Introduction
Distance learning may be defined as a form of education in which the teacher and the student are separated by space and sometimes also by time. This type of learning is normally characterized by voluntary participation by the student, pace of learning is normally determined by the student rather than the teacher and it is also characterized by lack of continuous communication by the teacher or student. The two latter parties are brought together by either the use of print media or through the use of technology.
Intelligence may be defined as the ability to adopt or change one’s behavior to suite their environment. Therefore artificial intelligence is when computers or other machines can be made to perform some duties that human beings can only perform when they use intelligence. Artificial intelligence can come in various forms. It may be sub- human where the computer or machines can only be able to perform simple tasks that are concentrated on a particular domain. Artificial intelligence can also be almost human or super human. The latter normally occurs when a machine is capable of carrying out all activities that human beings do while the latter will occur when a machines are capable of exceeding the potential of human beings. (Bransford et al, 2000)
Unfortunately, this is a problem that most researchers engaging in artificial intelligence research face. No computer or machine has been able to perform almost as many functions as the human brain. This is because what constitutes intelligence involves a complex number of factors. What human beings call common sense is a trait that is not possessed by computers. Currently, computers have only been able to perform certain specific functions. They have only been able to display some signs of artificial intelligence. These elements are what will be dealt with in the subsequent parts of the essay. They include;
language understanding-computers can be able to answer questions and answers in a certain programmed language problem solving-where computers can be able to tackle certain problems like given preconditions e.g. determining moves that are required to win in a chess game
Perception-the machine is able to sense areas within its environment and then react accordingly. An example are robots in assembly lines
Learning-computers are presented with a situation once then they store that information and use it for other similar circumstances e.g. using the format for past tenses of certain words then applying it to others.
Reasoning-when computers can be able to infer tasks appropriate to the given circumstances. (Gunther & Heide, 2002)
Technological applications in distance learning
In the olden days, distance learning was done through the use of print media or through the use of radios and televisions. Here, lecturers would talk to wide audiences who listened or watched. But this form was not very successful as most of these professors may have had good knowledge of their subject matter but lacked the ability to keep audiences interested in them. Then with the introduction of the Internet, there was a need to have more interaction between teachers and students. This has brought about the use of the following in distance learning. Some of these may incorporate some forms of artificial intelligence
-Audio-conferencing
-Video-conferencing
-Butellin board system
-Fiber optics
-Cable
-Satellite
-Audio-graphic teleconferencing
Areas that could require artificial intelligence in distance learning
There are two specific areas that will receive special emphasis in the essay. First of all, the issue of giving exercises during distance learning. Tutorials are a crucial aspect of distance learning because it enables the student to engage in active learning. They will be able to internalize concepts when there is a hand on approach. Tutorials are a good way to analyze a student’s understating of concepts since they can be able to apply them to solve problems.
The other area in distance learning that will receive emphasis in this section is the issue of communication. There are three parties that need to be coordinated in distance learning. These include the teacher, student and other members of the class i.e. his peers. There is a need for students to communicate ideas to these parties. Research done in distance education has indicated that courses in which there is adequate interaction through real time communication, retention rates increased to 75% from 20%. (Nicolai, 2003)
Currently exercises are normally administered through email. What has been happening is that students are given exercises through web servers that belong to their teachers. This is then followed by the student’s individual work. After completing the assignment, a student either has the option of sending the work to his teacher through email or they could actually hand it in manually to the lecturer. The lecturer does the marking on his or her own and the results may then be emailed back to the student. There are some shortcomings that come from such a system and will therefore necessitate the use of artificial intelligence. Some of them include;
1) There is a lack of individualization of the exercises to suite the particular needs of a student. In this traditional system, all the questions given to students of the class are the same.
