- Slides: 28
Managing Knowledge Chapter 11
Knowledge Management • Knowledge management is the process of systematically and actively managing and leveraging the stores of knowledge in the organization. • The major information systems that support knowledge management are: • Office automation systems (OAS): Office automation systems are systems that increase the productivity of office workers by document management and creation and communications. • Knowledge work systems (KWS): Knowledge work systems are specialized systems that support knowledge workers in their fields. • Group collaboration systems and artificial intelligence systems (AI): Group collaboration allows, primarily knowledge workers, to share documents, schedule meetings, develop shared databases, and send e-mail. Groupware is the primary group collaboration software. Artificial intelligence is the development of systems that behave as humans • Note that group collaboration systems and artificial intelligence systems are usually included in decision-support systems.
Relationship between information work and productivity in contemporary organizations • Information work => Work that primarily consists of creating or processing information. • Information work plays the dominant role in advanced industrial societies. • In 1989, over 70 percent of all capital investment in the U. S. was in the area of information technology with 70 percent of that going directly into offices. • Information technology has increased productivity in manufacturing. • However, the extent to which computers have enhanced the productivity of information workers remains under debate.
The roles of the office in organizations • Offices are now viewed as one of the most important work sites for professional knowledge and data workers. • They involve the work of professional, managerial, sales, and clerical employees. • They perform three organizational roles: Ø Coordinate and manage the work of local professional and information workers within the organization. Ø Link the work being performed across all levels and functions throughout the organization. Ø Couple the organization with the external environment.
The roles of the office in organizations The five major office activities are: Ø Managing documents. Ø Scheduling individuals and groups. Ø Communicating with individuals and groups. Ø Managing data on individuals and groups. Ø Managing projects.
Generic requirements of knowledge work systems • They require considerably more computing power than most systems. For example, graphics requires about six times more power to represent and process than do character based documents; document based functions, such as legal research, require a great deal of power to scan thousands of documents; and design and engineering systems such as CAD require a great deal of power to perform simulations. • They must have a user-friendly interface because such interfaces are much less time consuming than traditional interfaces, and knowledge workers' time is very costly. • They often require the use of workstations because workstations contain the power, speed, telecommunications, and user interfaces these systems need.
Systems support knowledge work • Computer-aided design (CAD) • Virtual reality, • Investment workstations.
CAD systems • CAD systems automate the creation and revision of designs using computers and sophisticated graphics software. Using CAD, business can benefit in many ways, including: Ø The design produced can be more sophisticated and functional than manual designs. Ø Design time can be significantly reduced, thus reducing cost and increasing competitiveness. Ø Expensive engineering changes during manufacturing can be reduced. Ø Using the modeling function, designers need to make far fewer prototypes of the product. Ø CAD software can be used to design the tooling and the manufacturing process from the design, resulting in a manufacturing process with far fewer problems.
Virtual reality • Virtual reality, like CAD, has modeling and simulation capabilities. • The difference is that virtual reality uses interactive graphics software to create computer-generated simulations that are so close to reality that users believe they are participating in a "real" world. • It is interactive in such a way that users actually feel immersed in the "world" the computer creates.
Investment Workstations • Investment workstations are computer systems which access and manipulate massive amounts of financial data to manage financial trades and portfolio management. • The people who develop the sophisticated mathematical models on these systems are called rocket scientists. • Sometimes it is applied to the people who use these workstations.
How does groupware support information work? • Groupware provides services and functions that support the collaborative activities of work groups. • Its goal is to improve the effectiveness of the work group by providing electronic links between members of that group. • The following slide describes the work of a group and the capabilities of groupware to support that work.
Groupware capabilities and the Internet capabilities for collaborative work. Group functions • Schedule meetings • Hold meetings • Communicate with each other Groupware support Electronic calendaring Electronic meeting software to increase productivity in meetings and to make possible a meeting of participants in scattered locations Electronic mail, compound documents communication databases, memo distribution and classification software
Groupware capabilities and the Internet capabilities for collaborative work. Group functions • Collaborate to develop ideas • Share preparation of documents Groupware support Communications databases, word processors with group editing and document management facilities Communication and collaboration capabilities already listed All the above listed facilities • Share knowledge and work information • • The Internet allows meetings, electronic mail, communication, and collaboration
Artificial Intelligence • Artificial intelligence is commonly defined as the effort to develop computer-based systems (hardware and software) that behave as humans. • Such systems would be able to learn natural languages, accomplish coordinated physical tasks (robotics), and develop and utilize a perceptual apparatus that informs their physical behavior and language (visual and oral perception systems), and emulate human expertise and decision making (expert systems). • Such systems would also exhibit logic, reasoning, intuition, and common sense. • Actually, however, successful artificial intelligence systems are neither artificial nor intelligent.
AI FAMILY ARTIFICIAL INTELLIGENCE NATURAL LANGUAGE ROBOTICS PERCEPTIVE SYSTEMS EXPERT SYSTEMS INTELLIGENT MACHINES
Why do we use artificial intelligence in business? • Preserve or capture expertise that might be lost through retirement, resignation, or death. • Store information in an active form—as a form of, say, an electronic manual or tutor. • Create a mechanism that is not subject to human feelings such as fatigue or emotional involvement. • Eliminate routine and unsatisfying jobs held by people • Enhance the organization’s knowledge base by suggesting solutions to problems that are too massive and complex for human beings to solve in a short period of time.
