An Intelligent Tutoring System for Multimedia Virtual

Paper ID #15525

An Intelligent Tutoring System for Multimedia Virtual Power Laboratory

Mr. Ning Gong, Temple University

Ning Gong is currently a third year PhD student in Electrical and Computer Engineering at Temple Uni-versity. His research is focused on Smart Grid and Consensus Control Theories. He is particularly inter-ested in power distribution system topology and resilience control applications. Before coming to TempleUniversity, he graduated in Polytechnic Institute of New York University with his M.S degree. Currentlyhe is a Graduate Research Assistant in the department. He can be contacted at: [email protected].

Dr. Saroj K Biswas, Temple University

Saroj Biswas is a Professor of Electrical and Computer Engineering at Temple University specializingin electrical machines and power systems, multimedia tutoring, and control and optimization of dynamicsystems. He has been the principle investigator of a project for the development of an intelligent tutoringshell that allows instructors create their own web-based tutoring system. His current research focuses onsecurity of cyber-physical systems based on multiagent framework with applications to the power grid,and the integration of an intelligent virtual laboratory environment in curriculum. He is an associate editorof Dynamics of Continuous, Discrete and Impulsive Systems: Series B, and is a member of IEEE, ASEE,and Sigma Xi.

Dr. Li Bai, Temple University

Dr. Li Bai is a Professor in the ECE department, Temple University. He received his B.S. (1996) fromTemple University, M.S. (1998) and Ph.D. (2001) from Drexel University, all in Electrical Engineering.He was a summer research faculty in AFRL, Rome, NY, during 2002–2004 and the Naval Surface WarfareCenter, Carderock Division (NSWCCD), Philadelphia, PA, during 2006–2007. His research interestsinclude video tracking, level 2+ information fusion, array signal processing and multi-agent systems,wireless sensor network and dependable secure computing. His research has been supported by Office ofNaval Research, Department of Transportation, U.S. Department of Commerce’s Economic DevelopmentAdministration (EDA), National Science Foundation, U.S. Army and Exxon Mobil, etc. Also, Dr. Baiserved as the Chair of the IEEE Philadelphia Section in 2007 and was Young Engineer of the Year inDelaware Valley, IEEE Philadelphia Section in 2004.

Dr. Brian P. Butz, Temple University

Dr. Brian P. Butz is a Professor Emeritus of Electrical and Computer Engineering at Temple University,Philadelphia, PA. In 1987, Professor Butz founded the Intelligent Systems Application Center (ISAC)which provided a focal point within Temple University for research in intelligent systems. ProfessorButz’s research efforts focused on expert/knowledge-based systems and intelligent tutoring systems. Hehas been the Principal Investigator for several projects that immerse users into a particular virtual envi-ronment in which they are able to learn both theory and application within a specific subject area. From1989 through 1996, Professor Butz was the Chair of the Electrical and Computer Engineering Depart-ment at Temple University. He has written many papers on intelligent systems and has received severalteaching awards including the Christian R. and Mary F. Lindback Award for Distinguished Teaching andthe Temple University Great Teacher Award. He is a m a Life Senior Member of the IEEE.

c©American Society for Engineering Education, 2016

INTELLIGENT TUTORING SYSTEM FOR MULTIMEDIA VIRTUAL

LABORATORY IN POWER ENGINEERING

Abstract

A laboratory practicum is considered a key element in traditional undergraduate education in

power engineering, however often ignored for various reasons, such as expenses for establishing

a physical laboratory, safety of students working with high voltage sources, and lack of qualified

teaching assistants. This research is a continuation of an NSF funded project on the development

of a Virtual Power Laboratory1 (VPL) environment that can simulate an Electric Machines

Laboratory. Development of the VPL architecture and its modules has been reported in previous

years. This paper describes the final stage of the VPL architecture which is instructional design

and implementation of an Intelligent Tutoring System. The virtual laboratory is supervised by a

virtual Intelligent Tutor that can track students’ progress and monitor their actions in the virtual

laboratory platform. The VPL offers a virtual experimental environment with 2D graphics, 3D

animations, audio guidance, simulations, knowledge concept bases, and virtual experiments; the

Intelligent Tutoring System is designed on top of these functionalities. It processes the user

actions and the resources existing in the virtual laboratory, and tracks students’ progress, answers

questions, monitors their actions, and if necessary, guides the students with prerequisite material

on the subject matter.

