📘 Lecture 37 — Capturing Tacit Knowledge and Expert’s Evaluation
📖 Overview: This lecture focuses on the process of capturing tacit knowledge from experts, which is crucial for building knowledge management (KM) systems. It covers how to identify and evaluate experts, methods for working with single or multiple experts, and the use of interviews as a primary tool for eliciting tacit knowledge. Understanding this process is vital for converting human expertise into usable system rules.
🗂️ Topics Covered
The lecture begins by defining knowledge capture and its important steps, then moves to expert evaluation including indicators of expertise, qualifications, and levels of expertise. It compares the advantages and disadvantages of working with single versus multiple experts. The lecture then discusses developing relationships with experts, including creating impressions, understanding expert styles, preparation, and different approaches for multiple experts. Finally, it covers interviewing as a tacit knowledge capture tool, including types of interviews, guidelines, reliability issues, and rapid prototyping.
📝 Lecture Summary
Capturing the Tacit Knowledge
Knowledge Capture is defined as the process using which the expert's thoughts and experiences can be captured. In this case, the knowledge developer collaborates with the expert in order to convert the expertise into the necessary program code(s). The important steps include: using appropriate tools for eliciting information; interpreting the elicited information and consequently inferring the experts' underlying knowledge/reasoning process; and finally, using the interpretation to construct the necessary rules which can represent the expert's reasoning process.
Expert Evaluation
Indicators of expertise include: the expert commands genuine respect; the expert is found to be consulted by people in the organization when a problem arises; the expert possesses self-confidence and has a realistic view of limitations; the expert avoids irrelevant information and uses facts and figures; the expert can explain properly and customize their presentation according to the audience level; the expert exhibits depth of detailed knowledge with exceptional quality of explanation; and the expert is not arrogant regarding their personal information.
Expert's qualifications require that the expert should know when to follow hunches and when to make exceptions, be able to see the big picture, possess good communication skills, tolerate stress, think creatively, exhibit self-confidence, maintain credibility, operate within a schema-driven/structured orientation, use chunked knowledge, generate enthusiasm and motivation, share expertise willingly, and emulate an ideal teacher's habits.
Experts levels of expertise include highly expert persons, moderately expert problem solvers, and new experts.
Capturing single vs. multiple experts' tacit knowledge: Advantages of working with a single expert include being ideal for building a simple KM system with few rules, ideal when the problem lies within a restricted domain, facilitating logistics aspects of coordination, easier resolution of personal conflicts, and greater confidentiality. Disadvantages include that the expert's knowledge is often not easy to capture, provides a single line of reasoning, experts are more likely to change meeting schedules, and knowledge is often dispersed. Advantages of working with multiple experts include benefiting complex problem domains, stimulating interaction, allowing consideration of alternative ways of representing knowledge, and formal meetings generating thoughtful contributions. Disadvantages include frequent disagreements, complicated scheduling, harder confidentiality, process loss from overlapping mental processes, and often requiring more than one knowledge developer.
Developing Relationship with Experts
Creating the right impression: The knowledge developer must learn to use psychology, common sense, and technical as well as marketing skills to attract the expert's respect and attention.
Understanding the expert's style of expression: Experts usually use one of these styles: Procedure type (logical, verbal, procedural); Storyteller type (focused on content at the expense of the solution); Godfather type (compulsive to take over); Salesperson type (spends time dancing around the topic, explaining why their solution is best).
Preparation for the session: Before the first appointment, the knowledge developer must acquire knowledge about the problem and the expert. Initial sessions are most critical. The developer must build trust, be familiar with project terminology, review existing documents, and make a quick rapport with the expert.
Deciding the location for the session: Protocol calls for the expert to decide the location. The expert is usually more comfortable having necessary tools and information available. The meeting place should be quiet and free of interruptions.
Approaching multiple experts: The individual approach has sessions with one expert at a time. The approach using primary and secondary experts involves sessions with the senior expert early for clarification, then asking other experts for detailed probing. The small groups approach has experts gather together to discuss the problem domain, providing a pool of information. Responses are monitored and functionality is tested against others. This requires experience in assessing tapped knowledge and cognition skills, and the developer must deal with power issues affecting opinions.
Interviewing as a Tacit Knowledge Capture Tool
Advantages of using interviewing as a tacit knowledge capture tool include: flexibility, excellent for evaluating validity of information, very effective for eliciting information about complex matters, and people often enjoy being interviewed.
Types of interviews: Unstructured types are difficult to conduct and used when exploring an issue. Structured types are goal-oriented and used when specific information is needed, with question types including multiple-choice, dichotomous, and ranking scale questions. Semi-structured types involve predefined questions but allow the expert some freedom in expressing answers.
Guidelines for successful interviewing include setting the stage and establishing rapport, phrasing questions, listening closely and avoiding arguments, and evaluating the session outcomes.
Reliability of information can be reduced by uncontrolled sources of error: expert's perceptual slant, failure to exactly remember what happened, fear of unknown, communication problems, and role bias. Errors in the knowledge developer's part involve interviewer effect where something about the developer colors the expert's response, including gender, age, and race effects.
Problems encountered during interviewing include response bias, inconsistency, communication problems, hostile attitude, standardizing questions, and setting interview length.
Process of ending the interview: The end should be carefully planned. One procedure calls for halting questioning a few minutes before the scheduled ending time and summarizing key points. This allows the expert to comment and schedule future sessions. Verbal and nonverbal cues can be used for ending (refer to Table 5.2, page 148 of textbook).
Issues to be prepared for include: eliciting knowledge from experts who cannot say what they mean, setting up the problem domain, dealing with uncertain reasoning processes, handling difficult relationships with experts, and dealing with situations where the expert dislikes the developer.
Rapid Prototyping in interviews: Rapid prototyping is an approach to building KM systems where knowledge is added with each knowledge capture session. This iterative approach allows the expert to verify rules as they are built. It opens up communication through demonstration, reduces risk of failure through instant feedback and modification, allows the developer to learn with each change, and is highly interactive. However, the prototype can create user expectations that become obstacles to further development.
⭐ Key Takeaways
Students must remember that knowledge capture is about converting expert thoughts into system rules through a structured process of elicitation, interpretation, and rule construction. Expert evaluation requires assessing both indicators of expertise and specific qualifications, with experts falling into different levels of proficiency. Working with single experts offers simplicity and confidentiality but limited perspectives, while multiple experts provide richer knowledge but introduce coordination and conflict challenges. Building effective relationships with experts requires understanding their communication style, proper preparation, and choosing appropriate approaches for multiple experts. Finally, interviewing is a flexible but complex tool that requires managing reliability issues, avoiding interviewer effects, and handling common problems, with rapid prototyping offering an iterative way to build systems while verifying rules.
🧠 Quick Revision Questions
- What are the three important steps in the knowledge capture process?
- List three indicators of expertise and three qualifications that an expert should possess.
- What are the advantages and disadvantages of working with a single expert compared to multiple experts?
- Identify and describe the four types of expert expression styles mentioned in the lecture.
- Define rapid prototyping and explain three benefits and one drawback of using this approach in interviews.
📘 Lecture 38 — Knowledge Elicitation: Data Mining and Knowledge Codification Methods
📖 Overview: This lecture covers methods for eliciting knowledge from experts and codifying it into usable forms for organizational use. It distinguishes between data mining purposes (description and prediction), explores various knowledge elicitation techniques (observation, brainstorming, Delphi, concept mapping, etc.), and details knowledge codification tools like decision tables, decision trees, frames, and production rules. Understanding these methods is critical for building effective knowledge management systems.
🗂️ Topics Covered
The lecture begins with the two basic purposes of data mining (descriptive and predictive), then systematically covers knowledge elicitation methods including on-site observation, brainstorming (traditional and electronic), nominal group technique, Delphi method, concept mapping, blackboarding, and knowledge capture systems. It transitions to knowledge codification, explaining modes of knowledge conversion, codification tools such as knowledge maps, decision tables, decision trees, frames, and production rules, case-based reasoning, knowledge-based agents, and finally the knowledge developer's required skill set.
📝 Lecture Summary
Data Mining and Web Mining Purposes
There are two basic purposes for data mining: descriptive and predictive. Descriptive data mining tries to understand patterns in people or things—for example, purchasing habits during Super Bowl week or financial dealings of people suspected of money laundering. This explains behavior of individuals or phenomena. Predictive data mining attempts to predict behavior by building a model from history, such as modeling the stock market to forecast future performance.
