knowledge
Model 2
Knowledge
Model n
Agent 1
Agent 2
Agent m
Agent
Intelligent
Systems
Agent BAgent A
Agent EAgent DAgent C Agent F
………….
…
…
…

_______________________________________________________________________________________________
40
_______________________________________________________________________________________
agent (knowledge extraction). The HIMA also has the
initiative feature through producing some questions to the
expert domain and it can extract knowledge and linguistic
expressions from experts' answers to save them in the
knowledge base and a conceptual database respectively.
Now let us discuss the components of proposed HIMA is
explained above chapters.
III. Table Template for production using HIMA approach
Situation Pattern Description
Main situation Disease(Slipped Disc) Slipped Disc
Symptom1 Condition1 low back pain in the button of the back bone
Symptom2 Condition2 leg pain in left or right one
Main situation Treatment("slipped disc", "Take drugs like Tilcotil20mg,
if the situation continue do surgery operation").
Take drugs like Tilcotil20mg, if the situation
continue do surgery operation
__________________________
//*Pseudo Code for propose algorithm*//
disease_ name("slipped disc"):-
condition1,
condition2,
condition3, condition4.
conditions(condition1," low back pain in the button of the back bone").
conditions(condition2," leg pain in left or right one").
conditions(condition3," tingling, weakness, and foot senseless").
conditions(condition4," little legs narcotize or numbness").
treatment("slipped disc", "Take drugs like Tilcotil20mg, if the situation continue do surgery operation")
Results and Discussions
The assessment process of the proposed system is
approved out across a medical domain. To be more precise,
the referred domain is back pain diseases as in the above
case study. For evaluating the show of the proposed system
we can divide the process into 2 parts: HIMA evaluation
and TMIA (Text Mining Intelligent Agent) evaluation.
In experimental setting environment, results are
obtained from java environment with help of eclipse
software with DB access. Each of the conventional and
propose approaches was evaluate using cross-validation, a
widely-used evaluation methodology for machine learning
and text classification systems. A 95-fold cross validation
was adopted, in which the 1500 ID in the data set were
divided into 50 equal portions, with 40 documents each.
Testing was performed for 100 iterations, in each of which
58 portions of the data (980 documents) were used for
training and the remain portion (30documents) was used for
testing. The data were rotated during the process such that
each portion was used for testing in exactly one iteration.
In the conventional way, the knowledge engineer
interviews domain experts to elicit problem solving
information and formulates it in the information base. In the
proposed architecture the HIMA interacts directly with area
experts to extract their information and save it in the
information base. We displayed HIMA to five domain
expert in the health domain. One of them understood his
task (interaction with HIMA) after an introductory
depiction, whereas the others understood their tasks after
relating each step in detail. Therefore, we put all description
about how HIMA works in the help option of the main menu
of that system. All questions which were produced by
HIMA to the domain experts are the same questions which
were produced by a knowledge engineer to domain experts.
This means, the answers also are same in both cases. From
the answers, the knowledge engineer and the HIMA were
able to extract information (construction rules) and save it in
the information base. That led us to conclude the knowledge
base in both cases is similar (equivalent).
The following experiments show six times when
the process was carried out on the same text documents both
by Text Mining Intelligent Agent and knowledge engineer.
The aim of this comparison was to analyze whether the Text
mining intelligent agent is capable to produce and create a
correct and effective knowledge base to the users. Table 4
shows the results of the above process, the column title
show us the total number of pages of text document files,
the number of paragraphs for each text file,

