Printer Friendly

Toward the development of a second generation computerized job-matching system for persons with disabilities: a conceptual framework.

The rapid development of integrated circuit technology has enabled computer manufacturers to condense the main component of the computer, the central processing unit (cpu), into a very inexpensive silicon chip smaller than a postage stamp (Restak, 1980). This technological breakthrough has resulted in a proliferation of low-cost but powerful mini-computer and microcomputer systems that can perform complex data processing and computing functions which once required costly large mainframe computer systems.

The same technology has also facilitated the development of super performance mainframe computers. These computers have made possible telecommunications, computer networks, office automation, and distributed processing systems (Synnott & Gruber, 1981). The widespread availability of computer telecommunication technology to the general public is changing the way American workers prepare for and perform their jobs. Naisbitt (1982) stated that the United States is moving rapidly from a manufacture-based to an information-based and service-based economy.

The field of rehabilitation is not, nor should it be, insulated from these developments. According to a recent national survey of the Stout Vocational Rehabilitation Institute (Menz & Bordieri, 1986), training in computers in information management is one of the categories ranked by rehabilitation facilities administrators and managers as having the greatest need for training. In response to this, rehabilitation professionals must learn to take advantage of, and adapt to, this new information technology. For the past few years, numerous descriptions and discussions of computer applications were published in professional journals in areas such as rehabilitation counseling, vocational evaluation, work adjustment, and job placement (Buckhead & Sampson, 1985; Chan, McCollum, & Parker, 1985; Chan, Matkin, Parker, & McCollum, 1988; Chan & Questad, 1981; Crimado & Goodley, 1985; McCollum & Chan, 1985).

Computerized Job Matching-Systems

One of the primary uses of microcomputer technology in rehabilitation has been in the area of "computerized job-matching" For example, in a recent review of computerized job-matching software, Botterbusch (1986) identified more than 15 major commercial software packages available to rehabilitation professionals. The proliferation of these software programs can be partially attributed to the long standing emphasis of job placement in rehabilitation.

According to Dunn (1974), the quality of job placement of each client is determined early in the exploration and planning phases of the vocational rehabilitation process. To facilitate vocational decision-making, a rehabilitation client must be exposed systematically to the world of work, develop insights for skills, abilities, interest and physical limitations, and be sensitive to labor market constraints. This, in turn, requires the skilled rehabilitation counselor to become knowledgeable about (a) the vocational implications of a variety of disabilities, (b) the work demands and requirements of different occupations, (c) the job trends and training opportunities in the local and national labor economy, and (d) the development and availability of job accommodation methods and assistive devices for people with disabilities (Roessler & Rubin, 1982).

The time and energy required for manually keeping up with the sophisticated and ever-changing world of work information is overwhelming. Computerized job-matching system were designed, to a large extent, to alleviate the counselor from those most time consuming tasks of collecting and retrieving up-to-date occupational data. As a result, these system have gained wide acceptance among professionals in the rehabilitation community (Chan McCollum, & Pool, 1985).

Specifically, a computerized job-matching system is a computer software program that is capable of systematically comparing an individual's employability profile (vocational strengths and weaknesses) against the requirements of a job, a cluster of closely related jobs, or a training program in a "world of work" data base (Botterbush, 1986). The strengths and limitations of rehabilitation clients are usually determined by clinical interview, counselling, psychological assessment, and vocational evaluation. The data bases used in these system are usually extracted from the fourth edition of the Dictionary of Occupational Titles (DOT).

Methodological Considerations for Developing Computerized Job Matching Systems

Although the value of computerized job-matching system to rehabilitation professionals is well established, these DOT based system are not without major problems (Botterbusch, 1986; Chan, Parker, Dial, Lam, & Chan, 1989). The accuracy of the 12,000 jobs describes in the DOT is frequently challenged by researchers (Bose, Grzesik, Geist, & Bryant, 1986; Miller, Trieman, Cain, & Roos, 1980). Recently, Bose et al., 1986), after analyzing representative sample of occupations in the Chicago area, concluded that the DOT definitions for many occupations are no longer valid due to rapid technological changes.