Knowledge Elicitation Methods: On-Site Observation (Action Protocol)
This process involves observing, recording, and interpreting the expert's problem-solving while it occurs. The knowledge developer listens more than talks, avoids giving advice or passing judgment, and never argues with the expert during task performance. Compared to interviewing, on-site observation brings the developer closer to actual steps and procedures. Disadvantages include some experts disliking observation, distraction from others in the setting, and concerns about accuracy and completeness of captured knowledge.
Brainstorming
Brainstorming is an unstructured approach for generating creative solutions involving multiple experts in a session. Questions can be raised for clarification, but no evaluations occur on the spot. Similarities that emerge are grouped logically and evaluated by asking: what benefits are gained if an idea is followed, what problems can be solved, and what new problems may arise. The general procedure includes introducing the session, presenting the problem, prompting idea generation, and looking for convergence. If experts cannot agree, the developer may call for a vote or consensus.
Electronic Brainstorming
Electronic brainstorming is a computer-aided approach for multiple experts. It begins with a pre-session plan identifying objectives and structuring the agenda, presented to experts for approval. During the session, each expert sits at a PC and engages in a predefined approach to resolve the issue, generating ideas. This allows experts to present opinions without waiting for turns. Comments are displayed electronically on a large screen without identifying the source, protecting introvert experts and preventing tagging of comments to individuals. Benefits include improved communication, effective discussion of sensitive issues, and closure with concise recommendations, leading to convergence of ideas and joint ownership of the solution.
Nominal Group Technique (NGT)
Nominal Group Technique (NGT) provides an interface between consensus and brainstorming. The panel of experts becomes a nominal group whose meetings are structured to pool individual judgment effectively. Idea writing is a structured group approach for developing ideas and exploring their meaning, usually resulting in a written report. NGT is itself an idea-writing technique.
Delphi Method
The Delphi method is a survey of experts where a series of questionnaires pool experts' responses for solving a specific problem. Each expert's contributions are shared with others by using the results from each questionnaire to construct the next one.
Concept Mapping
Concept mapping is a network of concepts consisting of nodes and links. A node represents a concept, and a link represents the relationship between concepts. It is designed to transform new concepts/propositions into existing cognitive structures related to knowledge capture. It is a structured conceptualization that allows groups to function without losing individuality. Concept mapping can be used to design complex structures, generate ideas, communicate ideas, or diagnose misunderstanding. The six-step procedure is: preparation, idea generation, statement structuring, representation, interpretation, and utilization. Similar to concept mapping, a semantic net is a collection of nodes linked together to form a net, where each idea is represented by a node connected by arcs showing relationships between nodes.
Blackboarding
In blackboarding, experts work together to solve a problem using a blackboard as their workspace. Each expert gets equal opportunity to contribute. It is assumed all participants are experts, though they may have acquired expertise in different situations. The process continues until a solution is reached. Characteristics include diverse problem-solving approaches, common language for interaction, efficient and flexible information representation, iterative problem-solving, and organized participation. Components include: the Knowledge Source (KS) — each an independent expert observing the blackboard and contributing partial solutions; the Blackboard — a global memory structure storing partial solutions; and a Control Mechanism — coordinating the pattern and flow of solution. The inference engine and knowledge base are part of this system.
Knowledge Capture Systems: Systems that Preserve and Formalize Knowledge
Stories may be elicited through anthropological observation using a naïve but interested interviewer, whose naïveté facilitates natural volunteering of stories. Using a group with common context (e.g., community of practice) forms storytelling circles. Useful methods include fish tales, alternative histories, shifting characters or context, indirect stories for security, and using metaphors to start storytelling.
Concept maps represent knowledge through concepts (patterns or regularities in objects/events) shown as text inside geometric shapes. Different concepts are connected by lines representing propositions, labeled with verb phrases or prepositions indicating relationship nature. More general concepts appear at the top with specialization toward the bottom. Cross-links represent inter-domain relations.
Context-based reasoning (CxBR) models tactical situations and operations during special tactical situations. Its three tenets are: (1) tactical situations call for specific sets of actions and procedures; (2) situations are dynamic and may require transition to new contextual sets of actions; (3) what is likely to happen in a situation is limited by the situation itself.
