PAGENO. INDEX SYLLABUS 2 Unit IV 1.Game playing techniques 1.1.Mini max procedure 2. Alpha-beta cut-offs 3. Planning 4. Study of the block world problem in robotics 5. Introduction to understanding and natural languages processing Unit V 6. Introduction to learning 7. Various techniques used in learning 8. Introduction to neural networks 9. Applications of neural networks 9.1. Real-life applications 9.2.Neural networks and neuroscience 10. Common sense 11. Reasoning 12. Some example of expert systems 12.1. Advantages 12.2. Disadvantages REFERENCES Page 1 3 3 5 6 7 9 11 11 11 17 17 17 18 19 20 20 21 21 23 IT 833 Artificial Intelligence Branch : Information Technology, VIII Semester Course: Artificial Intelligence Unit I: Meaning and definition of artificial intelligence, Various types of production systems, Characteristics of production systems, Study and comparison of breadth first search and depth first search. Techniques, other Search Techniques like hill Climbing, Best first Search. A* algorithm, AO* algorithms etc, and various types of control strategies. Unit II: Knowledge Representation, Problems in representing knowledge, knowledge representation using propositional and predicate logic, comparison of propositional and predicate logic, Resolution, refutation, deduction, theorem proving, inferencing, monotonic and nonmonotonic reasoning. Unit III: Probabilistic reasoning, Baye's theorem, semantic networks, scripts, schemas, frames, conceptual dependency, fuzzy logic, forward and backward reasoning. Unit IV: Game playing techniques like minimax procedure, alpha-beta cut-offs etc, planning, Study of the block world problem in robotics, Introduction to understanding and natural languages processing. Unit V: Introduction to learning, Various techniques used in learning, introduction to neural networks, applications of neural networks, common sense, reasoning, some example of expert systems. Page 2 Mini max procedure Minimax is a decision rule used in decision theory. 2. it has also been extended to more complex games and to general decision making in the presence of uncertainty. Beside the attraction of the topic to people. However. and the other's viewpoint is exactly the opposite. and its well-defined states and rules. This type of game is common. game theory. states also depend on the actions of other players (systems) who usually have different goals. A special situation that has been studied most is two-person zero-sum game. for most interesting games. in multi-player games. Given sufficient time and space. the minimax procedure can solve the problem given sufficient computational resources. For two-person zero-sum perfect-information game. The most common used AI technique in game is search. Game playing techniques Game playing has been a major topic of AI since the very beginning. though not all competitions are zero-sum! There are perfect information games (such as Chess and Go) and imperfect information games (such as Bridge and games where dice are used). each state can be evaluated by a score from one player's viewpoint. and easy to analyze. In some other problem-solving activities. if the two players take turn to move. that is. we Page 3 . it is also because its close relation to "intelligence". This algorithm assumes each player takes the best move in each step. usually an optimum solution can be obtained for the former by exhaustive search.Unit IV 1. state change is solely caused by the action of the system itself. First. covering both the cases where players take alternate moves and those where they make simultaneous moves. where the two players have exactly opposite goals. though not for the latter. However. such a solution is usually too inefficient to be practically used. statistics and philosophy for minimizing the possible loss for a worst case (maximum loss) scenario. Originally formulated for two-player zero-sum game theory. All nodes in the tree will all be scored. and the path from root to the actual result is the one on which all nodes have the same score. Example: Because of computational resources limitation. MAX and MIN. while the other player (MIN) takes the action that leads to the lowest score. At each step. determined by the depth of the search tree. in the state graph. the search depth is usually restricted. Page 4 . with the difference of possible win paths as the henristic function. Example: Tic-tac-toe.distinguish two types of nodes. and go backwards towards the root (the starting state). MAX). and estimated scores generated by a heuristic function are used in place of the actual score in the above procedure. Minimax procedure: starting from the leaves of the tree (with final scores with respect to one player. one player (MAX) takes the action that leads to the highest score. When applied to a standard minimax tree. Search is depth-first. Go. which never decreases. etc.3. In Alpha-Beta Pruning. These values are set and updated when the value of a child is obtained. It stops completely evaluating a move when at least one possibility has been found that proves the move to be worse than a previously examined move.). and stops at any MIN node whose beta value is smaller than or equal to the alpha value of its parent. Alpha-beta cut-offs Alpha–beta pruning is a search algorithm that seeks to decrease the number of nodes that are evaluated by the minimax algorithm in its search tree. it returns the same move as minimax would. Chess. It is an adversarial search algorithm used commonly for machine playing of two-player games (Tic-tac-toe. Such moves need not be evaluated further. which never increases. Page 5 . each MAX node has an alpha value. but prunes away branches that cannot possibly influence the final decision. as well as at any MAX node whose alpha value is greater than or equal to the beta value of its parent. each MIN node has a beta value. (6)MIN[==8] |-----------X |-----------X This method is used in a Prolog program that plays Tic-tac-toe. humans.Examples: in the following partial trees.