The closed system of the current computerized job-matching programs may be substantially unfair to the client. In a traditional manual search, the couselor, based on his or her clinical experience, will have the flexibility of adjusting th performance scores of a client to include jobs that may be appropriate with modification. Whereas in the typical present day computer search, the computer tends to utilize cutoff scores to select occupations with very little, if any, flexibility. Basically, the machine takes control of the job-matching process and the benefit of human judgement is not optimally solicited. It is easy to see how this type of job machine approach could inadvertently rule out many jobs that might be appropriate.

Another set of problems related to current computerized job matching system was addressed by Botterbusch (1986). He noted that the appropriateness of a computer job search output depends largely on the liability of the input data and the validity of the internal (computer) job search algorithm. In appealing to a broad base of customers, programs usually opt for flexibility of input data at the expense of accuracy. By design, these computer programs may be subjected to multiple sources of uncotrolled error., e.g., there are significant differences between data obtained as the result of a week of vocational evaluation and data obtained as the result of estimates based on little more than guessing. Even the substitution of test results obtained from one particular test (e.g., the GATB) with another similar test (e.g., the DAT) may confound the outcome of the job search results (Botterbusch, 1986). Bottherbusch strongly urged rehabilitation professionals to exercise caution in using occupational information generated form these computer systems.

According to the Minnesota Theory of Work Adjustment, it is important to consider both the job Satisfaction (person-job environment fit) and Job Satisfactoriness (person-job requirement fit) factors in vocational evaluation and counselling (Dawis, England, & Lofquist, 1964). Unfortunately, most DOT based job matching system currently on the market tend to emphasize job matching on the basis of person job requirement fit (i.e., physical capacities, skills, abilities, and aptitudes) alone. As a result, the person job environment fit factor is not incorporated into the computerized match. An example might be the case of a truck driver with a high school diploma. This person may have the "transferable job skills" to work as an office clerk. However, personality, work style, and work behavior may be such that the may not fit into the white collar (clerk) work culture (Chan et al., 1989). Regardless of the output of a computer matching system, the ultimate responsibility for job matching lies with the counselor. How much more efficient might the counselor be if a computerized system could help with the equally important consideration of whether the person fits in with his or her work environment.

Perhaps the most significant problem associated with current computerized job matching system is their inherent flaw in system design. Most job matching systemn utilize highly structured conventional database models in their design. As a result, the job search (query) process of these system tends to be very rigid and are quick to discard information that cannot be conveniently represented by some kind of rating category (e.g., Excellent 1, 2, 3, 4, 5 Poor, or in dichotomous format such as yes/no, on/off, or qualified /not qualified, etc.) (Career Evaluation System, 1985). This highly structured search method imposes a straight-line logic in vocational decision-making. It assumes that all jobs related variables as defined in the DOT have almost perfect reliability and validity derived from perfect job analysis data and assessed by perfect tests (Career Evaluation System, 1985). The clients are deemed appropriate or inappropriate for certain occupations on the basis of meeting rigid job attribute cut-off requirements defined by the DOT.

Job matching and placements is one of the most vital parts of the rehabilitation process. We must try to improve the current system in order to provide better services to people with disabilities. Because of the impression of the DOT data and the great variation in adaptability among individuals, it is important to develop a job matching system that is more flexible than conventional system. One possible strategy is the development of a microcomputer-based "expert" system comprised of the collective wisdom, knowledge and experience of a panel of rehabilitation experts who have extensive succesful experience placing clients. An "expert" system is usually defined as a computer system that can perform at, or near, the level of a human expert. The expert job matching system could serve as a desktop consultant to assist rehabilitation professionals with less experience and knowledge about job placement become more effective in this endeavor. Expert system have been highly succesful in training physicians in diagnosing diseases (e.g., the MYCIN System developed by Stanford University) (Clancey & Shortliffe, 1984). This approach would require the development of a knowledge-based system capable of "reasoning" and making decisions as would a vocational expert. In addition, the system must be capable of effecting self correction (learning) based on cumulative job matching experience and feedback from the users. Furthermore, it must be capable of handling query on the basis of vague, uncertain, and incomplete information as is commonly found in real life data. More importantly, when certain occupations are identified as appropriate for an individual with a disability, this ideal job matching system must be able to: explain to the counselor why they are appropriate in terms of logical decision rules and succesful experience of similar case studies available in the system, make suggestion as to whether job accommodation is needed for any of these occupations, and indicate what kinds of support services may be needed for these occupations.

