Multi Intelligent Agent System Computer Science Essay

Published Date: 02 Nov 2017

Department of Computer science, Faculty of management and IT, Jamia Hamdard,

Hamdard Nagar, New Delhi,110062

Synopsis of

Multi-intelligent gent system to build Knowledge base

In Rule-based Expert System

SUPERVISED BY

Prof. M.Afshar Alam

…………………………………

Dr. Harleen Kaur

………………………………….

BY

MOHAMMED ABBAS KADHIM

……………………………..

Field of research:

Artificial Intelligence (A.I.)

Research Title:

Multi-intelligent agent system to build Knowledge Base in Rule-based Expert System

Introduction

Computer system has become a part of our everyday lives, Artificial Intelligence (AI) is sometimes refer to machine intelligence which is concerned with devising computer programs to make computer more intelligent, expert system is one of the most common application of artificial intelligence that consists of user interface, inference engine and knowledge base. In this research, we focus on construct of knowledge base by using multi-agent system. The process of acquiring knowledge from expert and building knowledge base is called knowledge engineering [19]. The knowledge acquisition is bottleneck in the construct of expert system, because the extraction of knowledge in problem domain is an important step in the building of knowledge base [2]. In the traditional method of knowledge acquisition, the knowledge engineer interviews with domain expert to extract knowledge then translates it to representation method in knowledge base of expert system.

In this proposal research we will find automatic way by using multi-agent system to translate expert's knowledge, knowledge in databases, and extracted knowledge from text documents to facts and rules in knowledge base. The term agent is used to represent two orthogonal concepts, the first is the agent's ability for autonomous execution, the second is the agent's ability to perform domain oriented reasoning [3] The autonomous agents are computational system that inhabit some complex dynamic environment, sense and act autonomously in this environment, and by doing so realize a set of goals or tasks for which they are designed[18] .

Intelligent Agent

Recently, a more intelligent and user-friendly interface system has been introduced, in the form of an agent that can act in the place of a human being. An agent is an information process I assumed that this was meant to be a general introductory statement program that can be applied to numerous fields [8]. The concept of an agent has become important in both artificial intelligence and mainstream of computer science [16]. Agent can be classified into two major categories: resident and mobile. Resident agents stay in the computer or system and perform their tasks there. Mobile agents move to other systems, performing tasks there[18]. Researchers from the artificial intelligence scholarly community had developed several major areas of research during the 1980's, one of which was expert systems, this body of research has evolved into intelligent systems during the past decade. Within the area of intelligent systems research, expert system, knowledge management, machine learning, neural networks, data mining and intelligent agents have further development, researchers have define intelligent agent as consisting of very little Artificial Intelligent and primarily computer science[9].

A rule-based, agent control architecture provides several desirable properties such an agent could have an embedded expert system with a human expert's knowledge encoding in to rules to aid in high level decision making and control [7], this definition of agent is a good starting point but can be expanded in certain areas. First, an agent executes; it acquires input and produces output .Agent processing is domain-oriented: An agent "knows" about certain concepts, data structures, rules, and interfaces but is not necessarily capable of interpreting information outside its field. An agent also "reasons" by encapsulating rules that allow it to transform conditions into decision. And it operates autonomously by virtue of being persistent and capable of operating in a changing environment [3]. A multi-agent system is a system composed of multiple interacting intelligent agents. A multi-agent system is a loosely coupled network of problem-solver entities that work together to find answers to problems that are beyond the individual capabilities or knowledge of each entity [11] In our research will introduce software agent technology and concentrate specifically on ways that agent add value to knowledge base related processing such as knowledge discovery and other forms of data to knowledge base.

Rule-based expert system

An expert system is a computer program that simulates the judgment and behavior of a human or an organization that has expert knowledge and experience in a particular field. Typically, such a system contains a knowledge base containing accumulated experience and a set of rules for applying the knowledge base to each particular situation that is described to the program[10]. After the stage of problem domain selection the Knowledge engineers interact with human experts or collect documented knowledge from other sources (books, scientific research or databases) in the knowledge acquisition stage.

Knowledge acquisition is one of the main problem in development knowledge base system. Indicative learning technique tries to get the knowledge of a system from a set of examples[15]. The acquired knowledge is then coded into a representation scheme to create a knowledge base. The knowledge engineer can collaborate with human experts or use test cases to verify and validate the knowledge base. The validated knowledge can be used in a knowledge-based system to solve new problems via machine inference and to explain the generated recommendation [2,19].