(5)MIN[<=0] ----. Page 6 .(4)MAX[==3] | |-----------.(4)MIN[==3] | |-----------. We will only focus on the simplest AI planning problem.g. The world is typically viewed to consist of atomic facts (state variables).(3)MIN[==5] | |-----------. It involves choosing a sequence of actions that will (with a high likelihood) transform the state of the world. characterized by the restriction to one agent in a deterministic environment that can be fully observed. 4. and actions make some facts true and some facts false.(2)MIN[==3] ----. Planning The planning problem in Artificial Intelligence is about the decision making performed by intelligent creatures like robots. step by step.(3)MAX[==5] | |-----------. (1)MAX[>=3] ----. the other children of node (5) do not need to be generated. More complex forms of planning can be formalized e.(6)MAX[==0] |-----------X |-----------X (1)MIN[<=5] ----. and show how the planning problem can be solved automatically. in the framework of Marvov decision processes. so that it will satisfy the goal.(2)MAX[==5] ----. or computer programs when trying to achieve some goal. with uncertainty about the effects of actions and therefore without the possibility to predict the results of a plan with certainty.(5)MAX[>=8] ----. In the following we discuss a number of ways of formalizing planning. abstract domains for both analytical and empirical studies. For example. puts block a onto the top of the stack containing block b. The problem is to parse a series of commands that instruct a robot arm in how to manipulate blocks that lie on a flat table. move a over b where a and b are block numbers. Page 7 . Intelligent Control. All of the methods described below are equally applicable to all of these other problems as well. and so on. which has other important applications in Computer Aided Verification (reachability analysis. puts block a onto block b after returning any blocks that are stacked on top of blocks a and b to their initial positions. Initially there are n blocks on the table (numbered from 0 to n-1) with block bi adjacent to block bi+1 for all 0 <= i < n-1 as shown in the diagram below: The valid commands for the robot arm that manipulates blocks are: b move a onto where a and b are block numbers. In this problem you will model a simple block world under certain rules and constraints. you will "program" a robotic arm to respond to a limited set of commands. an early AI study of planning and robotics (STRIPS) used a block world in which a robot arm performed tasks involving the manipulation of blocks. discrete event-systems diagnosis. after returning any blocks that are stacked on top of block a to their initial positions. 5. and many of these methods were initially developed and applied in the context of these other problems. model-checking). Rather than determine how to achieve a specified state. Study of the block world problem in robotics Many areas of Computer Science use simple.The most basic planning problem is one instance of the general s-t reachability problem for succinctly represented transition graphs. onto block b. There will be no syntactically incorrect commands. All blocks on top of block b are moved to their initial positions prior to the pile taking place. If there is at least a block on it. Your program should process all commands until the quit command is encountered. You may assume that 0 < n < 25. and any blocks that are stacked above block a. Page 8 . moves the pile of blocks consisting of block a. The blocks stacked above block a retain their original order when moved. The blocks stacked above block a retain their order when moved. puts the pile of blocks consisting of block a. Each original block position numbered i ( 0 <= i < n where n is the number of blocks) should appear followed immediately by a colon. onto the top of the stack containing block b. Don't put any trailing spaces on a line. Output The output should consist of the final state of the blocks world. You may assume that all commands will be of the form specified above. followed by a list of blocks that appear stacked in that position with each block number separated from other block numbers by a space. the colon must be followed by one space. quit terminates manipulations in the block world. Input The input begins with an integer n on a line by itself representing the number of blocks in the block world. Any command in which a = b or in which a and b are in the same stack of blocks is an illegal command.pile a onto b where a and b are block numbers. All illegal commands should be ignored and should have no affect on the configuration of blocks. pile a over b where a and b are block numbers. one command per line. The number of blocks is followed by a sequence of block commands. and any blocks that are stacked above block a. There should be one line of output for each block position (i.. n lines of output where n is the integer on the first line of input). Introduction to understanding and natural languages processing Natural language understanding is a subtopic of natural language processing in artificial intelligence that deals with machine reading comprehension. Page 9 .e. Sample Input 10 move 9 onto 1 move 8 over 1 move 7 over 1 move 6 over 1 pile 8 over 6 pile 8 over 5 move 2 over 1 move 4 over 9 quit Sample Output 0: 0 1: 1 9 2 4 2: 3: 3 4: 5: 5 8 7 6 6: 7: 8: 9: 6. voice-activation. relatively simple tasks such as short commands issued to robots. ranging from small. for instance text classification for the automatic analysis of emails and their routing to a suitable department in a corporation does not require in depth understanding of the text. Many real world applications fall between the two extremes. text categorization. but is far more complex than the management of simple queries to database tables with fixed schemata Page 10 .The process of disassembling and parsing input is more complex than the reverse process of assembling output in natural language generation because of the occurrence of unknown and unexpected features in the input and the need to determine the appropriate syntactic and semantic schemes to apply to it. The umbrella term "natural language understanding" can be applied to a diverse set of computer applications. factors which are predetermined when outputting language.[dubious – discuss] There is considerable commercial interest in the field because of its application to newsgathering. to highly complex endeavors such as the full comprehension of newspaper articles or poetry passages. archiving and large-scale contentanalysis. Introduction to learning Learning can be described as normally a relatively permanent change that occurs in behavior as a result of experience. Various techniques used in learning The following components are part of any learning problem: Page 11 . it is possible to learn by practicing the skill until the performance converges on the desired model.e. Learning occurs in various regimes. it is possible to learn to open a lock as a result of trial and error. For example. Automatic performance of some skills by the brain points out that the brain is capable of doing things in parallel i. one part is devoted to the skill whilst another part mediates conscious experience. One begins by paying attention to what needs to be done." --Herbert Simon · "Learning is constructing or modifying representations of what is being experienced.. possible to learn how to use a word processor as a result of following particular instructions.. enables a system to do the same task more efficiently the next time. but with more practice. Once the internal model of what ought to happen is set.Unit V 1." --Marvin Minsky 2. There's no decisive definition of learning but here are some that do justice: · "Learning denotes changes in a system that . one will need to monitor only the trickier parts of the performance." --Ryszard Michalski · "Learning is making useful changes in our minds. g.for example. An intelligent tutoring system could try to explain why a student gives some answer in terms of what the student understands and does not understand. or improved speed .Task :-The behavior or task that is being improved Data :. Consider the agent internals of Figure . increasing accuracy in prediction. There are two principles that are at odds in choosing a representation scheme: Page 12 . This internal representation could be the raw experiences themselves. which is deriving consequences of a knowledge base.1..The experiences that are used to improve performance in the task measure of improvement. which is hypothesizing what may be true about a particular case. Figure 1. new skills that were not present initially. it can hypothesize what is happening in the world to explain why the light was not working. The problem of inferring an internal representation based on examples is often called induction and can be contrasted with deduction. about the experiences of the agent) and to create an internal representation (the knowledge base) that is used by the agent as it acts.: Offline and online decomposition of an agent The problem of learning is to take in prior knowledge and data (e. but it is typically a compact representation that summarizes the data. For example.). if an agent were to observe that some light was not working. and abduction(Abduction is a form of reasoning where assumptions are made to explain observations. How the improvement is measured . The most commonly studied learning task is supervised learning: given some input features. The richer the representation scheme. Page 13 . A very rich representation is difficult to learn because it requires a great deal of data. The representations required for intelligence are a compromise between many desiderata (see Section 2). some target features. This is called classification when the target variables are discrete and regression when the target features are continuous. the more difficult it is to learn. and a set of training examples where the input features and the target features are specified. The richer the representation. the more useful it is for subsequent problems solving. and often many different hypotheses are consistent with the data. the representation must be rich enough to express a way to solve the problem. For an agent to learn a way to solve a problem. predict the target features of a new example for which the input features are given. Figure 2: The role of representations in solving problems The ability to learn the representation is one of them. but it is not the only one. Learning techniques face the following issues: Task Virtually any task for which an agent can get data or experiences can be learned. In supervised learning. where the feedback in terms of rewards and punishments occurs after a sequence of actions.. Supervised learning of actions occurs when the agent is given immediate feedback about the value of each action.Other learning tasks include learning classifications when the examples are not already classified (unsupervised learning). Much of machine learning is studied in the context of particular representations (e. decision trees. or case bases). all of the training examples are available to an agent before it needs to act. Feedback Learning tasks can be characterized by the feedback given to the learner. Supervised classification occurs when a trainer provides the classification for each example. This leads to the creditassignment problem of determining which actions were responsible for the rewards or punishments. such as in reinforcement learning. what has to be learned is specified for each example. Representation For an agent to use its experiences. training examples arrive as the agent is acting. learning what to do based on rewards and punishments (reinforcement learning). In online learning. It is possible that it can learn what actions to perform without actually determining which consequences of the actions are responsible for rewards. The robot then must either learn what it is being rewarded for or learn which actions are preferred in which situations. Page 14 . a user could give rewards to the delivery robot without telling it exactly what it is being rewarded for. neural networks. the experiences must affect the agent's internal representation. and learning richer representations such as logic programs (inductive logic programming) or Bayesian networks. Unsupervised learning occurs when no classifications are given and the learner must discover categories and regularities in the data. For example. learning to reason faster (analytic learning). Feedback often falls between these extremes. This chapter presents some standard representations to show the common features behind learning.g. Online and offline In offline learning. The measure is usually not how well the agent performs on the training experiences. Agent P claims that all of the negative examples seen were the only negative examples and that every other instance is positive. A representation is built using the training set. A standard way to measure success is to divide the examples into a training set and a test set. Agent N claims that the positive examples in the training set were the only positive examples and that every other instance is negative. but how well the agent performs for new experiences. To know whether an agent has learned.An agent that learns online requires some representation of its previously seen examples before it has seen all of its examples. Typically. Consider the agents N and P defined earlier. In classification. Thus. the real measure is its performance on some future task. and then the predictive accuracy is measured on the test set. the learner must generalize: go beyond the specific given examples to classify unseen examples. being able to correctly classify all training examples is not the problem. Bias The tendency to prefer one hypothesis over another is called a bias. For example.both N and P Page 15 . Suppose that there were two agents P and N. Both of these agents correctly classify every example in the training set but disagree on every other example. Of course. Measuring success Learning is defined in terms of improving performance based on some measure. an agent never sees all of its examples. the agent must update its representation. Success in learning should not be judged on correctly classifying the training set but on being able to correctly classify unseen examples. this is only an approximation of what is wanted. we must define a measure of success. In active learning. As new examples are observed. Saying that a hypothesis is better than N's or P's hypothesis is not something that is obtained from the data . consider the problem of predicting a Boolean feature based on a set of examples. Active learning is a form of online learning in which the agent acts to acquire useful examples from which to learn. the agent reasons about which examples would be useful to learn from and acts to collect these examples. Noise In most real-world situations. an agent requires a bias. trying to find the representation or representations that best fits the data given the bias. the Ptolemaic system and heliocentric system of Copernicus made detailed models of the movement of solar system in terms of epicycles (cycles within cycles). The definition of the learning algorithm then becomes one of defining the search space. Interpolation and extrapolation For cases in which there is a natural interpretation of "between.but is something external to the data. the evaluation function. an agent will not be able to make any predictions on unseen examples. interpolation involves making a prediction between cases for which there are data. Unfortunately.accurately predict all of the data given . and often there are examples with missing features. the data are not perfect. Learning is a search through the space of possible representations. To have any inductive process make predictions on unseen data. Learning as search Given a representation and a bias." such as where the prediction is about time or space. and the search method. Extrapolation involves making a prediction that goes beyond the seen examples. there are inadequate features (the features given do not predict the classification). For example. the search spaces are typically prohibitively large for systematic search. What constitutes a good bias is an empirical question about which biases work best in practice. One of the important properties of a learning algorithm is its ability to handle noisy data in all of its forms. Noise exists in the data (some of the features have been assigned the wrong value). and. Without a bias. we do not imagine that either P's or N's biases work well in practice. Nearly all of the search techniques used in machine learning can be seen as forms of local search through a space of representations. The parameters for the models could be made to fit the data very well and they were Page 16 . the agent will not be able to resolve this disagreement. in ancient astronomy. The hypotheses adopted by P and N disagree on all further examples. Extrapolation is usually much more inaccurate than interpolation. if a learning agent cannot choose some hypotheses as better. except for the simplest of examples. the problem of learning can be reduced to one of search. however. as new methods have been created have shown brief promise. 4. How the brain enables human beings to think has remained a mystery until the present day. 3. and have suffered from some setback.very good at interpolation. but the learned model is used for extrapolation. the models were very poor at extrapolation. Neural Networks can be loosely separated into Neural Models. and it would be very profitable to be able to do so. Page 17 . Introduction to neural networks Artificial Intelligence has had its fair share from the field of neuroscience. It is very difficult to predict the price that a stock will be tomorrow. Network Models and Learning Rules. the earliest mathematical models of the Neuron pre-date Mcullock and Pitts who developed the first Network models to explain how the signals passed from one neuron to another within the network. Neuroscience is the study of nervous system. Weiners work allowed Mculloch and Pitts to describe how these different connection types would affect the operation of the network. When you hear of a network being described as a feed forward or feedback network. But significant leaps and bounds in the field have enabled scientists to come close to the nature of thought processes inside a brain. particularly the brain. it is often easy to predict a stock price on a certain day given data about the prices on the days before and the days after that day. As another example. Neural Network research has gone through a number of these lulls. This is particularly useful in applications where the complexity of the data or task makes the design of such a function by hand impractical. Applications of neural networks The utility of artificial neural network models lies in the fact that they can be used to infer a function from observations. they are describing how the network connects neurons in one layer to neurons in the next. However scientists have always come back to the technology because it is a real attempt to model neural mechanisms despite the hype. have been over-promoted. An agent must be careful if its test cases mostly involve interpolating between data points. clustering. financial applications (e. After training. Colorectal cancer has also been predicted using the neural networks. Real-life applications The tasks artificial neural networks are applied to tend to fall within the following broad categories: Function approximation. prosthesis. An ANN based hybrid lung cancer detection system named HLND improves the accuracy of diagnosis and the speed of lung cancer radiology. "KDD"). poker). Artificial neural networks have also been used to diagnose several cancers. The diagnoses can be used to make specific models taken from a large group of patients compared to information of one given patient. natural resources management). blind source separation and compression. or regression analysis. The models do not depend on assumptions about correlations of different variables. Robotics. process control. including time series prediction. chess. including filtering. including pattern and sequence recognition. speech. Neural networks and neuroscience Page 18 . Data processing. Application areas include the system identification and control (vehicle control.[43] b). fitness approximation and modeling. the networks could predict multiple patient outcomes from unrelated institutions. automated trading systems).a). including directing manipulators. handwritten text recognition).[41] game-playing and decision making (backgammon. face identification. novelty detection and sequential decision making. data mining (or knowledge discovery in databases. sequence recognition (gesture. pattern recognition (radar systems. Control.g. including Computer numerical control. visualization and e-mail spam filtering. medical diagnosis. Classification. Neural networks could predict the outcome for a patient with colorectal cancer with more accuracy than the current clinical methods. quantum chemistry. object recognition and more).[42] These networks have also been used to diagnose prostate cancer. it requires the machine to appear as intelligent as a human being. The problem is considered to be among the hardest in all of AI research because the breadth and detail of commonsense knowledge is enormous. and this is impossible in general unless the machine is familiar with all the same concepts that an ordinary person is familiar with. the following: An ontology of classes and individuals Page 19 . To gain this understanding. the field is closely related to cognitive and behavioral modeling. The aim of the field is to create models of biological neural systems in order to understand how biological systems work. neuroscientists strive to make a link between observed biological processes (data). object recognition. To do these tasks perfectly. the machine simply has to know what the text is talking about or what objects it may be looking at. Any task that requires commonsense knowledge is considered AI-complete: to be done as well as a human being does it. Common sense In artificial intelligence research. Such a database is a type of ontology of which the most general are called upper ontologies.Theoretical and computational neuroscience is the field concerned with the theoretical analysis and the computational modeling of biological neural systems. Information in a commonsense knowledge base may include. These tasks include machine translation. 5. Since neural systems are intimately related to cognitive processes and behavior. commonsense knowledge is the collection of facts and information that an ordinary person is expected to know. but is not limited to. represented in a way that it is available to artificial intelligence programs that use natural language or make inferences about the ordinary world. The commonsense knowledge problem is the ongoing project in the field of knowledge representation (a sub-field of artificial intelligence) to create a commonsense knowledge base: a database containing all the general knowledge that most people possess. text mining and many others. biologically plausible mechanisms for neural processing and learning (biological neural network models) and theory (statistical learning theory and information theory). Inferences are classified as either deductive or inductive. Reasoning To reason is to draw inferences appropriate to the situation. “Previous accidents of this sort were caused by instrument failure.” and of the latter. therefore this Page 20 . “Fred must be in either the museum or the café. Parts and materials of objects Properties of objects (such as color and size) Functions and uses of objects Locations of objects and layouts of locations Locations of actions and events Durations of actions and events Preconditions of actions and events Effects (postconditions) of actions and events Subjects and objects of actions Behaviors of devices Stereotypical situations or scripts Human goals and needs Emotions Plans and strategies Story themes Contexts 6. He is not in the café. therefore he is in the museum. An example of the former is. With an expert system the goal was to specify the rules in a format that was intuitive and easily Page 21 . Inductive reasoning is common in science. The first expert systems were created in the 1970s and then proliferated in the 1980s. an expert system is a computer system that emulates the decision-making ability of a human expert. Expert systems are designed to solve complex problems by reasoning about knowledge. There has been considerable success in programming computers to draw inferences. Deductive reasoning is common in mathematics and logic. where elaborate structures of irrefutable theorems are built up from a small set of basic axioms and rules. Expert systems were among the first truly successful forms of AI software. it involves drawing inferences relevant to the solution of the particular task or situation. especially deductive inferences. Inference engines can also include explanation and debugging capabilities. In a traditional computer program the logic is embedded in code that can typically only be reviewed by an IT specialist. Advantages The goal of knowledge-based systems is to make the critical information required for the system to work explicit rather than implicit. However. Some example of expert systems In artificial intelligence.accident was caused by instrument failure. The knowledge base represents facts and rules. The inference engine applies the rules to the known facts to deduce new facts. where data are collected and tentative models are developed to describe and predict future behaviour—until the appearance of anomalous data forces the model to be revised. 7. a). represented primarily as if–then rules rather than through conventional procedural code. whereas in the inductive case the truth of the premise lends support to the conclusion without giving absolute assurance.” The most significant difference between these forms of reasoning is that in the deductive case the truth of the premises guarantees the truth of the conclusion. true reasoning involves more than just drawing inferences. An expert system is divided into two sub-systems: the inference engine and the knowledge base. These problems were essentially the same as those of any other large system: integration. reviewed. debugging. and even edited by domain experts rather than IT experts. simply invoke the inference engine. when looking at the life-cycle of expert systems in actual use other problems seem at least as critical as knowledge acquisition. However. and performance. Ease of maintenance is the most obvious benefit. Page 22 . As a result of this problem a great deal of research in the later years of expert systems was focused on tools for knowledge acquisition. Performance was especially problematic because early expert systems were built using tools such as Lisp. Essentially. The benefits of this explicit knowledge representation were rapid development and ease of maintenance. to help automate the process of designing. such as C. System and database integration were difficult for early expert systems because the tools were mostly in languages and platforms that were neither familiar to nor welcomed in most corporate IT environments – programming languages such as Lisp and Prolog and hardware platforms such as Lisp Machines and personal computers. Interpreting provided an extremely powerful development environment but with the drawback that it was virtually impossible to match the efficiency of the fastest compiled languages of the time. b). and maintaining rules defined by experts. This was achieved in two ways. With an expert system shell it was possible to enter a few rules and have a prototype developed in days rather than the months or year typically associated with complex IT projects. First. by removing the need to write conventional code many of the normal problems that can be caused by even small changes to a system could be avoided with expert systems. Obtaining the time of domain experts for any software application is always difficult but for expert systems it was especially difficult because the experts were by definition highly valued and in constant demand by the organization. access to large databases. This also was a reason for the second benefit: rapid prototyping.understood. which executed interpreted rather than compiled code. the logical flow of the program (at least at the highest level) was simply a given for the system. Disadvantages The most common disadvantage cited for expert systems in the academic literature is the knowledge acquisition problem. New Delhi. and porting to more standard platforms. “Artificial Intelligence”. Page 23 . Rich E and Knight K. References:1. integration with large database systems.As a result a great deal of effort in the later stages of expert system tool development was focused on integration with legacy environments such as COBOL. TMH. These issues were resolved primarily by the client-server paradigm shift as PCs were gradually accepted in the IT world as a legitimate platform for serious business system development and as affordable minicomputer servers provided the processing power needed for AI applications. 2. Berlin.. “Principles of Artificial Intelligence”. Springer Verlag. Page 24 .J. Nelsson N.