Knowledge Codification
Knowledge codification means converting tacit knowledge to explicit knowledge in usable form for organizational members. Tacit knowledge (human expertise) is identified and leveraged through forms producing highest business return. Explicit knowledge is organized, categorized, indexed, and accessed—often using decision trees and decision tables. Codification must build the knowledge base, which supports training and decision making in: diagnosis, training/instruction, interpretation, prediction, and planning/scheduling. Before codification, note that recorded knowledge is often difficult to access, diffusion of new knowledge is slow, knowledge is hoarded rather than shared, and knowledge may not be in proper form, available at the correct time, present in the proper location, or may be incomplete.
Modes of Knowledge Conversion
- Tacit to tacit: socialization — developer looks for experience in knowledge capture.
- Tacit to explicit: externalizing — explaining/clarifying tacit knowledge via analogies, models, or metaphors.
- Explicit to tacit: internalizing — fitting explicit knowledge to tacit knowledge.
- Explicit to explicit: combining — categorizing, reorganizing, or sorting different bodies of explicit knowledge to lead to new knowledge.
Codifying Knowledge
Before codification, an organization must focus on: what organizational goals the codified knowledge will serve, what knowledge exists to address these goals, how useful existing knowledge is for codification, and how to codify knowledge. Codifying tacit knowledge entirely is often difficult because it is developed and internalized over long periods.
Codification Tools/Procedures: Knowledge Maps
Knowledge maps originated from the belief that people act on things they understand and accept, indicating self-determined change is sustainable. A knowledge map is a visual representation of knowledge that can represent explicit/tacit, formal/informal, documented/undocumented, and internal/external knowledge. It is not a repository but a directory pointing to people, documents, and repositories. It may identify strengths to exploit and gaps to fill. Knowledge mapping is useful for visualizing complex systems like ecosystems, the internet, or customer-supplier chains. It is a multi-step process where keys are extracted from databases or literature, placed in tabular form as lists of facts, and connected in networks. A popular example is a skills planner matching employees to jobs, built by: developing knowledge requirement structure, defining knowledge required for specific jobs, rating employee performance by knowledge competency, and linking the map to training programs.
Decision Table
A decision table consists of conditions, rules, and actions. Example: A phone card company gives discounts—5% for orders over $35, 4% for orders between $20 and $35, no discount under $20—all conditional on payment within two weeks. The decision table organizes these conditions (paid within two weeks? order>$35? $20≤order≤$35? order<$20?) into rules (1-4) with corresponding actions (5% discount, 4% discount, no discount).
🔑 Definition — Decision Table: A technique for knowledge codification that uses a matrix of conditions, rules, and actions to represent decision logic.
📌 Example: For the phone card company, Rule 1: Paid within 2 weeks=Y, Order>$35=Y → 5% discount. Rule 2: Paid within 2 weeks=Y, $20≤Order≤$35=Y → 4% discount. Rule 3: Paid within 2 weeks=Y, Order<$20=Y → No discount. Rule 4: Paid within 2 weeks=N → No discount (regardless of amount).
Decision Tree
A decision tree is a hierarchically arranged semantic network. The telephone company discounting policy can be represented as a decision tree showing the sequential decision points (paid within 2 weeks? then order amount ranges) leading to discount outcomes.
Frames
A frame is a codification scheme for organizing knowledge through previous experience, dealing with combined declarative and operational knowledge. Key elements: Slot — a specific object being described or an attribute of an entity; Facet — the value of an object/slot.
Production Rules
Production rules are conditional statements specifying an action to take if a certain condition is true. They codify knowledge as premise-action pairs using syntax: IF (premise) THEN (action). Example: IF income is 'standard' AND payment history is 'good' THEN 'approve home loan'. In knowledge-based systems, rules are based on heuristics. Rules can incorporate uncertainty levels; a certainty factor (synonymous with confidence level) is a subjective quantification of expert judgment. The premise is a Boolean expression that must be true for the rule to apply. The action clause follows THEN and may contain statements separated by AND's or commas.
In knowledge-based systems, planning involves: breaking the system into manageable modules, considering partial solutions linked through rules, deciding on programming languages and software packages, testing and validating, developing the user interface, promoting clarity and flexibility, and reducing unnecessary risk.
The role of inferencing implies deriving a conclusion from statements that only imply that conclusion. An inference engine is a program managing inferencing strategies. Reasoning is applying knowledge to reach a conclusion, depending on premise and general knowledge.
Case-Based Reasoning
Case-based reasoning (CBR) reasons from relevant past cases, similar to humans using past experiences to arrive at conclusions. It records and documents cases, then searches appropriate cases to determine usefulness for solving new cases. The aim is to bring up the most similar historical case matching the present case. Knowledge is expanded by adding new cases and reclassifying the case library. A case library may require considerable database storage and efficient retrieval.
Knowledge-Based Agents
An intelligent agent is program code capable of performing autonomous action in a timely fashion. Agents can exhibit goal-directed behavior by taking initiative and can interact with other agents or humans using agent communication language. In knowledge-based systems, an agent can learn from user behavior and deduce future behavior to assist the user.
Knowledge Developer's Skill Set
Knowledge requirements include: computing technology and operating systems, knowledge repositories and data mining, domain-specific knowledge, and cognitive psychology. Skills requirements include: interpersonal communication, ability to articulate project rationale, rapid prototyping skills, personality attributes, and understanding of job roles.
⭐ Key Takeaways
A student must remember the distinction between descriptive and predictive data mining, and know that knowledge elicitation methods range from on-site observation to blackboarding—each with specific procedures and purposes. The modes of knowledge conversion (socialization, externalization, internalization, combination) are foundational to understanding how tacit knowledge becomes explicit. For codification, the key tools are knowledge maps (directories, not repositories), decision tables (condition-action matrices), decision trees (hierarchical semantic networks), frames (slot-facet structures), and production rules (IF-THEN statements). Case-based reasoning mirrors human experience-based problem solving, while knowledge-based agents provide autonomous, goal-directed assistance. Finally, the knowledge developer must possess both technical knowledge (computing, data mining, domain expertise, cognitive psychology) and interpersonal skills (communication, articulation, rapid prototyping).
🧠 Quick Revision Questions
- What are the two basic purposes of data mining, and how do they differ?
- List and briefly describe five knowledge elicitation methods covered in this lecture.
- What are the four modes of knowledge conversion, and what does each mode accomplish?
- How does a decision table differ from a decision tree in representing knowledge codification?
- What are the three components of a blackboard system, and what role does each play?
📘 Lecture 39 — Knowledge Sharing and Transfer Systems
📖 Overview: This lecture explores systems that organize and distribute knowledge, focusing on the crucial requirements for successful implementation. It categorizes and explains different types of knowledge-sharing systems, from incident databases to expertise locators, highlighting how each serves distinct organizational needs.
🗂️ Topics Covered
The lecture covers five crucial requirements for successful implementation of knowledge-sharing systems: collection and systematic organization of information, minimization of up-front knowledge engineering, exploiting user feedback for maintenance and evolution, integration into existing environment, and active presentation of relevant information. It then discusses five different types of knowledge-sharing systems: incident report databases, alert systems, best practices databases, lessons learned systems, and expertise locator systems.
📝 Lecture Summary
Knowledge Sharing Systems: Systems that Organize and Distribute Knowledge
1. Describe the crucial requirements for the successful implementation of knowledge-sharing systems.
a. Collection and systematic organization of information from various sources.
Most organizational business processes require information and data including CAD drawings, e-mails, electronic documents such as specifications, and even paper documents. This requisite information may be dispersed throughout the organization. This first step requires the collection of this information throughout the organization.
b. Minimization of up-front knowledge engineering.
Knowledge-sharing systems must take advantage of explicit organizational information and data, such that these systems can be built quickly, generate returns on investment, and be able to adapt to new requirements. This information and data is mostly found in databases and documents.
c. Exploiting user feedback for maintenance and evolution.
Knowledge-sharing systems should concentrate on capturing the knowledge of the organization’s members. This includes options for maintenance and user feedback so the knowledge can be kept fresh and relevant. Furthermore, knowledge-sharing systems should be designed to support user’s needs and their business process workflows.
d. Integration into existing environment.
Knowledge-sharing systems must be integrated into an organization’s information flow, by integrating with the IT tools currently used to perform the business tasks. Humans, by nature, will tend to avoid efforts to formalize knowledge (ever met a computer programmer that enjoys commenting her code?). In fact, as a rule of thumb, if the effort required in formalizing knowledge is too high, it should be left informal, to be described by humans, and not attempt to be made explicit. For instance, consider the possibility of capturing the “how–to” knowledge, of how to ride a bicycle. Clearly an understanding of the laws of physics can help explain why a person stays on the bicycle while it’s moving, but few of us recall these laws while we ride. Other than the proverbial “keep your feet on the pedal,” which doesn’t explicate much about the riding process, most of us learned to ride a bicycle through hours of practice, and many falls, while we were kids. It would be impractical to try to codify this knowledge and make it explicit. On the other hand, it might be useful to know who’s a good bicycle rider, in particular if one is looking to put together a cycling team. 💡 Why this matters: This requirement underscores a critical design principle: forcing knowledge formalization when it is too difficult compromises system adoption and effectiveness. It's often better to leave certain knowledge informal and instead build systems that connect people who hold that knowledge.
e. Active presentation of relevant information.
Finally, the goal of an active knowledge-sharing system is to present its users with the required information when and wherever it’s needed. These systems are envisioned to become intelligent assistants, automatically eliciting and providing knowledge that may be useful in solving the current task, whenever and wherever it’s needed.
2. Discuss which the different types of knowledge-sharing systems are.
a. Incident report databases
These are used to disseminate information related to incidents or malfunctions, for example, of field equipment (like sensing equipment outages) or software (like bug reports). Incident reports typically describe the incident together with explanations of the incident, although they may not suggest any recommendations.
b. Alert systems
These were originally intended to disseminate information about a negative experience that has occurred or is expected to occur. However, recent applications also include increasing exposure to positive experiences.
c. Best practices databases
These describe successful efforts, typically from the reengineering of business processes that could be applicable to organizational processes. Best practices differ from lessons learned in that they capture only successful events, which may not be derived from experience.
🔑 Definition — Lessons Learned: knowledge gained from experience (successful or unsuccessful) that can be applied to future situations.
d. Lessons learned systems (LLS)
The goal of LLS is “to capture and provide lessons that can benefit employees who encounter situations that closely resemble a previous experience in a similar situation.” LLS could be pure repositories of lessons or sometimes intermixed with other sources of information (e.g., reports).
e. Expertise Locator Systems (ELS)
These serve the purpose to identify experts in the organization. Experts may need to be identified to help solve technical problems or staff project teams, to match employee competencies with positions within the company, or to perform gap analysis that point to intellectual capital inadequacies within the organization. The intent of these systems is to catalog knowledge competencies, including information not typically captured by human resources systems, in a way that could later be queried across the organization.
⭐ Key Takeaways
Knowledge-sharing systems must balance explicit knowledge capture with practical implementation constraints, minimizing upfront engineering effort while maximizing user feedback and integration into existing workflows. The five types of systems serve different purposes: incident report databases for malfunction documentation, alert systems for disseminating experiences (both positive and negative), best practices databases for capturing only successful efforts, lessons learned systems for experience-based knowledge that can benefit others in similar situations, and expertise locator systems for identifying internal experts and cataloging knowledge competencies. A critical design rule is that if formalizing knowledge requires too much effort, it should remain informal and instead be connected through human networks. Successful systems actively present relevant information to users when and where it is needed, acting as intelligent assistants. The distinction between best practices (only successful events) and lessons learned (all experiences, positive or negative) is a key exam concept.
🧠 Quick Revision Questions
- What are five crucial requirements for successful implementation of knowledge-sharing systems?
- How do best practices databases differ from lessons learned systems?
- What is the primary purpose of an Expertise Locator System (ELS)?
- When should knowledge be left informal rather than made explicit in a knowledge-sharing system?
- What distinguishes alert systems from incident report databases in terms of what they disseminate?
📘 Lecture 40 — Corporate Memory; Types of Knowledge Repositories
📖 Overview: This lecture explores corporate memory and its relationship to knowledge management, detailing the lessons learned process and various knowledge repository types. It focuses on practical implementations of knowledge-sharing systems, including expertise locator systems at major organizations, the role of taxonomies, and communities of practice in sharing tacit knowledge.
🗂️ Topics Covered
The lecture covers corporate memory definition and its relationship to KM, the five-step lessons learned process (collect, verify, store, disseminate, apply), the role of taxonomies in knowledge-sharing systems, comparing ELS characteristics at HP, NSA, and Microsoft, communities of practice for tacit knowledge sharing, document management systems, barriers to knowledge-sharing system use, and five specific types of knowledge-sharing systems.