A Second Generation Computerized Job-Matching System

A second generation computerized job-matching system should utilize a knowledge-based expert system. This will more accurately reflect the job searching process proposed by the Minnesota Theory of Work Adjustment by carefully considering both the Job Satisfaction and Job Satisfactoriness factors in the job searching process. This specific job searching process must be developed through the use of outcome prediction research (quantitative data) and the combined clinical experience of vocational experts (qualitative data).

Recent advancement in artificial intelligence, generalized database techniques, and information retrieval system using "fuzzy set" theory has made it possible to consider the development of expert (intelligent) computerized job matching systems. A primary advantage of this ideal computerized job matching system will be the ability to appropriate handle problems of the uncertainty, incompleteness, and the vagueness of job-matching information commonly found in real life situations.

Conceptual Framework

At this point in the discussion, the reader will encounter a number of terms with which he or she may not be familiar. Please consult the brief Glossary from Harmon and King (1985) provided in the appendix I.

According to Bonissone and Tong (1985), the objective of an "expert" system is to capture the knowledge of an expert in a modular expandable structure, and transfer it to other users in the same problem domain. To accomplish this goal, the conceptualized "expert" system borrows from the current literature on artificial intelligence and begins with three basic divisions, each comprised of specialized elements (see fig. 1)

To accomplish this goal, it is necessary to define and address six important components. These components are: Inference Engine, Knowledge Acquisition Subsystem, Knowledge Base Subsystem, Explanation Subsystem, User Interface, and Uncertainty Handling Subsystem. All of these components are basic elements in the current literature dealing with artificial intelligence (Harmon, P. & King, D., 1985).

Inference Engine.

The Inference Engine is the heart of the job matching system conceptualized. It contains the inference and control strategies of the expert system. It also consists of the computer instructions necessary to facilitate the interactions among the users - the Explanation subsystem, and the Knowledge Base sybsystem. Usually, the Inference Engine decides the inference mechanism, such as forward chaining, backward chaining, etc. This mechanism is included in the expert system tools or shells. For example, PROLOG employs backward chaining, CLIPS employs forward chaining, and MYCIN uses modus ponens and backward chaining Harmon, P. & King, D., 1985).

Knowledge Acquisition.

In order to construct an expert system which can process information like an expert, the first area to develop is the Knowledge Acquisition subsystem. Knowledge in an expert system may originate from many sources, such as textbooks, reports, data bases, case studies, empirical data, and personal experiences. The dominant source of knowledge in expert systems is the domain expert. The knowledge engineer usually obtains this knowledge through direct interaction with the expert. This process will involve: (1) defining the problem, (2) identifying requirements, (3) collecting information, (4) generating rules, and (5) checking conflicting information. This ideal computerized job matching system, as discussed, will utilize both prediction research data (quantitative data) and clinical experience of vocational experts (qualitative data). The checklist method, a straight-forward approach which collects all the pertinent information will be utilized.

Knowledge Base.

