 Hi, my name is Deepak Khimani, I work in IIT Padras and I am here to talk about my course Knowledge, Representation and Reasoning, which is a part of artificial intelligence. Now, in this course, which is a companion to a course that we did on problem solving using search, the focus is on representation, knowledge, representation and reasoning. And the key thing that we want to talk about is, what does an agent know and what else does an agent know as a consequence of what it knows. So, what does it know about the world and what can it infer about the world. Now, why do not we use something like natural language. Now, the problem with language is that it is too rich and too ambiguous as illustrated in this cartoon by in So, World's book. We instead focus on formal methods, which is based on formal logic, in which our focus is on looking at valid forms of argument. So, you some of you might be familiar with this argument known as a Socratic argument in which you say that if all men are mortal and Socrates is man, then Socrates is mortal. Now, the key thing here is that it is a validity of the argument that we are concerned with, not the truth value of the premises. We say that as long as the premises are true, we will accept the conclusion. So, other examples are for example, all cities are congested, Chennai is city, Chennai is congested. All politicians are honest, Sambeth is a politician and therefore Sambeth is honest. Now, you will notice that the three arguments have the same form and it is a form that we are interested in and we say that if the premises that you give us are true, then the conclusion will necessarily be true. Now, there is a whole variety of languages that we use in starting with propositional logic. For example, here are two examples which says that if the earth was spherical, it would cast shadows on the moon, curved shadows on the moon, it cast curved shadows on the moon. So, it must be spherical. And here is another example, he says that if he used good bait and the fish weren't smarter than he was, then he did not go hungry, but he used good bait and he did go hungry, so the fish must have been smarter than he was. Now, only one of these is a valid argument and I will leave it for you to think about that. We need richer languages to talk about real world situations, we need the notion of variables for example, and quantifier over variables. So, we need to make statements like one of Tinker Taylor, the soldier, spy is a culprit, the culprits stole the document. So, we are saying that exists somebody who stole the document. Now, to reason with statements like this, we need first a logic and then we can make inferences that if these things happen and this is what really happens, then this is what you can conclude. We also look at something called description logics, for example, you can say that a progressive high tech company is a high tech company. We describe noun phrases, it is a logic of noun phrases and it is basically the formal basis of ontology. So, we can give a definition of a progressive high tech company and we can give a definition of a tech company and then we can say that this is a subclass of this. We also look at something called default reasoning, we say that if a Tweety is born then conclude that Tweety can fly. Now, this will not necessarily be true because not all birds fly, but we think that generally birds fly and therefore, we should be allowed to make such a inference. And this inference may later be contradicted by new facts which emerge. So, this is in the area of default reasoning. We then move on to event calculus which talks about time and action and change. So, for example, Jogesh made a cup of tea and left it on the table and Smita saw it and drank it up. So, when Jogesh came back, he saw that the cup was empty. So, here we are talking about things which are changing, the cup was full at one time, it was empty at another time when people did actions and so on. So, event calculus allows us to reason with that. Then finally, we move on to something called epistemic reasoning which is how agents reason with belief and knowledge and what they think other agents know. So, if you look at the previous story about this tea that Jogesh made, at the end of the story we can say that he concluded that Smita had polished off his cup of tea and Smita knew that Jogesh knew that she drank the cup of tea. So, this kind of representation in reasoning is possible in something called epistemic logic and that will be the final part of this course. So, the syllabus that we have will basically begin with proportional logic and move on towards epistemic logic going through all these things that you can see here. And the books that we will follow are a book by Rackman and Levis, the title is Knowledge Representation in Reasoning and a book which I recently wrote first course on Artificial Intelligence and there are some reference books that we also will use. So, I hope you will find this course interesting and I welcome you to come and be part of it. Thank you.