 Welcome back, this is a series of talks on open and distance learning. In this part, we will discuss about process issues that are involved in ODL. There are a few other issues which we will talk in the rest of the parts. What are the process issues in an online course? I have organized them into eight categories. The first of them is on how the course is designed, then about content creation and delivery. There is an issue about tagging the course. There are identity management issues. There is an assessment. How do you conduct exams and tests? There is a collaborative aspect of the course which is actually very important on an online course and then there is certification issues. Let us look at each one of them. Course design. I have a small example here. I hope you can see the detail. It does not matter if you do not. I will explain what this is. So, I have here data structures. It is a topic in computer science. A very fundamental elementary topic often taught as the first course after you learn programming. So, it says data structures, there are several sub topics like binary trees, linked list, AVL trees, multivay trees, B trees, hashing, linear hashing. The interesting thing about these topics is they actually follow a dependency. That is, before you can talk of AVL trees, the students should know about binary trees. There is a dependency like this. Before you can talk of binary trees, you should have talked about linked list. Before you can talk about linear hashing, you should have talked about hashings per se. So, these set of topics subscribe to this dependency diagram. Now, you must notice that this topics and the dependencies are universal. I call them universal models. This is the property of the domain. It is not a property which has been imposed by individual expert. Now, what for each one of these topics you can have a learning object. You can have multiple learning object. We will see what learning objects are in the next slide, but essentially learning object is nothing but an explanation of what the topic is. So, I have a set of learning objects, may be more than one for each topic. So, now if there is a course offered by let us say IITM on data structures, they will pick up a set of topics that are listed here and say this is our data structures offering. Another institute may pick up a different set of topics, again a subset of this and say this is our data structure offering. These two offerings may be different from each other. There could be variations, but both these confirm to this dependency. In both offerings before you can talk about multivariate trees, you would have talked of binary trees. Multivariate trees come before binary trees. When here multivariate trees come before binary trees. So, when you are designing a course what you have to do is take the domain, identify the topics in it, identify the dependency, pick up a set of topics and then construct the course. This is what I am calling course design. Another point is about content creation and learning objects. This is a small extract from a standard. When you try to create content for an online course, you do not make it into a one hour talk that is difficult to follow on an online lecture. If you are sitting in front of the computer and staring at it for an hour, your attention spans do not work out. So, typically what you do is you break them down into small learning objects. Now, what are learning objects? Learning objects are small units of learning ranging from 2 to 15 minutes. They are self-contained and often you should be able to take a learning object on any topic and that is all you need to know. You want to know. They are reusable. This is very important. That means if I have created a learning object for a particular course offering, I may be able to use the same learning object in another course offering because the topics are the same. They can be aggregated into small collections. You can string them up half a dozen learning objects into a lecture. Again, this is another important point. Each learning object has some metadata associated with it. That is, you describe what this learning object is about. Describe by a keyword which is descriptive in the domain and of course, some metadata like who has delivered it when and how long and what language and so on. So, when you are creating content, you essentially do it through learning objects. The next aspect of an ODL is how this content is delivered. So, I have put three points here. The first point says we weave assessment into content streaming. That is, an assessment is nothing but in this case a small question which asks about, tests a concept about the content that is being talked about to make sure that the student has understood what is being said to make sure that the student is alert. So, I am talking about a technology not simple recording of a talk that is delivered, but some interactive component which is inserted in between the talk which can accept a response from the student against a question and say this is correct or not correct. Go forward or not. It would not go forward if the response is bad is incorrect. It will say no, try again. It may even prompt you what the reason is. So, an assessment woven in between the content that is what we mean. Of course, we also need to be able to deliver to multiple platforms. This is sounds easy, but actually difficult to do because the student may want to learn from a desktop, from a handheld, from any from a phone for example. So, it is become very critical that I should be able to handle any kind of client. Now, capacity. When I am talking about a course on online, I do not know how many students are going to access it and at what time. So, I should be able to handle a large number of students and the demand may be elastic. Suddenly, I may have 1000 users coming online and another time there may be only a few 4 or 5 of them accessing the content. So, I have to build my system to be able to handle an elastic demand. Capacity is very critical in the delivery. So, this is another issue in having an online course. So, the flipped classroom is again an issue related to delivery. Very often what people do is when we have content that is recorded, I might actually use it in a interesting model. I might have a face to face class that is floated in a landed university and what I do is I do not use the classroom to actually deliver the content, to deliver the lecture. I tell the students look the content is already recorded and put on YouTube go and listen to this episode and come back to me. And what we do is once they listen to the content and come to the classroom, the classroom time is used for interaction of different nature. It can be used as a tutorial we can discuss. Have you understood what is being said there? Are there any difficulties? Let us talk about it. I will ask a question and say what did the instructor say on this? We can actually try to use the classroom time to do a tutorial to take to solve a problem to apply the knowledge that is learned. The classroom time can be used for discussion amongst the students. So, you get multiple perspectives and