The Knowledge Base subsystem in this expert system is divided into two additional components: The Rule Base component and the Fact Base component. Through the process of knowledge engineering, experienced vocational experts can be utilized to develop a Rule Base component of clinical decision rules for matching people with disabilities to jobs. This process involves: (1) selecting numerous cases representing clients with a wide range of disability types, severity, and employability, (2) providing the vocational experts with the case history, interview information, assessment data, and behavioral observation data for each client under study, and (3) understanding how vocational experts utilize case information to determine the appropriate jobs. By using this process, one can determine the relative importance of different job attributes for different levels of occupations (e.g., the cognitive factor may be more important for professional occupations while psychomotor functioning may be more important for unskilled occupations) and decide upon the priority for which job attributes should be evaluated in the job matching process. From this knowledge engineering process, a comprehensive clinical decision-making Rule Base component can be achieved. Similarly, a prediction (quantitative) decision-making Rule Base component can be developed by conducting a thorough study of the statistical prediction literature. One will need to identify statistical formulae useful for predicting vocational functioning levels, determine the most reliable and valid functional assessments, and the best psychometric measures or work samples that are commonly used in rehabilitation outcome prediction. One will also need to further investigate the interrelationships among job attributes defined in the DOT and the vocational literature. For example, Bolton (1985) has successfully demonstrated that he can predict, quite accurately, Holland's occupational types (R-I-A-S-E-c) from personality measures (i.e., the 16 Personality Factor Questionnaire)- Holland and his associates have been able to predict DOT codes from Holland's R-I-A-S-E-C codes (Holland, 1985). Empirical studies such as these will allow estimates of important missing job-matching variables from their known relationships with other job attribute variables. The Rule Base component is also the technical area for handling ambiguous information and is the working area where "fuzzy sets," for dealing with uncertainty, are used to manipulate vague, incomplete or uncertain data.

The second major component of the Knowledge Base Subsystem is the Fact Base component. The Fact Base component will consist of the job attributes information. In this suggested model it should consist of DOT information such as General Education Development [GED], Special Vocational Preparation [SVP], Physical Demands, Aptitudes, Environmental Condition. The Fact Base component will also consist of additional person-environment fit information obtained from the Dictionary of Holland Occupational Codes and other appropriate sources such as the Strong Campbell Interest Inventory Manual, Career Assessment Inventory Manual, and the Minnesota Importance Questionnaire Manual.

Explanation Subsystem.

The Knowledge Acquisition Subsystem, as developed through the knowledge engineering process, will also provide the necessary information to build the Explanation Subsystem. The Explanation Subsystem serves two major functions: (1) as a user friendly system, the Explanation Subsystem will allow the construction of extensive HELP functions in the User Interface end of the system. That is, if the user has any technical problem with the computer or conceptual problem about the job matching process, the Explanation Subsystem will be able to dialogue with the user and provide step-by-step instructions throughout the session. The Explanation Subsystem will also allow provision of detailed explanation (interpretative statements) as to why certain occupations are selected for a given client. If assistive devices are needed or accommodations are required on the job, the computer will also provide the necessary resources information.

User Interface.

The User Interface of the system is where the user (the counselor or evaluator) can interact in simple English for information and answers in order to find the best fit occupations for clients. The output results, a list of matched jobs is also conveyed through the User Interface. The user can also obtain the reasons why these jobs are chosen for their clients through the User Interface.

Working memory. The Working Memory is a technical area for storing short term data with respect to client information. It will process supplemental data from the user in a form which is easy to read and understand. The Working Memory will be comprised of all of the attribute-value relationships that are established while the consultation is in progress. After each job-matching process, the user will be asked about the temporary rules which are created for a particular case. If the user thinks some rules of one case can also apply to other cases, the system will add these rules into the Rule Base component. This process will include a consistency check which makes sure the new rules have no conflict with existing rules.

Figure 2 presents a relatively complete diagram of the architecture of the Knowledge-Based Computer System.

Dealing with Uncertainty with the Uncertainty Handling Subsystem

In most expert systems the degree of implication is artificially expressed as a scalar value on an interval scale (certainty factor). A more natural way to express such a degree of implication would be achieved by using fuzzy quantifiers such as "most," "almost all," etc. (Bonissone & Tong, 1985). Additionally, based on Zadeh's (1983) concepts of necessity and possibility, uncertainty in the data can be modeled by using a fuzzy extension of modal logic, in which a fuzzy interval is suggested to represent the degree of uncertainty rather than a scalar. The degree of necessity is represented by the lower bound of the fuzzy interval and the degree of possibility by its upper bound. Unlike probabilities, necessity and possibility do not require normalization. The value of the necessity of an hypothesis is always smaller than or equal to the value of its possibility.

