CHAPTER 5
CAPTURING TACIT KNOWLEDGE
CHAPTER 5
CAPTURING TACIT KNOWLEDGE
TEST YOUR UNDERSTANDING
1. In your own words, define tacit knowledge capture. What makes it unique?
Knowledge capture is the process of extracting the knowledge of how an expert arrives at a solution for a particular problem. This includes the actual steps and reasoning involved in arriving at the solution as well as the subjective logic that an expert uses in addressing the problem.
Knowledge capture is unique, in that the procedure does not follow an algorithmic flow or a particular syntax to solve a problem.
2. Are there any particular steps involved in knowledge capture? Explain briefly.
Knowledge capture involves three steps:
Using an appropriate tool to elicit the information from the expert. Extensive interview with the expert usually accomplishes this step.
Interpreting the verbal information and inferring the expert’s underlying knowledge and reasoning process. In this step, the knowledge developer decides where the information gathered fits into the development process of the knowledge-based system. Throughout the interviewing process, the knowledge developer gathers information on the expert’s rationale for arriving at a decision. It is important that the knowledge developer thoroughly questions the expert on all angles of the problem domain.
Taking the results from step two and using it to build the rules that represent the expert’s thought process or solutions. This step may require several checks to ensure the resulting system meets the needs of the user and has captured “the expert” as closely as possible. Flowcharts, flow diagrams, decision trees, decision tables, and other graphic representation can be used to depict the rules for the expert’s solution.
3. “The problem solution should be based on symbolic knowledge rather than numerical computation.” Do you agree? Explain.
The problem solution should be based on symbolic knowledge rather than numerical computation. This is so subjective factors may be taken into account and analyzed rather than simply doing calculations with known data. The knowledge-based system then is able to make judgment consistently in situations where the human aspect would create numerous, varied results.
Subjective factors are important and exist in the expert’s process in arriving at the solution. If this were not the case, then an algorithmic language such as Cobol could be used to determine the problem solution.
4. How would one identify expertise?
The collection of the several indicators of expertise would help the knowledge developer identify who would be an appropriate expert for a problem domain. These include:
a. Genuine respect from peers with regard to the expert’s decisions as good decisions
b. People consult the expert when a problem arises
c. Admitting to not knowing everything about a problem which demonstrates his or her confidence and provides a realistic view of limitations, avoiding irrelevant information and focusing on the facts
d. Working with a clear focus
e. Being able to explain the information to different audience levels
f. Depth of detail and exceptional quality in explanations
g. Demonstrating no arrogance regarding personal credentials
h. Years of experience
i. Strong ties with people in power
5. Working with experts requires certain skills and experience. What suggestions or advice would you give to an inexperienced knowledge developer concerning:
a. working with or approaching an expert
b. preparing for the first session
A and b are interrelated in the overall effort and interactions with the expert(s). Any knowledge developer, no matter how well experienced, must educate himself or herself in the expert’s area and be fully prepared for the knowledge acquisition phase. Perceptions are extremely important in knowledge capture.
6. Working with multiple experts has definite benefits and limitations. Cite an example in which the use of multiple experts is a must. Explain your choice.
An example in which the user of multiple experts is a must could be for development of a knowledge-based system to predict the next direction of a given stock on the New York Stock Exchange. The reasons you would need multiple experts are the complexity of the problem domain, listening to a variety of views on stock exchange theory and behavior before attempting an approach or a solution. With this example, there is no single individual who is an expert in all aspects of the company stocks or the stock exchange or even the economy.
7. Use an example of your own to illustrate the conditions under which you would be willing to build a KM system based on a single expert. Justify your choice.
Most knowledge-based systems that require fewer than 50 rules are normally done with a single expert. Sometimes, the nature of the problem domain warrants multiple experts, especially if a number of users can be seriously affected by the solution if it is incorrect. Take the case of the author’s work on the diabetic foot advisor. The serious nature of diabetes and diabetic foot problems made it mandatory for two or more specialists to coordinate their views before the 40-rule prototype was finally accepted as a step toward the full-blown package. It would be interesting what the students come up with in answering this question.
8. Why should the knowledge developer understand the differences among the levels of experts? Isn’t an expert an expert regardless of level?
It is important for the knowledge developer to understand the differences among the levels of experts, because it influences the communication level and quality. Depending on the level of the expert, the knowledge developer adapts his questions to his or her understanding of the problem.
Highly expert persons generally give concise explanations and assume the knowledge developer has enough knowledge about the problem. This means focusing on the key steps, often skipping vital detailed information.
Moderately expert problem solvers may be more attentive in their explanations, yet they tend to provide detailed explanations. They are quicker to give answers than the highly expert person and more often adapt their description to the level of the knowledge developer.
In contrast, new experts are more likely to offer answers that are brief and fragmented, which suggests shallow knowledge of the domain. The knowledge developer must adapt the questions to the level of the expert and understand that the types of questions asked of the expert will be different for each expert type.