Difference between artificial intelligence and natural or human intelligence • Artificial intelligence is an effort to fashion computer systems that behave like human beings. • Such systems would have the ability to learn natural languages, accomplish coordinated physical tasks, utilize complex visual and oral perceptual systems, and use complex logic, reasoning, and intuition. • To date, the AI systems that have been developed do not exhibit all of the qualities of human intelligence, and in no way can be considered fully "intelligent". • AI systems are based on human expertise, but they can use only very limited reasoning patterns and perform very limited tasks. • Example might be self-adjusting thermostats or autopilots or a robot that welds an automobile door. They also can't learn on their own and must be programmed by a human being to follow certain rules or produce certain solutions.
An expert system • An expert system is a computer system programmed to use knowledge in the form of rules or frames to solve problems by capturing the expertise of a human expert in limited domains of knowledge and experience. • Expert systems can interact with humans and consider multiple hypotheses simultaneously. • The knowledge incorporated in an expert system must be supplied by a human expert. • Such systems can't really think like humans and can merely assist decision making by asking relevant questions and explaining the reasons for adopting certain actions.
An expert system • • Some of the uses of expert systems are displayed in slide 15. The major use is to capture the knowledge of the organization. How do we repair boilers? How do we configure computers? The kind of hidden, expert knowledge that expert systems capture is usually not found in the database
Role of rule base, frames, semantic nets, inference engine in expert systems • Rule base is a list of if-then rules which are interconnected. • A semantic net represents knowledge and expertise that is composed of easily identified chunks or objects of hierarchically interrelated characteristics. Semantic nets are more efficient than rules if the data can be represented this way. • Frames also organize knowledge into chunks, but chunks that are related by shared characteristics. For example, cars and tanks are separate frames. • Inference engine is the software that embodies the strategy used to search through the rule base
What is case-based reasoning? • CBR uses descriptions of past experiences of human specialists, representing them as “cases” and storing them in a database for later retrieval when the user encounters a new case with similar parameters. • The system searches for stored cases similar to the new one, locates the closest fit, and offers the solution to the old case for use with the new case. • If the new case fits the solution, it is added to the case database. • If not, the case will be added with a new solution or explanations as to why the solutions didn’t work. • CBRs differ from expert systems in that they capture the knowledge of the organization rather than a single expert, and the knowledge is captured as cases rather than if-then rules.
Problems and limitations of expert systems • Expert systems are limited to certain problems, working successfully only with problems of classification that have few alternative outcomes. In addition, the outcomes must be known in advance. • The knowledge base is fragile and brittle because these systems rely upon IF-THEN representation. Such representation exists primarily in textbooks and cannot be used for deep causal models or temporal trends. They cannot represent knowledge that is essentially intuitive. • They are expensive to teach and have no ability to learn over time. Therefore, keeping them up to date in fast-moving fields such as medicine and computer sciences is a critical problem.
Neural network and its applications • Neural networks are usually physical devices (although they can be simulated with software) which emulate the physiology of animal brains. • The resistors in the circuits are variable, and can be used to "teach" the network. • When the network makes a mistake, i. e. , chooses the wrong pathway through the network and arrives at a false conclusion, resistance can be raised on some circuits, forcing other neurons to fire. • Used after a false conclusion, intervention teaches the machine the correct response. • If this learning process continues for thousands of cycles, the machine "learns" the correct response. • The simple neurons or switches are highly interconnected and operate in parallel so they can all work simultaneously on parts of a problem. • Neural networks are very different from expert systems where human expertise has to be modeled with rules and frames. • In neural networks, the physical machine emulates a human brain and can be taught from experience.
Neural network and its applications • An expert system is highly specific to a given problem and cannot be retrained. • Neural networks do not model human intelligence or aim to solve specific problems. • Instead of putting human expertise into programs, neural network designers put intelligence into the hardware in the form of a generalized capacity to learn. • Neural networks can solve entire classes of problems. • The neural network can be easily modified. • Neural networks therefore promise a substantial savings in development cost and time. • They allow much greater generality and more closely approximate what we consider intelligence. They are especially useful for visual pattern recognition problems.
Fuzzy logic & its applications • Fuzzy logic is a rule-based AI technology that tolerates imprecision, even using that imprecision to solve problems we could not solve before. • Fuzzy logic creates rules that use approximate or subjective values and incomplete or ambiguous data. • Fuzzy logic is closer to the way people actually think than traditional IF-THEN rules. • For example, if we all agree that 120 degrees is hot and -40 degrees is cold, is 75 degrees hot, warm, comfortable, or cool? The answer is fuzzy at best and cannot be programmed in an IFTHEN manner.
Genetic algorithms & its applications • Genetic algorithms (adaptive computation) are a variety of problem-solving methods that are conceptually based on the method that living organisms use to adapt to their environment-the process of evolution. • Genetic algorithms control the generation, variation, adaptation, and selection of possible problem solutions using geneticallybased processes. • As solutions alter and combine, the worst ones are discarded and the better ones survive to go on and produce even better solutions.
Genetic algorithms & its applications • They are particularly suited to the areas of optimization, product design, and the monitoring of industrial systems. • Organizations can use them to minimize costs and maximize profits, and both to schedule and to use resources efficiently. • They are ideal when problems are dynamic and complex and involve hundreds of variables or formulas. • For example, General Electric used a genetic algorithm to help them design a jet turbine aircraft engine that required the use of about 100 variables and 50 constraint equations.
Intelligent agent & applications • Intelligent agents are software programs that use built-in or learned knowledge base to carry out specific, repetitive tasks for a user, business process, or as part of the software application. • By watching the user of a program or system, an intelligent agent may customize the software system for the user to meet the user’s needs, reducing software support costs. • Intelligent agents can be used as so-called wizards to help users do or learn how to do a task. • Intelligent agents can be used to carry out “smart” searches of the database, data warehouse, or the Internet, reducing search costs and avoiding the problems of misdirected searches.