As is known, it is very difficult (if not impossible) for anyone to say that there is complete or certain information for any specific domain in the real world. The reason is that human cognition is limited by various factors such as time, space, speed safety, techniques, materials, etc. Futhermore, the object situations are changing all the time so the information we obtain may not always match the real situations. In some cases, resources of the information have disappeared or are not available, and the further acquisition of information can not be carried out. In other cases the limitation of conditions makes information uncertain. These limitations and restrictions are commonly found in the job matching and job placement processes.

For solving uncertainty problems in expert systems, a number of approaches have been developed. One of the most popular methods is the Bayesian model theorem which involves probability to handle uncertainty (Forsyth, 1984). The basic assumption of a Bayesian model approach is the prior probability of an hypothesis given no evidence and the posterior probability of an hypothesis given some evidence (Forsyth, 1984).

For purposes of illustration, an example from medical research may be helpful. Let P(H) denote the prior probability of an hypothesis H given no evidence and P(HIE) denote the posterior probability of H given evidence. The formula for the probability of an hypothesis under some evidence is: [Mathematical Expression Omitted]

For example, the approximate ratio (prior probability of influenza for patients P(H) is known. Now, the task is to find the probability for the existence of influeza in a particular patient who has a fever (the evidence of E). P(EIH) = 1, because if a patient has influenza, a fever is certain. We may calculate P(E) by the following formula: [Mathematical Expression Omitted]

If we consider more than one hypothesis under evidence (a fever may result in several diseases), the following formula will be used: [Mathematical Expression Omitted]

The Certain Factor (CF) approach (Shortliffe, 1976) has been successfully used in MYCIN, one of the earliest expert systems for medical diagnostics which is based on the Bayesian theory. The CF is a number between -1 and +1. Positive CF's indicate that there is evidence that the hypothesis is valid. The larger the CF, the greater the belief in the hypothesis. When CF = 1, the hypothesis is assumed to be correct. On the other hand, negative CF's indicated that the weight of evidence suggests that the hypothesis is false. When CF = 0, there is either no evidence regarding the hypothesis, or the supporting evidence is equally balanced by evidence suggesting that the hypothesis is not true.

It is difficult to apply the Bayesian inference rule at each step of the inference chain in a rule-based system. This is because not only the prior probability but also joint probabilities are required to compute the probability of hypotheses from evidence. This calls for an enormous amount of data which is simply not possible in a practical situation (Mamdani, Efstathion, & Pang, 1986). In the mid-1960s, Zadeh (1965) introduced the concept of fuzzy sets. A fuzzy set is a class that admits the possibility of partial membership, that is, in general, membership in a fuzzy set is a matter of degree. A membership function is defined as a real number between 0 and 1. Let A represent a fuzzy set, then a membership function denoted as, [.sub.m]A. [.sub.m]A(X) = 0.8 means that X is a member of set A to an extend 9.8. [.sub.m]A(X) is defined as the grade of membership of X in A. for example, Heavy (90 Pounds) = 0.8, means that 90 Pounds is compatible with the concept (set) of Heavy to a degree 0.8, where X can also be a predicate: Heavy [Lift (John)] = 0.8, where Lift (John) = 90 Pounds.

It can also be considered as the interpretation of the proposition of "John can lift heavy object," which can also be viewed in this example as a fuzzy assertion of a restriction on the possible values of John's physical strength, rather than as an assertion concerning the membership of John in a class of individuals. Therefore, a fuzzy restriction may also be interpreted as a "possibility" distribution.

The basic operations that are applied to fuzzy sets are: Not [Mu]A(X)= 1 -[Mu]A(X) [Mu]A(X) union MB(X) - max [[Mu]A(X),[Mu]B(X)] [Mu]A(X) intersection [Mu]B(X) = min [[Mu]A(X),[Mu]B(X)]

Basically, fuzzy set models deal with the concept of "possibility." These include complete information, unreliable information, loosely defined concepts and imprecise quantification. This approach, in essence, is able to deal with some of the problems the current computerized job-matching systems encounter and cannot resolve.

The advantage of the fuzzy set theory is that joint possibilities are not required. Furthermore, the fuzzy set theory makes it possible to be free from the law of contradiction and therefore handle conflicting propositions. Its potential application is the management of uncertainty in expert systems and, more generally, in the design of decision-support systems in which common sense knowledge plays and important role.