9. Explain the relationships between an expert’s:
a. motivation and willingness to share knowledge
b. credibility and perceptual ability
c. creativity and well-developed perceptual abilities
a. There are many aspects that must be understood in determining the relationship between an expert’s motivation and willingness to share knowledge. Personality places an important role in how the expert will react to being selected as the expert for the system. Praise and recognition go a long way in helping the expert feel that you are not there to gather expertise and then replace him or her with a system.
b. Credibility relates to believability or how well someone can be trusted with the information or advice he or she offers. An expert who demonstrates competency in the way they field questions gain a reputation for believability. Perceptual ability, however, has to do with how well a person sees through things. For example, if the expert perceives the knowledge developer as trying tricky questions that might adversely affect the expert’s credibility, the expert is expected to reach or adjust his or her cooperation for the rest of the session or in fact the rest of the capture phase.
c. Creativity addresses one’s inventiveness when solving a problem or explaining the reasoning behind a solution. Creative approaches can clarify the expert’s thought process that is more easily carried out with ability to perceive events or outcomes.
10. How does pattern matching adversely affect the quality of an interview?
Pattern matching is the line of reasoning used by experts that is analogies—comparing a problem to a similar one encountered in a different setting months or years ago to make judgment. It is a fuzzy comparison that must be clarified before the information captured can be usable. If one is interviewing an expert who uses this line of reasoning, one must learn more about the expert’s examples and pay attention to how the expert forms analogies, rather than learning about the problem domain, per se.
11. In what ways is visual imagery helpful in recalling chunks of experience?
Visual imagery is very helpful in recalling chunks of experience, because the expert uses the image to mentally arrive at an answer. The author feels that if a person is poor on imagery, he or she is likely to be poor on explaining things.
Visual imagery is also represented by phrases the expert uses in a descriptive answer. It is important that the knowledge developer ask questions about the image as the expert is explaining their answer.
12. Distinguish between:
a. validity and reliability
b. multiple-choice and dichotomous questions
c. question format and question content
d. question wording and question content
a. Validity means integrity and consistency in the way a question or a phrase is interpreted. For example, on an exam, if a question means different things to different students, then there is a validity problem. This, in turn, affects reliability, which means trustworthiness and dependability. It is a measure of truthfulness or credibility.
b. Multiple-choice questions offer specific choices, faster tabulation, and less bias due to the way the answers are ordered. Because more people have a tendency to favor the first choice item, alternating the order of the choices may reduce bias. However, additional time is needed to pick the choice item question.
In contrast, dichotomous (yes/no) questions are a special type of multiple-choice. They offer two answers. The sequence of questions and content are also important.
c-d. Students should demonstrate creativity in the way they answer a and b above. Sometimes they do well discussing this question in a team environment. The basis for including this question here is to alert them to the importance of taking time to properly phrase the question they ask during knowledge capture. In addition, question content and sequence are equally as important.
13. Briefly explain each of the uncontrolled sources of error that the knowledge developer needs to consider. Which source do you consider most seriously? Why?
There are several uncontrolled sources of error that may reduce information reliability:
a. the expert’s perceptual slant that varies among experts
b. with the passing of time, exact details of an event generally become more difficult to describe
c. an expert may distort the descriptions of events for fear that the knowledge-based system will replace him or her
d. some experts have difficulty communicating their knowledge
e. the risk of role bias increases when the expert is aware of his or her importance as the critical element in building the knowledge-based system.
Perhaps the most serious error is that the expert will distort his or her descriptions. If this happens, the integrity of the entire knowledge capture process is in question and the trust between the knowledge developer and the expert has truly not developed. The course of action may be to choose another expert or try to explain the importance of the project and allay the expert’s fears that his/her job is in jeopardy.
14. What is the interviewer effect? How likely is the occurrence of this problem?
The interviewer effect is something about the knowledge developer that colors the interview response of the domain expert. This includes age, race, and gender that bias the response. Although we like to think that bias does not exist, knowledge developers are human and may be swayed in their reasoning due to these effects. If this is the case, the knowledge developer must find ways to correct these tendencies. If resolution is not possible, he or she should not be involved in the knowledge capture process
15. Review briefly some of the problems encountered during the interview. Which problem do you consider most serious? Why?
Problems encountered during the interview range from biased responses, the way the question is phrased, to personality clashes between the expert and the knowledge developer. A response bias occurs when experts answer questions on the basis of their interpretation of the wording of the question and in response to certain constraints—lack of time, lack of motivation, perceived hostility, an attempt to please the knowledge developer, and so on. The problem of inconsistency is most likely to occur when the knowledge developer interviews two domain experts and is inconsistent when asking the questions.
Communication is a critical issue for both the knowledge developer and the expert(s). Not everyone is an “expert” at the art of communication. Nonverbal communication is just as relevant. The length of the questions and the duration of the interview could cause problems with attention spans breaking down and the quality of the thought process degenerates.
The most serious problem is the hostile attitude that the expert and the knowledge developer may have toward each other. Under these conditions, the quality of the information captured is unreliable and there needs to be a correction of the hostility. If the hostility cannot be resolved, the expert, the knowledge developer, or both must be replaced.
16. The chapter suggests a guide to a successful interview. Does it allow enough flexibility? That is, what determines how strictly one should follow the guide?
Guide or no guide, when it comes to conducting interviews with the expert, an experienced knowledge developer should assess the situation and react extemporaneously to it. It is nice to be reminded through a guide, but in real-life interview settings, the guide is not the answer to every situation that comes up. Therefore, there should be enough flexibility to adjust to an extenuating situation.
For example, an expert comes to the meeting an hour late, what do you say? How do you say it? Much depends on how well you know the expert, whether he or she would voluntarily provide a reason for being late, or the mood in which he or she comes to the meeting. This is what makes it risky to prepare and to conduct an interview. The guide can only offer general suggestions that may or may not fit the situation. Implicit in the guide, then, is enough flexibility to address a given situation accordingly.