📝 Lecture Summary
Corporate Memory and KM Relationship
Corporate memory is the collection of all explicit and tacit knowledge that may or may not be explicitly documented, but is specifically referenced. Corporate memory is crucial to the operation and competitive advantage of an organization. A focus of KM is the development of mechanisms and technologies that prevent corporate memory loss through knowledge sharing mechanisms, technologies, and applied systems. Such loss may result from the lack of appropriate technologies for the organization and exchange of explicit information and a lack of support for communication.
💡 Why this matters: Without deliberate KM systems, organizations lose valuable knowledge when employees leave or retire, directly affecting competitive advantage and operational efficiency.
The Lessons Learned Process (LLS)
1) Collect the Lessons
This task involves collecting the lessons (or content) that will be incorporated into the LLS. There are six possible lesson content collection methods:
a) Passive - the most common form of collection. Contributors submit lessons through a paper or Web-based form.
b) Reactive - where contributors are interviewed by a third party for lessons. The third party will submit the lesson on behalf of the contributor.
c) After-action collection - where lessons are collected during a mission debriefing, as for example, in military organizations.
d) Proactive collection - where lessons are automatically collected by an expert system, which may suggest that a lesson exists based on analysis of a specific content. For example, an expert system could monitor individual’s e-mail and prompt him/her when it understands that a lesson is described.
e) Active collection - where a computer-based system may scan documents to identify lessons in the presence of specific keywords or phrases.
f) Interactive collection – where a computer-based system collaborates with the lesson’s author to generate clear and relevant lessons.
2) Verify the Lessons
Typically a team of domain experts performs the task required by this component, which requires the verification of lessons for correctness, redundancy, consistency, and relevance. The verification task is critically important, but sometimes introduces a significant bottleneck in the inclusion of lessons into the LLS, since it’s a time-consuming process. Some systems, like for example Xerox’s Eureka LLS, provide a two-staging process.
3) Store the Lesson
This task relates to the representation of the lessons in a computer-based system. Typical steps in this task include the indexing of lessons, formatting, and incorporating into the repository. In terms of the technology required to support this task, LLS could be based on structured relational or object-oriented databases as well as case libraries (case-based reasoning) or semi-structured document management systems. LLS can also incorporate relevant multimedia such as audio and video, which may help illustrate important lessons.
4) Disseminate the Lesson
This task relates to how the information is shared to promote its reuse. Six different dissemination methods have been identified:
a) Passive dissemination -- where users look for lessons using a search engine.
b) Active casting – where lessons are transmitted to users that have specified relevant profiles to that particular lesson.
c) Broadcasting – where lessons are disseminated throughout an organization.
d) Active dissemination – where users are alerted to relevant lessons in the context of their work (for example by a software help-wizard that alerts a user of related automated assistance).
e) Proactive dissemination – where a system anticipates events used to predict when the user will require the assistance provided by the lesson.
f) Reactive dissemination – when a user launches the LLS in response to a knowledge need, for example when he launches a Help system in the context of specific software.
5) Apply the Lesson
This task relates to whether the user has the ability to decide how to reuse the lesson. There are three categories of reuse:
a) Browsable – where the system displays a list of lessons that match the search criteria.
b) Executable – where users might have the option to execute the lesson’s recommendation (like when the Word processor suggests a specific spelling for a word).
c) Outcome reuse – when the system prompts users to enter the outcome of reusing a lesson, in order to assess if the lesson can be replicated.
Role of Taxonomies in Knowledge-Sharing Systems
Taxonomies, also called classification or categorization schemes, are considered to be knowledge organization systems that serve to group objects together based on a particular characteristic. Knowledge taxonomies are used to organize knowledge (or competencies) relevant to the organization. In the case of ELS, the knowledge taxonomy is used to describe the organization’s critical knowledge areas used to index people's knowledge.
🔑 Definition — Taxonomy: A classification or categorization scheme that groups objects together based on a particular characteristic; used to organize knowledge relevant to the organization.