Job-matching.

By capturing the wealth of clinical knowledge developed by experienced job placement experts over years of clinical practice and incorporated in the computer, we believe that we can develop an expert job-matching system capable of reasoning, making sensible decision, learning from cumulative job-matching experience, and handling ambiguous data commonly found in real life job-matching situations. This goal of increasing the sophistication of job-matching systems is related to the concept of "fidelity" discussed by Berven (1987) as it applies to computer based training systems. Enhanced fidelity in computer-based job-matching systems will likewise enhance the value of job-matching systems for rehabilitation professionals, resulting in better quality job placements for people with disabilities.

There is no technical reason why the professions of rehabilitation counseling or vocational evaluation cannot develop and utilize a computerized job-matching system conceptually as presented. Of course, the usefulness of this kind of system must be validated through careful research. For example, research must be conducted to study: (a) what impact will the proposed knowledge-based system have on the decision-making processes of rehabilitation clients? (b) what differences exist between the proposed knowledge-based system and the current DOT based job-matching systems with respect to the time required in processing occupational information? (c) what impact will the proposed knowledge-based system have on the rehabilitation "users" since the proposed system will be able to communicate and dialogue with them? (d) whether the information obtained from the proposed knowledge based system lead to improved vocational outcomes for rehabilitation clients?

As sophisticated expert systems are being developed and gaining increased acceptance by the business, industry, and the scientific communities, concert research and development efforts should also be undertaken by rehabilitation researchers to explore ways of adopting this technology in rehabilitation to benefit both the professionals and people with disabilities.

Summary

In summary, we have presented a conceptual model for developing a second generation computer-based job-matching system utilizing the expert system approach. The central theme of our approach is that in addition to the use of quantitative methods, qualitative decision rules should also be incorporated in computer job-matching. By capturing the wealth of clinical knowledge developed by experienced job placement experts over years of clinical practice and incorporated in the computer, we believe that we can develop and expert job-matching system capable of reasoning, making sensible decision, learning from cumulative job-matching situation. This goal of increasing the sophistication of job-matching systems is related to the concept of "fidelity" discussed by Berven (1987) as it applies to computer-based training systems. Enchanced fidelity in computer-based job-matching systems will likewise enhance the value of job-matching systems for rehabilitation professionals, resulting in better quality job placement for people with disabilities.

There is no technical reason why the professions of rehabilitation counseling or vocational evaluation cannot develop and utilize a computerized job-matching system conceptually as presented. Of course, the usefulness of this kind of system must be validated through careful research. For example, research must be conduced to study: (1) What impact will the proposed knowledge-based system have on the decision-making processes of rehabilitation clients? (2) what differences exist between the proposed knowledge-based ssytem and the current DOT based job-matching systems with respect to the time required inprocessing occupational information? (3) What impact will the proposed knowledge-based system have on the rehabilitation "users" since the proposed system will be able to communicate and dialogue with them? (4) whether the information obtained-from the proposed knowledge-based system lead to improved vocational outcomes for rehabilitation clients?

As sophisticated expert systems are being developed and gaining increased acceptance by the business, industry, and scientific communities, concert research and development efforts should also be undertaken by rehabilitation researchers to explore ways of adopting this technology in rehabilitation to benefit both professionals and people with disabilities.