a deeper understanding into what is being said. So, this is called the flipped classroom model sounds very attractive. So, sometimes the content that is recorded and put online may be delivered in a flipped classroom model as well. So, again that is why I call it a delivery issue tagging. So, when you have a learning object, when you have a lecture recorded and put online, what you would like to do is to search for it. You may have a thousand videos recorded. You want to know, you want to pick up a particular one, that one on data structures, B trees. So, how do you locate that? If you can assign metadata, then you can use the metadata to identify that. Metadata is data about data. The other advantage that you will have if these are tagged is you may be able to actually compose a lecture through a program. I would say please get me that lecture which I have delivered in 2010, which talks about B trees and that lecture which I have delivered in 2011 on binary trees and put them together into one lecture, string them together. So, I have this what I said is a query and the actual extraction of the video and its delivery is done through a program which is composed at runtime. So, this is a very powerful technique if we can do tagging appropriately. So, tagging is a very important aspect when you are doing online delivery, identity management. This is very interesting. So, typically what people found is when they have an online course, individuals register with multiple identities, same person registers twice, twice, several times in the same course. Sometimes you may want to figure out that the person who has registered in that course with this name is the same person as X in this course. Across courses you may want to resolve registration because you may be offering a degree and to make sure that the same person has done both courses. You need to resolve across courses. How do you find out over the web that this is the same person? It is still an open problem. You know you could use biometrics for example or you can use identities like Facebook and Google login through your Facebook profile and so on. But identity is a very critical important issue and let me quickly add that we do not have good solutions as yet because it is possible to masquerade unless you go use powerful technology like this and these are expensive and difficult to roll out. The next problem is assessment. It is again very critical. Here we are talking about certifying a student that he or she has done this course. That means I need to conduct an exam, an online exam. Student is somewhere in the universe. So how do I make sure the student does not copy? The student does not get assistance while taking the exam. So that is the problem with assessment. The second problem is if you have a large number of students, 100,000 students registered correcting them is going to be as necessarily to be automated. So manual correction is difficult. Now if you are going to do manual correction, you have this issue of if you are going to do automated correction, multiple choice questions kind of exams cannot test the conceptual depth of the learning process. Often I require some detailed descriptive answers as well. So the question type that is allowed in an automated process is very limited which can test only certain kind of concepts and I cannot have any other kind of testing because I have to deal with large number. So assessment is again an issue which is still open. There is some handle on this. People are working on this. An example of this is a system built by Professor Jeffrey Ullman from Stanford. So he conceptualized something called a root question. If I am looking for a long answer and I need to correct that using a machine, I should be able to pose that question which requires a long answer as a series of short answer questions, multiple choice questions. So if the student manages to navigate them, we could certify that he knows enough to have answered the long answer question, an equivalent long answer question. So that is the hypothesis there. So they have built a system and this will probably give a handle, things like this will give a handle on how do you scale up long answer questions to large numbers. Of course Pearson view has a different kind of approach to this. They actually have test centers in 165 countries. So they have physical centers where the students go sit down and take the exam. So this is going to handle my identity management problem. It is not addressing this question but of course I need to finally have to correct that physical paper that is conducted but I can make sure that there is no masquerading that is happening while taking the test. The other process issue when an online course happens is collaboration. It is a well known fact that students often learn better from their peer group. Interaction with the brethren is a very powerful learning mechanism and if it is an online course which I take sitting in my room, I am going to miss out on that. A classroom interaction or sitting in a lab and doing it together where I shout out to the student across saying hey how do I delete this file which is missing when I am doing an online course. So collaboration issues have to be built into the online course and another aspect is these interactions actually produce content. We will see more about it when we talk about the kind of online courses that you could have. So very fact that two different minds have come together and talking about it will allow you to get two different perspectives and connect up different bodies of knowledge perhaps and knowledge gets created in the process and that is another major take away from a course. So when you are doing an online course you have to provide facilities where students can collaborate and connect between concepts and that is another major learning of it. So what we are saying is the connections that happen between content and of course between students, between the participants they are also a major take away. I come to know that there is an expert on normalization from there. So that is a big take away from me. So collaboration is the other issue on open distance learning certification. What happens in certification? It is related to assessment. So when students take this online course what kind of certificate do I give? Do I give a participation certificate? If I make sure that they have all or each one has seen all the videos then I can which I can track through technology then I can say he has participated in the course. Do I give a participation certificate or do I give a competency certificate? He has not only participated but he has learnt certain kind of things. Of course if I want to give a competency certificate I need to do assessment. I need to make sure that he is competent in the body of knowledge and the next issue is who gives the certificate? Does the instructor give the certificate or the institute gives the certificate? If you see the courses at this point of time the instructor who is giving the certificate. Institutes are not yet giving certificates in MOOCs. So that is all about process issues in ODL. Thank you.