Key Words: Virtual Power Laboratory, Intelligent Tutor, Electric machines, Web application.

1. Introduction

The “Task Force on America’s Future Energy Jobs1” reports that there will be “a critical

shortage of trained professionals to maintain the existing electric power system and design, build,

and operate the future electric power system”, and “new workers will be needed to fill as many

as one-third of the nation’s 400,000 current electric power jobs2”. In order to deal with the

shortage, both the Task Force1 and the National Science Foundation

3 suggest a major revision in

engineering curricula regarding power, and recommend a significant investment in education,

research, and hiring of faculties in the power area. A laboratory practicum is an essential element

of a student’s learning process, however for courses on electric power and machines, it is often

difficult to give students a hands-on learning opportunity due to safety issues, expense, and lack

of qualified teaching assistants. It has been reported that out of 118 schools and colleges that

participated in a recent survey1, 26 schools that offered power-related courses had no laboratory

support.

With the development of Machine Learning technology and Artificial Intelligence, computer

programs for virtual lab environment have become more adaptive and friendly to users.

Graphical User Interface based virtual laboratories have made it possible for the user to learn the

concept more naturally. Recent advances in computer technology and availability have allowed

the computer industry to develop hardware and software applications that address the needs of

the physically disabled. According to the Center of Disease Control, approximately 109,700

persons with motor disabilities are employed in the United States4,5

. After the US Congress

1 This research was supported in part by the National Science Foundation under grant DUE-1140306

amended the Rehabilitation Act Section 508 Compliance in 1998 to require educational and

federal agencies to make their electronic and information technology accessible to people with

disabilities, technologies to ensure the accessibility of computer software have developed rapidly.

The VPL is an important development for students with physical handicaps as it allows access to

a “lab” that might otherwise be unavailable. Recent advances of mobile devices with touch

screen and fast wireless capabilities have made it possible to develop intelligent multimedia

virtual tutoring systems that can be accessed by people with motor disabilities

This paper presents the implementation of an Intelligent Tutoring System on Virtual Power

Laboratory (VPL)6,7

(http://vpl.temple.edu) that was developed in previous years. The virtual

laboratory is supervised by a virtual Intelligent Tutor that can answer students’ questions, track

their progress and monitor actions as they perform experiments, and if necessary, guides the

student with correct actions. The Intelligent Tutor processes the laboratory material on electric

machines, including knowledge concepts, experiment instructions and videos, as it responds to

students’ questions. The Intelligent Tutoring System is based on a Keyword Matching

Likelihood (KML) estimate algorithm that maps the questions asked by the users to the desired

answers from the database so that the question can be answered in real-time. The Intelligent

Tutoring System also uses Natural Language Processing functionality to decipher students’

queries that have lower KML estimate scores. Questions that cannot be answered by the

Intelligent Tutoring System are forwarded to the instructor for providing off-line response to the

user. The Intelligent Tutoring System has a knowledgebase that can be extended and updated by

the instructor as required. The VPL system is universally accessible as it is developed for most

Operating Systems and current mobile devices.

The rest of the paper is organized as follows: Section 2 is an overview of virtual tutoring

systems available in the literature. Section 3 summarizes the authors’ previous work6,7

on the

development of VPL. The architecture of the Intelligent Tutoring System is presented in Section

4. The paper is concluded in Section 5 with discussions and plans for the future.

2. Virtual Tutoring Environment

Computer-aided instruction (CAI) systems were introduced as early as 1960's as a means of

assisting students outside the classroom8. The first CAI programs were either computerized

versions of textbooks or drill and practice monitors9 that presented a student with problems and

compared the student’s responses to the pre-scored answers. If necessary, these CAI programs

provided the student with canned remedial responses. Improvements were continuously made

until computer-aided instruction systems evolved into intelligent tutoring systems when artificial

intelligence techniques were used to embed explicit knowledge of the subject matter. An