_______________________________________________________________________________________________
41
_______________________________________________________________________________________
IV. .Table Comparison between TMIA with information engineer
paragraphs Rules didn’t
need
modification
pages in text
document
accepted
paragraphs
Rules need
modification
Rules
extracted by
knowledge
engineer
Total rules
extracted by
proposed
TMIA
7 0 2 5 5 5 5
6 1 1 5 4 4 5
8 2 3 7 4 6 7
The above table 4 results show that, the comparison of
various information with respect to both knowledge
engineer and propose approach (TMIA). In various
paragraph contents levels (7, 6 & 8), where rule need
modification in average paragraph like 4 to 6. The results
from knowledge engineer having less accuracy of rules than
TMIA propose approach.
V. Table Results comparison between existing vs proposed with various parameter like accuracy, precision with time extraction
in seconds
Conventional
method Vs
Proposed
method
Accuracy of
Data
extraction in
% (2D &
3D)
Precision
In %
Time of data
extraction in
seconds
Conventional
method Vs
Proposed
method
Accuracy of
Data
extraction in
% (2D &
3D)
Precision
In %
Lexicon 82.83 65.40 40 Lexicon 82.83 65.40
Knowledge
Extraction
method
79.28 81.03 37
Knowledge
Extraction
method
79.28 81.03
SVM-world
Extractor
84.57 84.57 35
SVM-world
Extractor
84.57 84.57
In above table (Table 5) conclude that the propose method
(HIMA and TMIA) obtained more accuracy and optimum
precision than conventional methods like both SVM and
knowledge extraction). In lexicon method having complex
paragraph obtaining in various text and paragraph
environment. Time of extraction is maximum in
conventional methods than proposed method.
In mining system, time of execution with data
extraction should complex for calculation. In our proposed
method (HIMA) includes knowledge extraction, text mining
extraction etc . With some intelligent agent system. In Time
computation is calculated with dimentional of the data sets
like 2D and 3D with some data set size. Here we evaluated
with some 2D and 3D attribute data types with data sets with
mining process, we evaluate the time extraction (Execution
time).
VI. Table Data Table for 0.5MB Data set 2D (0.5 MB data
set, 2-D)
Execution Time ( seconds)
Lexicon ( Existing) HIMA with TMIA
3 3
6 4
10 8
11 9
16 12
30 18

_______________________________________________________________________________________________
42
_______________________________________________________________________________________
Fig. 6 Performance graph of Time execution between
Lexicon with proposed approach (HIMA)
Data set 2D (0.5 MB data set, 2-D)
In Fig.6 shows that performance time execution between
existing (lexicons) with proposed method (HIMA with
TMIA). The performance shows between numbers of runs in
db with time extraction with respect to 2D data sets.
VII. Table Data Table for 0.5MB Data set 3D (0.5 MB data set, 3-
D)
4 3
8 5
11 7
13 8
18 12
25 17
4 3
8 5
11 7
13 8
18 12
25 17
Fig. 7 Performance graph of Time execution between Lexicon with proposed approach (HIMA)
Data set 3D (0.5 MB data set, 3-D)
In Figure7 shows that performance time execution
between existing (lexicons) with proposed method (HIMA
with TMIA). The performance shows between numbers of
runs in db with time extraction with respect to 3D data sets.
Conclusion
The method presented in this work offers new
approaches for acquire information from field experts and
text documents based on a Hybrid intelligent Multi agent
(HIMA) and using an automatic way for constructing a
knowledge base in a specific diagnosis domain. In this
research paper we produce expert mining as a new concept
to mean extracting useful knowledge or meaningful patterns
from human domain experience and we present a novel
method for text mining based on a medical database, causal
words, phrase structure trees, and predefined template to
extract production rules from text document files. The