Differentiating Characteristics of ELS at HP, NSA, and Microsoft
| Characteristic | CONNEX (HP) | KSMS (NSA) | SPuD (Microsoft) |
|---|---|---|---|
| Purpose of the system | To share knowledge, for consulting and to search for experts | To staff projects and match positions with skills | To compile the knowledge and competency of each employee |
| Self-Assessment | Yes | Yes, supervisors also participate in data gathering | No, supervisors rate employee's performance |
| Participation | Only those who are willing to share | Whole personnel | Whole personnel in the IT group |
| Knowledge Taxonomy | US Library of Congress | INSPEC Index Own | Department of Labor (O*NET) Own |
| Levels of Competencies | No | Yes | Yes |
| Data Maintenance | User (nagging) | User and Supervisor | Supervisor |
| Company Culture | Sharing, Open | Technology, Expertise | Technology, Open |
| Platform | HP-9000 Unix Sybase Verity | OS/2, VMS, and Programming Bourne shell SQL | MS Access |
Communities of Practice for Sharing Tacit Knowledge
A community of practice, also known as a knowledge network, is an organic and self-organized group of individuals who are dispersed geographically or organizationally but communicate regularly to discuss issues of mutual interest. Communities of practice are supported through technology that enables interaction and conversations amongst its members.
🔑 Definition — Community of Practice: An organic and self-organized group of individuals dispersed geographically or organizationally who communicate regularly to discuss issues of mutual interest, supported by technology enabling interaction and conversation.
Document Management Systems and Knowledge Sharing
Document management systems are composed of two pieces: a repository of documents and technology support for classifying, organizing, storing, and retrieving documents. The repository itself may be centralized or distributed and access points into the repositories are normally distributed. Most document management systems provide a knowledge portal that is a common, yet customizable, platform independent interface to distributed repositories. The common interface and support for finding information through document classification and organization and support for retrieval make finding information and getting information easier for individuals within the organization.
Barriers to Use of Knowledge Sharing Systems
A business culture may exist in which an organizational unit considers information from outside the unit as worthless. Typically this culture discourages knowledge consumers from participating in the knowledge market and organizational rewards are tied to creating knowledge (even when unnecessary) and not to sharing or knowledge re-use. Additionally, those organizations that separate knowledge from the knowledge owners/producers tends to discourage knowledge owners from volunteering knowledge to be shared. On the human side, knowledge sharing was anathema to traditional hierarchical organizations where knowledge was equated to power hence employees need to be properly motivated to engage in knowledge sharing initiatives. Without proper management support and motivation to share, knowledge sharing is not likely to occur.
💡 Why this matters: Understanding these barriers is critical because technology alone cannot make knowledge sharing successful—organizational culture and motivation are equally important.
Five Specific Types of Knowledge Sharing Systems
-
Incident report databases record and disseminate information related to either incidents or malfunctions.
-
Alert systems disseminate information regarding either negative or positive experiences.
-
Best practices databases describe successful (winning) efforts, typically regarding business process reengineering.
-
Lessons learned systems capture and disseminate lessons that benefit users that encounter similar situations.
-
Expertise locator systems catalogue knowledge competencies.
⭐ Key Takeaways
Corporate memory encompasses all explicit and tacit knowledge in an organization and KM systems are designed to prevent its loss. The lessons learned process has five critical stages—collection (with six methods from passive to interactive), verification (often a bottleneck), storage (with indexing and multimedia options), dissemination (with six methods including active casting and broadcasting), and application (browsable, executable, or outcome reuse). Taxonomies are essential for organizing knowledge and indexing expertise in systems like ELS. Three different ELS implementations—CONNEX at HP, KSMS at NSA, and SPuD at Microsoft—show differing approaches to purpose, self-assessment, participation, and data maintenance based on organizational culture. The five specific types of knowledge-sharing systems each serve distinct purposes: incident report databases, alert systems, best practices databases, lessons learned systems, and expertise locator systems.
🧠 Quick Revision Questions
-
What are the six methods for collecting lessons in a Lessons Learned System, and which is considered the most common?
-
Why is the verification step in the LLS process considered a potential bottleneck, and how did Xerox's Eureka system address this?
-
What are the three categories of lesson reuse in the Apply the Lesson step, and give an example of the "executable" category.
-
How did the purpose, self-assessment approach, and data maintenance differ between CONNEX (HP), KSMS (NSA), and SPuD (Microsoft)?
-
What are the five specific types of knowledge-sharing systems, and what is the primary purpose of each?