2) Students who get problems when trying to solve a given problem are not able to get immediate responses to their queries. This could make work seem excessively hard for the student and could also make them lack motivation to continue with the course if everything seems to be so difficult
3) Teachers evaluate grades manually and this causes a lot of time and resource wastage on both the part of the student and the teacher
4) Lastly, students are not able to interact with their course mates to exchange ideas and help one another in the course of performing exercises or tutorials (Brunsmann, et al 1999)
In light of these inadequacies, Artificial intelligence will come in to solve some of them. Artificial intelligence could be used to measure how proficient a particular student is in doing a particular exercise. This is then followed by measuring proficiency of other students and comparing them. The student can then be given exercises that suite his/her level in terms of proficiency. Artificial intelligence can also help the student to be able to communicate with his or her peers who fall under the same level. On top of that, the concept can also help teachers from spending too much time when marking student wok as this can be done by the machine. Therefore incorporating all these aspects into one will result into a tutoring system that has some from of artificial intelligence. There is some work that has been started on such a project by the University of Mannheim.
Artificial intelligence can be able to merge a number of issues at the same time to help toward the solution of these problems. First of all, it is possible to program a computer to enable it to generate problems for students and also to be able to perform symbolic calculations. It is also possible to be bale to make the machine adapt to a given set of multiple choice questions. On top of that, one can be able to control an online system by himself. This is crucial aspect to any student in that they feel that they will be learning at their own pace. Some researchers claim that when a student possesses control over his or her learning, then they are able to develop meta-cognitive abilities. Besides this, artificial intelligence also enables the student to get immediate feedback about his or her progress and this assists greatly in the process of learning. (Ossi, 2000)
Therefore there is need to integrate all these components of artificial intelligence into one. This means that one can be able to utilize a hybrid model. The model can consist of a feedback system and also a system that allows the student to be able to control his or her learning. These two aspects will allow the system to be able to merge two aspects i.e. an aspect of behavior where a student is able to practice continuously and consistently and the aspect of cognition since students are able to control their own learning.
Artificial intelligence can therefore be applied to distance learning through the following criteria; it can be able to use some algorithms to measure how proficient a certain student is. Then the student is given feedback, this is then followed by immediate adaptation to the kind of proficiency that students posses. It is a known fact that students require something that is challenging to them in order to maintain their interest. This is what will be achieved by this hybrid model of artificial intelligence. It is also crucial to monitor exercises such that a student also does not have to do something that is beyond his or her means. When exercises are too difficult for the learner, they are bound to loose motivation and may even drop out of their course or class. (Mao and Lin, 1992)
What makes artificial intelligence unique to current systems in place during distance learning?
Currently there are some online systems designed to measure the capability of a given student say in the solving of a mathematical problem. However, these systems simply measure outcome, they do not send feedback to students and they also do not allow communication to classmates through video conferencing. On the other hand, there are systems that can be able to facilitate video conferencing but such systems do not measure performance levels of students. This implies that the students who interact may not necessarily belong to the same category. When students of different capabilities are lumped together, it may not lead to constructive learning. Therefore, the current systems of video conferencing are mainly between students and their tutors but not between students especially those of the same level.
Academic reasons why group interactions are important
Group interactions help students in a number of ways. First of all, they make the student active. This is because the student is now aware that there are other people who may be waiting for him/her and this helps the student to stay on his or her toes. On top of that, a teacher will be able to know hat kind of level a student has reached just from the kind of group which that student belongs to. A student is also greatly motivated. This is because they are able to enjoy the fact that they are working together with others and conferring with them. Group interactions give a platform for doing exercises. These exercises can then be evaluated by a teacher and student performance can then be determined. Group work also draws less attention from the tutor as the sole source of knowledge in the class. It allows students to access knowledge from their peers and also boosts their self confidence.
Researchers in the field of education have shown that for one to able to understand given facts in a concrete manner, then it is necessary for others to explain it to them in depth. By doing so, a student will be able to apply that knowledge to solve certain skills. Mayes et al (2000) recommends dialogue for most students to enable them to be able to fill in any gaps or misunderstandings in the course. He claims that in every learning cycle, there must be some form of two way interaction which he calls ‘the externalization processes. Research has also shown that the use of team work in learning helps students to memorize facts quite easily. This is improved by the fact that there will be increase motivation on the part of the student. It is also a method that is acceptable by many students out there.
Traditionally, most online systems favored the interaction of teacher to student only. This could present several drawbacks to the learning process. First of all, the student felt isolated since he or she did not know about the whereabouts of other students in his class. On top of that, the student may also have some restrictions bout constantly disturbing his/her tutor since this person is viewed as the ultimate expert. These are all not conditions that can bring out the optimum level of performance from a student. This is where artificial intelligence will come in. It will aid in the creation of peer communication rather than the traditional and often impractical approach of help desk.
Problems in distance learning that need to be solved with the help of artificial intelligence
Distance learning by its very nature poses the problem of conducting exercises. What is currently happening is that teachers give students assignments through the Internet or students can download the information. This assignment is then done by the student and then it is emailed back or it can be up loaded onto a particular teacher’s web server. While the student is doing their work, there is no room for interaction and this causes feelings of isolation. Students and teachers have no way of knowing performance levels of the students and communication between peers is not a common feature.
How the different aspects will be merged to make hybrid model of artificial intelligence
Artificial intelligence will allow the student to be able to work on exercises that go with their level of intelligence. The more advanced a student gets, the more challenging the questions that student will be receiving. This is also coupled with the fact that there are feed backs given to that student. This will help the student to assess his or her progress.
The system will also be monitoring who is online, measuring their performance at any one time and can be able to gauge who is working on similar exercises, then this can allow it to merge members belonging to the same groups automatically. This allows the student who wishes to interact with others to alert the system through some form of input like pushing a button. In case the student would like to work with some other groups that may not necessarily belong to his/her performance level, then all they have to do is notify the system and they can start enjoying video conferencing through the aid of the system. (Mao and Lin, 1992)
Design of the hybrid model of artificial intelligence
Since we have looked at the needs that should be addressed by the artificial intelligence model, then it is crucial to incorporate those needs into the design of the model. The artificial intelligence model system can be made up of a server-client web based technology.
The first portion will consist of a web client. This will act as interface for the student. There will be a screen shot of the student and a bar for navigation that will allow the student to deal with specific exercises. The main exercise is placed at the centre of the screen and the student can be able to perform this given exercise as often as he/she likes. It should be noted that this type of exercise is dynamic in nature and changes with the advancement level of the student. The exercise is conducted through a Java applet. Then when the student has been able to complete the assignment, they should be able to hand in information through the server. This exercise is then marked by the system and results are given. The system will have certain performance criteria installed and will be able to measure the student’s performance against these criteria. After doing that, the server is then able to place the student in different ranking. This will be after the student has completed a given round of questions. (Allen, 1997)
There will be a ranking list placed on one side of the screen for the student to be able to access it. On top of that, there will be a process that will lead to video conferencing. This can be achieved through the use of a URL that will call a certain student. This will be achieved through a Session Initiation Protocol. When the student on the other side sees this message, then they can be able to accept the call and video conferencing can then begin. This may apply to a student that is within the same level as the original learner since a list will be displayed. However, the same process can also apply to tutors because the student may be in need of expert advice on the subject matter. The beauty about the use of artificial intelligence to achieve all these tasks is that there will be flawless communication between members of a class or course; there will also be the aspect of video or audio communication without a break in between of telephony or the computer.
How learning performance can be measured
There is a need to integrate some performance indicators into the artificial intelligence system. They will be as follows
time
proficiency
reliability
Time will be measured by the system in milliseconds. It will be an indication of the duration a student takes from the onset of a given exercise to the point when they have completed that given task. This is a crucial aspect in any form of learning because we can liken it to a typical classroom scenario. Here, a teacher can be able to measure a student’s speed when giving them exercises since they are in face to face contact with each other. This aspect is a crucial indicator of how a student performs because there may be times when the student gives a correct response but they take hours to come up with that answer.
Proficiency will show how difficult the problem is and to what level did the student reach when attempting to solve the problem. This is because some students re capable of completing the entire solution while others may come very close to the solution without necessarily completing it. Others can also start with the initial pats of the solution then they come to a stop.
Reliability is used to indicate the consistency level which student possesses in answering given questions. Most of the time it is possible to find that there are numerous cases of inconsistency. A student is able to do a certain type of question at one time and then at another instance they are defeated. It would not be a good idea to infer that a student is good at a given topic just because they were able to answer one type of question. In this sense, reliability is a combination of the proficiency level in answering certain types of problems. This criterion is also known as confidence level.
The performance level is kept in the database and is quite important because it will be used to determine what rank students falls in. It will also be used to place that given student in a category of group members. This will then form the basis of video conferencing tools that will eventually take place between the peers. Performance rankings will also be the criterion upon which a different set of questions will be offered to the student
How an adapted exercise can be generated
These can be achieved in two ways.
1) First of all, there can be numerous questions placed in the database of the server. Then each question is given a certain level of difficulty. When need arises, then a specific question can be chosen
2) Alternatively, questions can be generated at any one time by the use of a program. This is a process that will be continuous with time
Selecting questions at will from a given database can be achieved through the use of an algorithm that was suggested by Huang (1998). This programmer suggested that for the algorithm to work there must be a huge sample of questions. It utilizes an algorithm called CBAT-2. This algorithm can adopt to determine the difficulty level of the question. In this algorithm questions are weighed by two factors i.e. the difficulty level and the level at which one can be able to guess the right answers. This is what is called the guessing factor.
Selecting the right questions is a two step procedure. First of all, the system is able to choose the type category to analyze the students in. This is then followed by picking out questions from the relevant area of study. All these will be based on the extent of difficulty of the question. The reason why Huang suggests a large amount of questions to choose from is because if the questions were few then there may be chances of cheating or guessing.
Automatic evaluation of questions can be done by the sever only if the questions fall under the following categories
multiple choices that have one answer
Multiple choice with many answers
mathematical problem with one solution
filling in blanks
ordering problem
other types of questions that be graded
The first bunch of questions; multiple choices with one answer, is the most well know one and they do not require any complicated systems for analysis. However such systems may have their own drawbacks. This is because they leave lot of room for guesswork especially if the sample space is small. On top of that, there is also the possibility of cheating since students do not give in their exact names. (Liang, et al, 2005)
The second types of questions are similar to the first but with a slight variation. First of all, there are many right answers that may be available to the student. He or she must make sure that they only choose one correct question. Then the questions can be analyzed such that those who select a right answer get a point for it and any wrong answer will lead to a loss of points. There are a number of shortcomings that may arise from such a system. These include the fact that chances of guessing the right answer are incrassated if there is more than one right answer that corresponds to the question. On top of that, scores are not clear since it is possible fro students to loose marks and to gain them.
The next set of questions is a mathematical problem that can only have one solution. The computer can be programmed to ascertain that the student gets the right answers and scores are only given to those who get it correct. However, this can be improved when dealing with more complex problems in that the system can be able to cater for some predictable or allowable arithmetic errors and hence scores are tallied against those results as well. This will be important in eliminating the all or nothing law.
Other types of questions that can be generated include filling in blanks. What normally happens is that a student is meant to fill in the gaps and his or her solution is matched against certain built in solutions within the system. Correct answers may vary so a number of key words ought to be provided for this. On top of that, there can also be the possibility of setting answers that may be very near the true ones, Students who fill in those near ones can be awarded half the marks. This will go a long way in encompassing some students who may have an idea but have not yet fully developed it. (Allen, 1997)
The system can also generate an ordering problem. As the name suggests, such kinds of systems are normally evaluated through the correct arrangement of a series of steps. If a student is able to put all of them in the right order, then he/she should be able to score the maximum points. However, there is also room given to students who might be somewhere near the answer. Those who miscalculated one step will loose few marks, those who missed two or three will result in loss of more marks. This goes to show that the system can be adjusted to evaluate students not simply on the right and wrong criteria but also based on their attempts
During the generation of the questions, what normally happens is that the kinds of questions that are essential to the course are labeled. Levels of difficulty to the questions are also given but this must be done manually. There will be a formula that will help the system to compute what is the level of performance of a given student. After doing this, then a Java applet is utilized to do the following actions
-retrieving data from student
-presenting questions to student
-measuring and recording performance criteria
-giving feedback of the results to the server
The main advantage of such a system, where there is a vast number of questions that are generated, is that this can be used in content domains. On top of that, teachers may get some assistance from an editor called WILD. This will help them in the process of generating questions.
Generating questions to be used at any one time in the server
The system can be programmed such that it is able to identify and scrutinize questions on the basis of three levels. First of all it should be able to asses the level of proficiency if the user, this can be called the level of discrimination. The second parameter to check for is the level of difficulty. This aspect was in built into the system by tutors and they are the ones to determine how easy or difficult a question can be. Lastly, the system can be able to respond to the level of guessing which student poses. (Mitsuru, 1997)
Normally what occurs is that a system will be given the moat important parameter to look out for. This is essential because the most important parameter among the three is the level of proficiency of the student.
Implementing the server containing artificial intelligence
Implementation of the server is done by incorporating any of three perspectives
template applets
independent subject
clients for the artificial intelligence model
a server for the artificial intelligence model
The first thing one need to examine is the artificial intelligence server. The program that can be used for such a server is Linux in relation with DBMS. Here, there will be a table in which all the information regarding a particular student can be entered. These include their results, names, groups and even their logins. In addition, there should also be the use of Java for conducting the exercise and also for communication. (Blackboard Inc., 2002)
There should be an interface for the web and this can most likely be done through the use of Apache which is a web sever. Administration can be done through the use of a PHP script.
An example of application of an artificial intelligence server on distance learning
We will examine the use of the system to be able to evaluate students doing fractional arithmetic.
It is important for the tutor to define his/her objectives. In this case it can be that the teacher wants his students to be able to add, subtract, multiply, divide and subtract tractions. This should then be followed by designing the problems to be done in the exercise. It should be noted that the questions should be varied in their complexity. Measurement to be utilized in calculation will come out automatically. Values in the form, of integers are then given to the questions that have different levels of complexity. It is then up to the tutor to apply the Java applet that puts forward the questions to the student.
The results are then meant to be integrated into the Java framework where students are supposed to access the information from. The framework’s purpose is to determine the kind of performance level that students attain. It also allows the student to be able to access his/her results. It is important that after the tutor has completed al the above procedures, then they should register their applet so that students may be able to access it.
Conclusion
Distance learning by its very nature presents certain problems to the tutor and the student. First of all the student is not able to interact with members of his class. This is worsened by the fact that communication with tutors may also be a problem. Tutors are also faced with the difficulty of marking scripts manually and can therefore spend too much time on this. Artificial intelligence can be incorporated into distance learning through a server that is able to perform many crucial tasks.
First of all, the server will present questions to a student, mark these questions and assign a performance level. The system will then be able to use this proficiency level to rank the student and the next set of questions that the student will do will depend on the type of score. The more student scores, the more difficult the next question. On top of that, artificial intelligence will also be able to choose other students that may fall in the same category as the given client and he may be able to interact with them and exchange ides. This will also apply to tutors. Another fascinating feature that can be incorporated into distance learning is the fact that students can portray a level of control over the questions they choose. (Claude & Gilles, 1990)
Questions can either be generated through the use of a large pool of questions arranged according to level of difficulty or through the generation of a given question at any one time. Implementation of the idea requires four main aspects. There will be a general sever and some templates whose purpose will be to act as a framework. On top of this, there will be the use of a web interface and then the clients for the system. By applying these steps, artificial intelligence will have played a crucial part in the improvement of distance learning.
0 comments:
Post a Comment