The primary people involved in building expert system are the knowledge engineer, the domain expert and the end user ,the main task of knowledge engineer select the software and hardware tools for the project, help the domain expert to focus on necessary knowledge and implement that knowledge in correct and efficient knowledge base [10]. Figure (1) illustrates a simplified model of the knowledge acquisition process that will serve as a useful for understanding the problem involved in acquiring and formalizing human expert performance, knowledge engineer interviews with domain expert to extract expert knowledge which is then translated to facts and rules. We will search more automatic way to construct knowledge base. There are many programs that interact with domain experts to extract expert's knowledge these programs provide support for the following activities:

Entering knowledge

Maintaining knowledge base consistency

Ensuring knowledge base completeness

The most useful knowledge acquisition programs are that restricted to particular problem solving domain such as diagnosis, design, prediction… etc.[10].

If p(x)Λq(x,y)

then t(y)

if u(x)Λv(y)

then s(x,y)

Expertise Implemented system

Figure(1) a simplified model of the knowledge acquisition process

Objective of research

The complex stage in building expert system is knowledge acquisition from domain experts and other resources of knowledge (books, scientific researches, and databases) and translating it to representation method in knowledge base of expert system.

In this research, automatic way is used to build knowledge base in rule-based expert system ,therefore ; this research aim to construct multi-agent system which is able to extract the problem solving knowledge from human experts, text documents, and databases then translate it to facts and rules in knowledge base of rule-based expert system, that means there are three intelligent agents, the first one is interact with domain experts to extract knowledge directly and put it in knowledge base this agent called Experts Intelligent Agent (EIA), the second one is extract knowledge from text documents (books, scientific researches, E-mail …etc.) this agent called Text Mining Intelligent Agent (TMIA), and third agent is extract knowledge from databases this process called Knowledge Discovery in Databases (KDD), and data mining is particular step in KDD process this agent called Knowledge Discovery in Databases Intelligent Agent (KDDIA) . .

Proposed language for implementation

1-Prolog programming language

The importance of proposed research

In this research we have discussed how we can build multi-agent system to construct knowledge base in rule-based expert system, the importance of this research can be summarized as the following :

1-The proposed multi-agent system can quick up construct of rule-based expert system, that means instead of using traditional methods to extract knowledge we can use multi-agent system for that job.

2-We can translate the conceptual expertise for experts and knowledge in text documents and databases to facts and rules in knowledge base using proposed multi-agent system.

3-The proposed intelligent agent which extract knowledge from text documents (TMIA) may be lead to more study of Natural Language Process (NLP) and its applications

4-In this research we produce a general model for construct multi-agent system in diagnosis domain.

5-We can use one application of artificial intelligent (multi-agent system) to construct another application (rule-based expert system).

Literature Review

In 2002, Hudson and Cohen produce Intelligent Agents in the Diagnosis of Cardiac Disorders which discuss Diagnosis of cardiac disorders requires the combination of many different types of data, including family and patient histories, laboratory results, physical findings, genetic information, electrocardiogram analysis, and imaging results. Intelligent agents, an approach that has been used chiefly in business applications, provides a structure that can combine not only data types but also a variety of reasoning methodologies in the same decision support system. The user is included as an agent in the system and can interact directly with any of the components[4].

In 2007, Insung Jung and others Intelligent Agent Based Graphic User Interface (GUI) for e-Physician In this paper, they present an intelligent agent based user interface for physician. Although a lot of user interface system has been built for the healthcare system but the most sophisticated and powerful system will be next to useless without an effective intelligent agent. This framework can be applied to develop an intelligent agent based user interface, especially GUI (Graphical user interface), self-governing, self customized with wise decision making competency[6].

In 2008, Tamanna Siddiqui produce paper A KKD for automatic Discovery of knowledge which provides a survey of the available literature on knowledge discovery and data mining. Different type of knowledge representation, KDD approaches and KDD tools are highlighted. Data mining is highly interdisciplinary area; machine learning is one of them. Inductive Learning Algorithm is demonstrated here as intelligent data mining tool[14]

In 2008, Priti Srinivas Sajja produce Multi-Agent System for Knowledge-Based Access to Distributed Databases which include a framework for knowledge discovery, knowledge use, and knowledge management is presented in this article to provide knowledge-based access of the domain databases using multi-agent systems approach. This framework encompasses five different agents: namely, knowledge management agent, data filter agent, rule induction agent, dynamic analysis agent, and interface agent[11].

In 2010, Mehnroush Shamsfard produce Lexico-syntactic and Semantic Patterns for Extracting Knowledge from Persian Texts in This paper he introduces some lexico-syntactic and semantic patterns and templates for extracting conceptual knowledge from Persian texts. The described patterns are general and domain application in dependent and work at sentence level. They are used to extract taxonomic and non-taxonomic relations and axioms from phrases and sentences[13].

In 2011, Patel Sanskruti and others produce a knowledge representation in distributed environment for health care applications using multi-agent system in this paper they introduce, The World Wide Web is distributed by nature (inherently) and hence knowledge representation in distributed environment has become core aspect in efficient knowledge management process to be implemented in distributed environment. In a distributed environment such as Web, it is very much essential that information created/captured at different places and stored on different hardware and software architecture are inter operable; so that different information system[12]

Research Methodology

I-Proposed multi-agent system

The multi-agent system which is used to knowledge acquisition more efficient when we are restricted to a particular problem solving domain. In proposed multi-agent system we are restricted to a diagnosis domain using production system to represent knowledge in knowledge base, that mean the output of proposed multi-agent is a knowledge base in rule-base expert system in diagnosis domain. Figure (2) below illustrated the input and output of proposed multi-agent system which is consist of three intelligent agents :

Proposed

Multi-agent system

(EIA, TMIA, KDDIA)

Knowledge base in rule-based expert system

Figure(2) Input and output of proposed multi-agent system

Experts

Text documents

Databases

1-Experts Intelligent Agent (EIA)

Software agents are computer programs different from non-agents programs in their ability to run autonomously sensing and acting on changing environment condition because they run autonomously [16]. The following paragraphs show the main components of (EIA) :

A-Interface

The proposed agent can be interacted with expert by interface, the interface of proposed agent must be friendly of expert and more comfortable for the human and must be hide the complexity of other components of agent. The interface built by using question and answer in natural language to become closest to expert. Figure (3) illustrate main components of (EIA). The interface also permits for expert to be update of knowledge in knowledge base. That means the ability of interact combination between adding knowledge and previous contains of knowledge base. The output of interface is knowledge description from the input conceptual description.

B-Extraction of Knowledge

After knowledge description, the second phase has been started which involves extracting facts and rules from knowledge description. The fact is a relationship between two objects (or more) without condition, while the rule consist of two elements, the first one (action or head of rule) is the relationship between two objects (or more), the other one is the conditions (body of the rule) which is the set of relations that must be satisfied to carry out of head of that rule (action). In diagnosis domain the head of rule represent the complex situation and the body of rule represent the observed symptoms. After extraction , knowledge (facts and rules) will be written in knowledge base.

C-Knowledge base

Before knowledge (facts and rules) have been written in knowledge base the (EIA) must be check that facts and rules consistency with other contains of knowledge base.

interface

Extraction of knowledge

Knowledge base

knowledge

description

facts

rules

Check consistency

Knowledge base in rule-based expert system

Experts

conceptual environment description

Figure (3) Modal of EIA

2- Text Mining Intelligent Agent (TMIA)

The term "text mining" was coined since it is a natural extension of data mining and is the extraction (or mining) of patterns, useful information or knowledge from natural language text. The process of text mining is not a new development as it is used in statistical natural language processing and information extraction. text data mining tasks can be classified as (i) question answering (information retrieval), (ii) information extraction, or (iii) thesaurus generation. Text mining can be used to discover prevalent concepts in a collection of documents, to summarize documents, or to classify documents into categories[5].The (TMIA) function is extract knowledge from text documents, figure (4) and following steps illustrate the main components of (TMIA):

A-Text documents test

In this step we will test the text document, Is it concern with particular domain which we want to build knowledge base or not ?. This function can be accomplishes by produce sentences segmentation process to tokens that means segmentation text document to set of words, after that compare between all vocabularies which used in interview between experts and EIA with set of tokens which represent text document, if the comparison result is more than 10% then text document is accepted otherwise text document is rejected.

B-Text analysis

In this step we will perform text analysis for each sentence in text document as follow as, When the proposed multi-agent works in diagnosis domain; the diagnosis is one of general expert system problem categories which means determining the cause of malfunction in complex situation based on observable symptoms, that means we will search about main situation which needs to diagnosis and a set of observable symptoms which satisfy this situation in each sentence in that text, for example in medical diagnosis the disease name represent the main situation and the symptoms of this disease represent the observable symptoms, show the following example for eyes disease:

The disease is Primary Glaucoma if Obligation of the felnation angle and No pain except ache and Abnormal facility of aqueous outflow and May be loss vision of one eye, The treatment of Primary Glaucoma are Anticholneslenose drugs and Epinephnes drugs and Pilocarpein or Surgical operation

From this paragraph can summarize the table (1), This table summarization depend on set of keywords such as (disease name, if, and, treatment) which always used in medical diagnosis domain, if we work in different diagnosis domain may be used different keywords.

Table (1) summary of above example

situation

Description

pattern

Main situation

Primary Glaucoma

Disease(primary Glaucoma)

Symptom1

Obligation of the felnation angle

Condition1

Symptom2

No pain except ache

Condition2

Symptom3

Abnormal facility of aqueous outflow

Condition3

Symptoms4

May be loss vision of one eye

Condition4

Main situation

Treatment of primary Glaucoma are Anticholneslenose drugs and Epinephnes drugs and Pilocarpein or Surgical operation

Treatment(primary glaucoma, Anticholneslenose drugs and Epinephnes drugs and Pilocarpein or Surgical operation)

C-Semantic patterns

In this step we depend on the description in second column to construct patterns in third column, these patterns may have one argument such as disease pattern and may don’t have arguments such as condition1,condition2,…,or may have two arguments such as treatment argument, these patterns are very important in next step.

D-Relationships generation

In this step we will construct relationship between main situation and symptoms using semantic patterns which extracted from step (C) for above example we can construct the following relations:

Disease(primary glaucoma):-condition1,condition2,condition3,conditon4. ….(1)

Treatment(primary glaucoma, "Anticholneslenose drugs and Epinephnes drugs and Pilocarpein or Surgical operation"). ….(2)

relations (1,2) represent final form which are ready to written in knowledge base, where relation (1) represent rule and relation (2) represent fact in knowledge base of rule based expert system. Before knowledge (facts and rules) have been written in knowledge base the (TMIA) must be check that facts and rules consistency with other contains of knowledge base.

rejected

Document test

Relationships generation

Knowledge base

facts

rules

Check consistency

Knowledge base in rule-based expert system

Text document

Text analysis

Figure (4) Modal of TMIA

accepted

patterns

Semantic patterns

description

3- Knowledge Discovery in Databases Intelligent Agent (KDDIA)

KDD refers to the overall process and discipline of extracting useful knowledge from databases and includes data warehousing, data cleansing and data manipulation tasks right through to the interpretation and exploitation of results. Data Mining now refers to one stage within the KDD process for extracting useful rules and patterns from the data [17]. A major issue in Data mining is the form in which the knowledge is to be represented. This has significant repercussions on the efficiency of search, intelligibility and power of the system. In most cases, the knowledge is meant to be used for human use - in planning, etc. Hence a symbolic representation is preferred[14].

There are many approach for KDD, in proposed (KDDIA) we will use automated tools which can be designed for learning rules from databases, the Inductive Learning Algorithm (ILA) is a suitable automatic tools for proposed (KDDIA), the following steps illustrate Rules3 Inductive Learning Algorithm[1]

Step 1. Define ranges for the attributes which have numerical values and assign labels to those ranges

Step 2. Set the minimum number of conditions (Ncmin) for each rule

Step 3. Take an unclassified example

Step 4. Nc = Ncmin -1

Step 5. If Nc < Na then Nc = Nc+1

Step 6. Take all values or labels contained in the example

Step 7. Form objects which are combinations of Nc values or labels taken from the values or labels obtained in Step 6

Step 8. If at least one of the objects belongs to a unique class then form rules with those objects; ELSE go to Step 5

Step 9. Select the rule which classifies the highest number of examples

Step 10. Remove examples classified by the selected rule

Step 11. If there are no more unclassified examples then STOP; ELSE go to Step 3

where (Nc=number of conditions, Na=number of attributes).

II-Embedded Knowledge base

Now we capture complete knowledge base which used production system to represent that knowledge, then knowledge engineer take that knowledge base and complete rule-based expert system by adding inference engine and user interface.

When the construct of knowledge base is the bottleneck of construct of rule-based expert system, therefore; the proposed agent focuses on construct of knowledge base. In other word, the main task of proposed multi-agent is to build the knowledge base in rule-based expert system. The other components of rule-based expert system such as user interface and inference engine are the task of knowledge engineer. Knowledge engineer will proposed search strategy and a way of user interface built to capture a complete rule-based expert system which can be provided an advice in diagnosis domain. Figure (5) illustrated the place of proposed agent in the components of rule-based expert system :

User Interface

Inference Engine

Knowledge base

Proposed

Multi-Agent

EIA, TMIA, KDDIA

Design and implementation

Expert, text documents,and database

Knowledge Engineer

facts

rules

Rule-based expert system

Figure (5) Proposed multi-agent and rule-based expert system