_______________________________________________________________________________________________
43
_______________________________________________________________________________________
proposed architecture (HIMA with TMIA) can speed up
construction of a knowledge base by reducing the amount of
time (retrieval and execution t ime) that a domain expert
may take when trying to explain his experience to
knowledge engineer and reducing the time also for the
knowledge engineer when he reads text documents to
extract and formulate knowledge from these documents
References
[1] M. Bilenko, R. Mooney, W. Cohen, P. Ravikumar, and
S. Fienberg. Adaptive name matching in information
integration.IEEE Intelligent Systems, 18(5):16–23,2003.
[2] M. Bilenko and R. J. Mooney. Adaptive duplicate
detection using learnable string similarity measures. In
Proceedings of the Ninth ACM SIGKDD International
Conference on Knowledge Discovery and Data Mining
(KDD-2003), pages 39–48, Washington, DC, Aug.2003
[3] Bing Liu. Web Data Mining - Exploring Hyperlinks,
Contents and Usage Data. Springer, 2007
[4] R. Bunescu, R. Ge, R. J. Kate, E. M. Marcotte, R.
J.Mooney, A. K. Ramani, and Y. W. Wong. Comparative
experiments on learning information extractors for
proteins and their interactions.Artificial Intelligence in
Medicine (special issue on Summarization and Informa-
tion Extraction from Medical Documents), 33(2):139–
155, 2005.
[5] M. E. Califf and R. J. Mooney. Bottom-up relational
learning of pattern matching rules for information
extraction.Journal of Machine Learning Research,
4:177–210, 2003
[6] K.-T. Chen, C. Chen, and P.-H. Wang. Network aware
loadbalancing via parallel vm migration for data centers.
In Computer Communication and Networks (ICCCN),
2014 23rd International Conference on, pages 1–8. IEEE,
2014.
[7] C. Clifton and R. Cooley. Topcat: Data mining for topic
identification in a text corpus. In Proc. European Conf on
Principles of Data Mining and Knowledge
Discovery(PKDD 99), pages 174–183, 1999
[8] Das-Neves, F., Fox, E. A. and Yu, X. Connecting Topics
in Document Collections with Stepping Stones and
Pathways. CIKM‟05, ACM Press, 2005, pp. 91-98
[9] M. A. Hearst and C. Plaunt. Subtopic structuring for
fulllength document access. In Proc. 16th Int’l ACM
SIGIR Conf. Research and Development in Information
Retrieval, pages 59–68, 1993.
[10] R.Hemamalini, Dr.L.Josephine Mary ―An Analysis on
Multi-Agent Based Distributed Data Mining System‖,
International Journal of Scientific and Research
Publications, Volume 4, Issue 6, June 2014.
[11] T. Kohonen. Self-Organizing Maps. Springer-Verlag,
Berlin,1997.
[12] C. Y. Lin. Knowledge-based automatic topic
identification.In Proc. Meeting of the Association for
Computational Linguistics (ACL 95), pages 308–310,
1995.
[13] Lokesh Kumar, Parul Kalra Bhatia ―TEXT MINING:
CONCEPTS, PROCESS AND APPLICATIONS”,
Journal of Global Research in Computer Science, 4 (3),
March2013,36-39
[14] http://www.scism.lsbu.ac.uk/inmandw/ir/jaberwocky.htm
[15] http://www.freepatentsonline.com/
[16] http://searchbusinessanalytics.techtarget.com/definition/t
ext-mining
[17] Mohammed Abbas Kadhim, M. Afshar Alam , Harleen
Kaur A Multi-intelligent Agent Architecture for
Knowledge Extraction: Novel Approaches for
Automatic Production Rules Extraction, International
Journal of Multimedia and Ubiquitous Engineering
Vol.9, No.2 (2014)
[18] J. M. Ponte and W. B. Croft. Text segmentation by
topic.In Proc. European Conference on Digital Libraries
(ECDL 97), pages 113–125, 1997
[19] G. Salton and A. Singhal. Automatic text theme
generation and the analysis of text structure. Technical
Report TR 94-1438, Dept. Computer Science, Cornell
Univ., Ithaca, NY, 1994
[20] Santosh Kumar Paul, Madhup Agrawal, Shyam Rajput,
Sanjeev Kumar ―An Information Retrieval(IR)
Techniques for text Mining on web for Unstructured
data‖, International Journal of Advanced Research in
Computer Science and Software Engineering
4(2),February-2014, pp.67-70.

Structured and Unstructured Information Extraction Using Text Mining and Natural Language Processing Techniques

Recommended

More Related Content

What's hot (20)

Similar to Structured and Unstructured Information Extraction Using Text Mining and Natural Language Processing Techniques (20)

More from rahulmonikasharma (20)

Recently uploaded (20)

Structured and Unstructured Information Extraction Using Text Mining and Natural Language Processing Techniques