References

Berven, N.L. (1987). Improving evaluation in counselor training and credentialing through standard simulations. In B.A. Edelstein & E.S. Berler (Eds.), Evaluation and accountability in clinical training (pp. 203-229) New York: Plenum. Bolton, B. (i985). Discriminant analysis of Holland's Occupational Types using the Sixteen PersonalityFactor Questionnaire. Journal of Vocational Behavior. 27(2), 210-217. Bonissone, P.P., & Tong, R.M. (1985). Editorial: Reasoning with uncertainty in expert system. International Journal of Man-Machine Studies, 22, 24i-250. Bose, J.L., Grzesik, T.A., Geist, G.0., @ Bryant, D.R. (1986). Misuse of occupational information in social security disability cases. Rehabilitation Counseling Bulletin, 30, 83-93. Botterbusch, K. (1986). A comparison of computerized job-matching systems. Menomonie, WI: Materials Development-Center, University of Wisconsin-Stout. Burkhead, E.J., & Sampson, J.P. (1985). Computer-assisted assessment in support of the rehabilitation process. Rehabilitation Counseling Bulletin. 28. 262-274, Career Evaluation System (1985). The use of fuzzy logic in career evaluation system. Niles, IL: Author. Chan, F., Matkin, R., Parker, H.J., h McCollum, P.S. (1988). Computer applications and issues related to their use in rehabilitation counseling. In S.E. Rubin and N. Rubin (Eds.), Contemporary challenges in rehabilitation counseling (PP- 167-174). Baltimore, MD: Paul H. Brookes. Chan, F., McCollum, P.S., & Parker, H.J. (1985). Computer assisted job placement: Selected applications. American Rehabilitation, 11, 18-21. Chan, F., McCollum, P.S., & Pool, D.A. (i985). Computer assisted rehabilitation services: A preliminary draft of the Texas casework model. Rehabilitation Counseling Bulletin, 28, 219-232. Chan, F., Parker, H.J., Dial, J.G., Lam, C.S., & Chan, Y.K. (1989). Implementing a computerized job-matching program with a hierarchicai structure: A person-environment fit approach. Journal of Rehabilitation, 55, 38-4-3. Chan, F., & Questad, K. (198i). Microcomputers in vocational evaluation: An application for staff training. Vocational Evaluation and Work Adjustment Bulletin, 14, 153-158, Crimando, W., & Godley, S.H. (1985). The computer's potential in enhancing empioyment opportunities of persons with disabilities. Rehabilitation Counseiing Bulletin, 28, 275-282. Dunn, L.D. (1974). Placement services in the vocational rehabilitation program. Menomonie: University of Wisconsin-Stout, Research and Training Center (RT-22). Dawis, R.V., England, G.W., Lofquist, L.H. (1964). A theory of work adjustment. Minnesota Studies in Vocational Rehabilitation. XV. Forsyth, R. (Eds.)(1984). Expert systems. New York: Chapman h Hall. Harmon, P., & King, D.(1985). Artificial intelligence in business: Expert systems, New York: John Wiley & Son, Inc. Holland, J.L. (1985). Making vocational choices. A theory of vocational personalities and work environments (2nd edition). Eng!ewood, NJ: Prentice-Hall. Lam, C.S., Lustig, P., Chan, F., & Leahy, M. (1987). Matching individual job satisfaction needs from a position focal point. Vocational Evaluation and Work Adjustment Bulletin, 20, 7-1O. Mamdani, A., Efstathion, J., & Pang, D. (1986). Inference under uncertainty. In M. Merry (Eds.) Expert System 85, (PP- 182-194). New York, NY: Cambridge University Press. McCollum, P.S. & Chan, F. (Eds) (1985). Implementing computer technology in the rehabilitation process. (Special Issue). Rehabilitation Counseling Bulletin, 28(4). Menz, F.E., & Bordieri, J.E. (1986). Rehabilitation facility administrator training needs: Priorities and patterns for the 1980's. Journal of Rehabilitation Administration. 10, 89-97. Miller. A.R., Trieman, D.J., Cain, P.S., Roos, P.A. (Eds). (1980). Work, jobs and occupations: A critical review of the "Dictionary of Occupational Titles". Washington, D.C.: National Academy Press. Naisbitt, J. (1982). Megatrends. New York: Warner Books, Restak, M.R., (1980, March). Smart machines learn to see, listen, and even think for us. Smithsonian, pp. 48-56. Roessler, R.T., Z Rubin, S.E. (1982). Case management and rehabilitation counseling. Baltimore: University Park Press. Shortliffe, E.H. (i976). Computer-based medical consultation: MYCIN. New York: Elsevier. Synnott, W.R., & Gruber, W.H. (1981). Information resources management. New York: Wiley. Zadeh, L.A. (1965). Fuzzy sets. Information and Control, B. 338-353. Zadeh, L.A. (1983). A computational approach to fuzzy quantifiers in natural languages. Computer & Mathematics, 9(1), 149-184.