Intelligent Tutoring System attempts to simulate the behavior of an intelligent human tutor in

addition to acting as a domain expert. The characteristics of an Intelligent Tutoring System

include the ability to teach a given subject, and to detect student’s errors, and to assist the student

correct her mistakes10,11

. Advancements in software over the last decade along with interactive

multimedia development tools have created software environments that make real Intelligent

Tutoring Systems a possibility12,13,14

. Recently15,16,17

promoted intelligent agents and systems

with intelligence have been proposed for a multitude of tasks18-20

. Intelligent agents are usually

envisioned as assisting a computer user by scanning through and sorting email, performing

routine tasks in an operating system or doing other mundane and time-consuming tasks.

There is also a continued debate21-23

over comparative outcomes between a real laboratory and

a virtual laboratory experiences in engineering education. While a real world laboratory gives the

student an opportunity to gain practical experience, the student must rely on self-learning of

concepts or on a human tutor who may not be available at a time when the student needs the

tutor most. Moreover, undergraduate laboratories at many universities depend on graduate

student teaching assistants who may or may not be conversant in the field represented by the

laboratory.

3. Virtual Power Laboratory

3.1. System Architecture

In the past few years, the authors have undertaken an effort to develop a virtual power

laboratory for undergraduate students in Electrical Engineering. Its goal is to complement a

hands-on laboratory but to be realistic enough to provide a satisfactory experience to students

who do not have access to a physical laboratory. VPL is developed on a universal web-based

platform that can be accessed anywhere by most mobile devices and modern computers6,7

. As a

proof of concept, nine virtual experiments have been developed for DC motors and generators.

Machine concepts are summarized using text, 2D and 3D graphics as well as multimedia

animation. The animated graphical user interface (GUI) plays an important role as it enables

students to review and retain basic concepts by building a bridge from the virtual environment to

the real laboratory. Multimedia is employed in order to provide visual guidance to the students

for better understanding of machines.

.

Figure 1. Architecture of Virtual Power Laboratory

As is shown in Fig. 1, the virtual machine laboratory is implemented in two parts: a Front-end

Client and a Back-end Server. The Front-end Client is an e-learning platform consisting of 1) a

Core Concept Knowledgebase, 2) an Experiment Knowledgebase, 3) Mathematical Tools,

and 4) a Virtual Instrumentation Core that are interfaced with 5) the User Interaction

Module through 6) a Multimedia Graphical User Interface Module. All these modules are

managed and supported by the Back-end Server that has the ability to offer secured http service

as well as dynamic database management. The Back-end Server shown in Fig. 1 has 7) a

Scalable Information Management Module, 8) System Administration Module and 9) an

Intelligent Tutoring Module to supervise and guide the students by providing necessary

guidance. Both the Front-end Client and the Back-end Server are designed to be secure, stable

and scalable. The VPL laboratory application can be accessed through the Internet on popular

mobile devices.

3.2. Front-end-Back-end Architecture

The Front-end-Back-end Architecture is applied as a universal framework for all the modules

in the VPL. This allowed the VPL to be developed in a distributed manner so that each module is

independent. Furthermore, each module inside the VPL architecture is able to apply its own

controller to control the flow of the data and manage data processing. The whole architecture is

implemented on Dreamhost Cloud Service. To minimize costs, we utilize Micro Instance24

which

provides small but consistent computing resources; capacity can also be dynamically increased in

bursts as needed. The Micro Instance is well suited for our system due to its low throughput and

periodic administration cycle by a small number of users. On the Micro Instance, the system is

built using the MEAN framework which is a high level web framework that stands for

MangoDB, ExpressJS, AngularJS and NodeJS. It is a combination of Backend technologies that

can provide a seamless architecture that includes a layer based client-server model25,26

. Also, a

Javascript/PHP based web server was hosted in parallel to provide intelligent tutoring services.

Javascript was chosen as the mathematical tool for machine simulations. The Front-end client is

designed using Adobe Captivate 7.0 and Bootstrap 3.0 framework so that the multimedia

interfaces and user interfaces are accessible for both the computer and most mobile devices. The

VPL server (http://vpl.temple.edu) is hosted and maintained at Temple University.

4. Intelligent Tutoring System

4.1 Architecture

This section presents the overall computational architecture of the Intelligent Tutoring System

(ITS) in VPL. The components of the ITS are divided into Front-end Client and Back-end Server,

as illustrated in Fig. 2. This Intelligent Tutor is a distributed client-server based web application,

capable of handling multiple services to groups of users with different roles. The user group

contains students, instructors and administrators. They have different authentication privileges

assigned in the system administration module in VPL so that they play different roles in using

the Intelligent Tutoring System. The Front-end Client for the Intelligent Tutoring System

contains the Logging Module and the Notification Module while the Back-end Server has

Question Mapping and Language Management Module.

The Logging Module manages all dialog data between the student and the

saves all dialogs in the database for future use with other students.

the core module that monitors the user’s behavior while he/she is working on the virtual

experiments and provides the user with correct action

makes repeated mistakes in executing an experiment

System is the Question Mapping Module

it to the pre-stored knowledgebase.

questions using a natural language processor.

Figure 2. Architecture of Front

The Intelligent Tutoring System is developed under the MEAN framework so

module is distributed and controlled by each processing sub

databases that are independent so that the queries can be processed more efficiently. Moreover,

in order to synchronize the data that is transferred betw

data link is created and implemented in Javascript and PHP.

ITS is similar to that of the AutoTutor

and output state objects, which greatly simplifies communication and interoperability between

modules. The modules operate synchronously by processing the data sent from the client or

requested from the database.

manages all dialog data between the student and the other

for future use with other students. The Notification Module


and provides the user with correct actions from the experiment database

makes repeated mistakes in executing an experiment. The central core of the Intelligent Tutoring

Question Mapping Module that processes questions asked by the users and maps

stored knowledgebase. The Language Management Module processes users’

questions using a natural language processor.

Figure 2. Architecture of Front-end Back-end based Intelligent Tutoring System

The Intelligent Tutoring System is developed under the MEAN framework so

module is distributed and controlled by each processing sub-module. It is built on several


in order to synchronize the data that is transferred between the Front-end and the Back

data link is created and implemented in Javascript and PHP. The basic architecture concept

is similar to that of the AutoTutor27

. Various modules are defined by their ability to input

ects, which greatly simplifies communication and interoperability between


other modules and

Notification Module is


experiment database if the user

Intelligent Tutoring

asked by the users and maps

processes users’

end based Intelligent Tutoring System

that each

module. It is built on several


end and the Back-end, the

The basic architecture concept of

odules are defined by their ability to input

ects, which greatly simplifies communication and interoperability between


4.1.1. Logging Module

The Logging Module manages all dialog data between the student and the other modules. It

also saves dialogs in the memory of the server for future use with other students. Site

redirections are also sent to the student if it is needed. The Logging Module performs as the

direct data transition hub between the Intelligent Tutoring server and the student. The dialog

exchanged in this module is capsulated to an object so that the data is secure and easily

transferrable to other modules.

4.1.2. Notification Module

The Notification Module is the core module that monitors the user’s behavior while he/she is

working on the virtual experiments. It connects to the database that stores the information of the

Experiment Knowledge Base. Thus, once an improper action is made by the user while doing the

experiment, the Notification module fetches the correct action from the experiment knowledge

base. In order to encourage the students to solve the problem by themselves, a counter with a

constant trigger number k is selected. If the same error is detected for k times then the

Notification Module will be triggered, and the student will be notified with the possible solution

to the error.

4.1.3. Question Mapping Module

The central core of the Intelligent Tutoring System is the Question Mapping Module that

processes the questions asked by the users and maps it to the pre-stored knowledge base. The

algorithm for the mapping is based on the calculation of Keyword Matching Likelihood (KML)

estimate between the extracted keywords of the question and the keywords in the knowledge

base. This module automatically returns three most likely matched answers and redirects the

user to the appropriate piece of knowledge or directly answers the question. If the Keyword

Matching Likelihood estimate is low for a certain question, it is passed to the Language

Management Module for further processing. The threshold for the KML estimate in this module

is selected by the knowledge expert.

The question-answer database is generated initially by the authors in this paper. The questions

include technical questions about machines and instructions for experiments as well as on user’s

guide. Table I shows the number of question-answer pairs that are pre-designed for the system.

The database is fully flexible and can be expanded by the instructors and administrators.

4.1.4. Language Management Module

The Language Management Module is implemented to handle the Keyword Matching

Likelihood estimate exceptions from user inputs. A Natural Language Processing unit is

included in this module to process the user inputs that are below the KML estimate threshold.

When the Language Management Module receives a question that cannot be handled by the

Question Mapping Module (i.e. the KML is below the threshold), it subsequently queries the

dialogue information state in the Logging Module and determines a response dialogue. The

updated dialog is passed to the Intelligent Tutor Server, which forwards it back to the student.

The natural language dialogue facilities are not expected to provide satisfactory response in all

conversational contexts. The quality of discourse depends on the subject matter, the knowledge

of the learner, and the expected sophistication of the dialogue strategies in the database.

TABLE I. QUESTION-ANSWER PAIRS FOR VIRTUAL POWER EXPERIMENTS.

Experiment

ID Virtual Experiment Titles

Number of

Question-Answer

Pairs

1 Magnetization characteristics of a separately

excited machine 10

2 Voltage buildup of a shunt generator

11

3 Loading of a shunt generator 16

4 Series generator load characteristics

15

5 Compound generator load characteristics

15

6 Starting of a shunt motor

11

7 Speed control of a shunt motor

14

8 Torque speed characteristics of a shunt motor

14

9 Torque speed characteristics of a series motor

Under

Development

4.2 ITS Function Methodology

The function methodology for the Intelligent Tutoring System is carefully designed by

combining the uses of all the distributed modules. It also considers the division of usages for

different user groups. Figure 3 illustrates the flowchart of the methodology. As the student user

asks a question through the Intelligent Tutor interface, the question will be passed to the

Question Mapping Module in the Back-end Server for likelihood estimate. The likelihood

calculator is designed to parse the question into keywords and calculate the likelihood estimate

by looping over all questions in the database. The likelihood estimate is then compared with

pre-defined threshold ��; if it is greater than ��, then the answer of the question will be queried

from the database and will return a response to the student. However, if the likelihood is less

than the threshold ��, then the question is passed to the Natural Language Processing Module.

Here the question is again parsed into keywords to compare the likelihood to the Natural

Language Question database. A Keyword Matching Likelihood is calculated and is compared

with a predefined threshold ��. If there is a database hit from the Natural Language Processor

Database, then the answer is generated and is prepared for sending. If there is no hit, the question

will be sent to the instructor group for an answer. This is an offline request so that the answer

will be sent once it gets answered by the instructor user. The output from the database is a JSON

format object so that it contains the available materials in the Front-end Client, such as learning

materials, links and videos and so on. These objects will be decoded at the Logging Module and

will be displayed on the Front-end side so that the student user can view it.

Figure 3. Intelligent Tutoring System flowchart

The roles of different user groups are defined differently as the student users have the right to

ask questions and view the answer

questions. The role of the administrator

pattern and update the database by manually inserting new data entries. It

party interrupts to the system so that the life cycle of the system

synchronously.

A brief interaction of a student with the Intelligent Tutoring System

First, the student asks a non-technical question which

processed by the Natural Language Processing unit. In the second instance,

intelligent tutor: “Why don’t I connect the series field in all experiments?”

processed by the Question Mapping Module, and

from the database: “Series field needs to be connected only when you are working with a series

generator/motor or a compound generator/motor.”

also points the student to the app

additional details on the topic. In the third instance, the intelligent tutor determines that the

student incorrectly connected the circuit

correct circuit configuration to the student.

Intelligent Tutoring System flowchart of the Methodology


ask questions and view the answer, while the instructor users are asked to answer offline

ns. The role of the administrator user is to regularly monitor the offline question

by manually inserting new data entries. It also enables the third

party interrupts to the system so that the life cycle of the system is maintained and updated

action of a student with the Intelligent Tutoring System is shown in Figure 4

technical question which didn’t meet the KML estimate so that it

processed by the Natural Language Processing unit. In the second instance, the student asked the

n’t I connect the series field in all experiments?” This question is

processed by the Question Mapping Module, and based on the KML estimate, returns the answer


generator/motor or a compound generator/motor.” If necessary, the Intelligent Tutoring System

also points the student to the appropriate page in the knowledgebase where the student can find

In the third instance, the intelligent tutor determines that the

student incorrectly connected the circuit 5 times so that the Notification Module provided the

correct circuit configuration to the student.

ethodology


while the instructor users are asked to answer offline

is to regularly monitor the offline question-answer

enables the third-

is maintained and updated

is shown in Figure 4.

didn’t meet the KML estimate so that it is

the student asked the

is question is

returns the answer


If necessary, the Intelligent Tutoring System

ropriate page in the knowledgebase where the student can find

In the third instance, the intelligent tutor determines that the

so that the Notification Module provided the

Figure 4. Question-Answer session of a student with the Intelligent Tutor

5. Conclusion

The intelligent Tutor functions as a laboratory instructor in VPL that oversees the student’s

progress in the laboratory and responds to requests and questions posed by the student. The

Intelligent Tutoring System can be used by students to learn basic DC machine concepts and

conduct virtual experiments similar to those that are typically done in a physical laboratory.

Instructors can enroll students for their course in VPL and update the question-answer database

regularly. The Intelligent Tutoring System is developed as scalable modular architecture so that

all modules can be running reliably and fast in a distributed manner using independent databases.

Further research is in progress to include other modules for virtual experiments on induction

motors, synchronous machines, and transformers.

With the addition of the Intelligent Tutoring System, this virtual laboratory serves as an

intelligent learning platform to bring the best of both worlds in the learning environment: the

richness of domain knowledge concepts and the tutoring supported by multimedia interactive

simulation that mimics a physical laboratory environment.

6. References

[1] Bipartisan Policy Center, National Commission on Energy Policy, “Task Force on America’s Future Energy

Jobs”, Final Report, Oct 2009. http://bipartisanpolicy.org/library/report/task-force-americas-future-energy-

jobs

[2] “IEEE Power Engineering Society Committee Report: The Power Engineering Education Committee (PEEC)

Task Force on Educational Resources”, IEEE Transactions on Power Systems, Vol 23, No 1, 2008.

3

1

2

[3] “National Science Foundation Workshop on the Future Power Engineering Workforce,” November 29-30,

2007. http://ecpe.ece.iastate.edu/nsfws/Report%20of%20NSF%20Workshop.pdf

[4] National Health Interview Survey, 1994, Center for Disease Control (1995)

[5] SBIR Phase II: Computer Simulation of Science and Technical Laboratory Exercises for Physically-

Disabled Students National Science Foundation (1997). Award Abstract 9710548.

[6] Gong, N, Butz, B.P, Bai, L & Biswas, S, 'A realistic intelligent multimedia virtual laboratory for power

engineering'. in 121st ASEE Annual Conference and Exposition: 360 Degrees of Engineering Education,

Indianapolis, IN, 15-18 June, 2014.

[7] Ning Gong, Brian Butz, Saroj Biswas, Li Bai, “Web Based Scalable Intelligent Multimedia Virtual

Laboratory for Power Engineering,”, ASEE Annual Conference and Exposition, Seattle, June 2015.

[8] Barr A. and E. Feigenbaum, “The Handbook of Artificial Intelligence”, Addison-Wesley, Vol. 2, Ch 9, 1982.

[9] Alonso, F. et al (1995). Teaching Communication Skills to Hearing-Impaired Children. IEEE Multimedia,

2 (4) 55-67.

[10] Sleeman D. and J. S. Brown, Eds., Intelligent Tutoring Systems, New York: Academic, 1982.

[11] Brown J.S. and R. R. Burton, “Diagnostic models for procedural bugs in basic mathematical skills,” Cogn.

Sci., vol. 2, pp. 155–192, 1978.

[12] Polson M. and J. J. Richardson J.J., Eds., Foundations of Intelligent Tutoring Systems, Addison-Wesley,

New York.

[13] Van Lehn, K. “Conceptual and meta learning during coached problem solving,” in Proc. 3rd Int. Conf.

Intelligent Tutoring Systems (ITS), C. Frasson,GGauthier, and A. Lesgold, Eds. NewYork: Springer-

Verlag.

[14] Butz, B.P. “The Learning Mechanism of the Interactive Multimedia Intelligent Tutoring System (IMITS)”,

Proceedings of the Frontiers in Education 2000 Conference, October 2000, pp T3D-11-T3D-16

[15] Hayes-Roth, B., “On Building Integrated Cognitive Agents: A Review of Alan Newell's Unified Theories

of Cognition”, Artificial Intelligence, Vol 59, pp. 329-341, 1993.

[16] Tambe M., Johnson W.L., Jones R.M., Koss F, and John Laird, “Intelligent Agents for Interactive

Simulation Environments”, AI Magazine, Vol 16, pp. 15-39, 1995.

[17] Mayberry, M.T. (ed.), Intelligent Multimedia Interfaces, AAAI/MIT Press, Menlo Park, CA, 1993.

[18] Maes, P., “Agents That Reduce Work and Information Overload”, Communications of the ACM, Vol 37,

No. 7 , pp. 30-40, 1994.

[19] Mitchell, T. et al, “Experience With a Learning Personal Assistant”, Communications of the ACM, Vol

37, No. 7, pp. 80-91, 1994.

[20] Etzioni, O. and Weld, D.S., “A Sofbot-based Interface to the Internet”, Communications of the ACM, Vol

37 No. 7, pp. 72-76, 1994.

[21] Bernard, R.M. and P. C. Abrami, “How Does Distance Education Compare with Classroom Instruction? A

Meta-Analysis of the Empirical Literature”, Review of Educational Research, Vol. 74, No. 3, pp. 379-439,

2004

[22] McPherson, T. , “A Rule Set for the Future”, MacArthur Foundation, MIT, pp. 1-23, 2007.

http://www.mitpressjournals.org/doi/abs/10.1162/dmal.9780262633598.001

[23] Sohn S.Y., H.Y. Park, and I.S. Chang, “Assessment of a complementary cyber learning system to offline

teaching”, Expert Systems with Applications, Vol. 36, no. 3, pp. 6485-6491, 2009.

[24] Dreamhost, https://www.dreamhost.com/, 2008

[25] Wenbin Jiang; Lei Zhang; Weizhong Qiang; Hai Jin; Yaqiong Peng, "MyStore: A High Available

Distributed Storage System for Unstructured Data," High Performance Computing and Communication &

2012 IEEE 9th International Conference on Embedded Software and Systems (HPCC-ICESS), 2012 IEEE

14th International Conference on , vol., no., pp.233-240, 25-27 June 2012

[26] Korostelev, Michael; Gong, Ning; Hu, Angela; Bai, Li; Wen, Kuang-Yi, "M2-PASS: SMS-based mobile

patient support and responding to challenges of transitional care," 2014 7th International Conference

on Biomedical Engineering and Informatics (BMEI), pp.762-768, 14-16 Oct. 2014

[27] Graesser, A.C.; Chipman, P.; Haynes, B.C.; Olney, A., "AutoTutor: an intelligent tutoring system with

mixed-initiative dialogue," in Education, IEEE Transactions on , vol.48, no.4, pp.612-618, Nov. 2005

Documents

An Intelligent Tutoring System for Multimedia Virtual