Appendix I

1. Backward chaining One of the several control strategies that regulate the order in which inferences are drawn. In a rulebased system, back-ward chairning is initiated by a goal rule. The system attempts to determine if the goal rule is correct. It backs up to the if clauses of the rule and tries to determine if they are correct. This, in turn, leads the system to consider other rules that would confirm the if clauses. 2. Certainty The degree of confidence one has in a fact or relationship. As used in AI, this contrasts with probability, which is the likelihood that an event will occur. 3. Certainty Factor A numerical weight given to a fact or relationship to indicate the confidence one has in the or relationship. 4. Expert Systems It refers to any computer system that was developed by means of a loose collection of techniques associated with AI research. Thus, any computer system developed by means of an expert system building tool would qualify as an expert system even if the system was so narrowly constrained that it could never be said to rival a human expert. Some practitioners would prefer to reserve "expert system" for systems that trul7 rival human experts and use "knowledge system" when speaking of small systems developed by means of AI techniques. 5. Explanation This refers to information that is Presented to justify a particular course of reasoning or action. In knowledge systems this typically refers to a number of techniques that help a user understand what a system is doing. 6. Forward chaining One of several control strategies that regulate the order in which inferences are drawn. in a rule-based system, forward chaining begins by asserting all of the rules whose if clauses are true. It then checks to determine what additional rules might be true, given the facts it has already established. This process is repeated until the program reaches a goal or runs out of new possibilities. 7. Inference Engine The portion of a knowledge system that contains the inference and control strategies. More broadly, the inference engine also includes various knowledge acquisition, explanation, and user interface subsystems. Inference engines are characterized by the inference and control strategies they use. 8. Knowledqe An integrated collection of facts and relationships which, when exercised, produces competent performance. The quantity and quality of knowledge possessed by a person or a computer can be judged by the variety of situations in which the person of program can obtain' successful results. 9. Knowledqe Acquisition The process of locating, collecting, and refining knowledge. This may require interviews with experts, research in library, or introspection. The person undertaking the knowledge acquisition must convert the acquired knowledge into a form that can be used by a computer program. 10. Knowledge Base The portion of a knowledge system that consists of the facts and heuristics about a domain. 11. Knowledge Engineer An individual whose specialty is assessing problems, acquiring knowledge, and building knowledge systems. 12. Modus ponens A basic rule of logic that asserts that if we know that A implies B and we know for a fact that A is the case, we can assume B. 13. Reasoning The process of drawing inferences or conclusions. 14. Rule A conditional statement of two parts. The first part, comprised of one or more if clauses, establishes conditions comditions that must apply if a second part, comprised of one or more then clauses, is to be acted upon. 15. Rule-based Program A computer program that represents knowledge by means of rules. 16. Working Memory It is comprised of all of the attributevalue relationships that are established while the consultation is in progress. Since the system is constantly checking rules and seeking values, all values that are established must be kept immediately available until all the rules have been examined.
COPYRIGHT 1992 National Rehabilitation Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Author:Hattori, Kanetoshi
Publication:The Journal of Rehabilitation
Date:Jan 1, 1992
Words:5896
Previous Article:Ramps not steps: a study of accessibility preferences.
Next Article:The Americans with Disabilities Act: From Policy to Practice.
Topics:


Related Articles
Meeting the vocational needs of individuals with psychiatric disabilities through supported employment.
The institutionalization of disability myths: impact on vocational rehabilitation services.
Forgotten: elderly persons with disability - a consequence of policy.
The ADA and employment accommodations: what now?
Business and rehabilitation factors in the development of supported employment programs for adults with developmental disabilities.
Partners for Independence: an industry managed job placement model.
The role of the entrepreneur in job placement.
Linking workers with severe disabilities to social supports in the workplace: strategies for addressing barriers.
Employers' Attitudes Toward Hiring Persons with Disabilities and Vocational Rehabilitation Services.
Changing the term "job placement" to reflect choice.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters