PKMC Module II flashcards

33 cards. Read the term and the definition here. The deck at the top of this page is optional.

Tacit knowledge

Personal, experience-based knowledge such as judgment, skill, and intuition that is hard to articulate or codify; it transfers mainly through observation, apprenticeship, and direct collaboration rather than documents.

Explicit knowledge

Knowledge that has been articulated and recorded in documents, databases, procedures, or models, making it transferable without the original knower being present; it still requires context to be interpreted correctly.

SECI model

Nonaka and Takeuchi's four knowledge conversion modes: socialization, externalization, combination, and internalization. Each conversion needs a shared context (ba) where people interact and meaning is negotiated.

Codification versus personalization

Hansen's two KM strategies: codification reuses knowledge through repositories and person-to-documents links for standardized problems; personalization shares knowledge person-to-person for novel, judgment-heavy problems.

Absorptive capacity

Cohen and Levinthal's concept that prior related knowledge determines how well an organization recognizes, assimilates, and applies new external knowledge; built through research, training, and experience.

Knowledge strategy fit

The alignment of KM approach with competitive strategy and task type: standardized, repeatable work favors codification and reuse economics, while bespoke advisory work favors personalization and expert networks.

Critical knowledge risk ranking

A prioritization formula multiplying the business impact of losing a knowledge asset by the likelihood of loss, then dividing by the ease of replacing it; high scores trigger capture or retention action.

After-action review

A structured team debrief immediately after an activity asking what was expected, what happened, why the difference, and what to change; it captures reasoning while memory and context are fresh.

Cognitive task analysis

A family of interview methods that elicit the cues, strategies, anomalies, and decisions experts use when performing complex tasks; it captures expert reasoning rather than only procedural steps.

Exit interview for knowledge

A structured session held well before a knowledgeable employee departs to extract undocumented expertise, key contacts, decision rationale, and risks; scheduled months ahead, not on the last day.

Codification quality test

A captured artifact passes only if a competent outsider can act on it without contacting the author; otherwise it is a pointer to expertise, not reusable codified knowledge.

Knowledge retention plan

A documented set of actions for at-risk expertise: successor assignment, mentoring, job rotation, shadowing, and scheduled handover reviews, each with an owner and a completion date.

Community of practice

Wenger's group defined by three elements: a shared domain of interest, a community built on relationships and trust, and a practice of shared repertoires, tools, and stories.

Legitimate peripheral participation

Lave and Wenger's model in which newcomers learn a practice by performing peripheral but genuine tasks, gradually moving toward full participation as competence and trust grow.

Structural holes

Burt's gaps between groups that lack direct ties; individuals who broker across those gaps gain information advantage, and organizations benefit when brokers share what they learn.

Social network analysis metrics

Degree counts direct ties, betweenness identifies brokers, density measures group cohesion, and structural holes reveal gaps; together they expose isolated experts and single points of failure.

Psychological safety

A shared belief that raising questions, admitting error, and asking for help will not be punished; it predicts team learning and knowledge-sharing behavior more strongly than individual expertise.

BM25 ranking

A probabilistic retrieval scoring function summing, per query term, inverse document frequency times a saturated term-frequency factor normalized by document length; parameters k1 and b tune saturation and length effects.

Precision and recall

Precision is relevant results divided by all returned results; recall is relevant results divided by all relevant items in the collection. F1 is their harmonic mean, balancing the two.

Knowledge graph triple

A subject-predicate-object statement, such as a person authored a document, stored in an RDF graph; ontologies define the classes and properties, and SPARQL queries the resulting graph.

Semantic search trade-off

Embedding or ontology-based retrieval improves recall for synonyms and paraphrase but can lower precision; a relevance threshold, hybrid keyword matching, or reranker restores precision.

Faceted search

A retrieval interface exposing metadata dimensions such as author, date, type, and topic so users narrow results progressively; facets derived from a controlled vocabulary remain consistent across content.

Controlled vocabulary spectrum

Increasing semantic control across term lists, synonym rings, taxonomies with hierarchy, thesauri with broader, narrower, and related links, and ontologies with classes, properties, and logical axioms.

Taxonomy design rules

One preferred term per concept, synonyms recorded as non-preferred entry terms, mutually exclusive sibling nodes, and a single primary parent per node to prevent ambiguous retrieval and navigation.

Metadata types

Descriptive metadata supports discovery (title, creator, subject); structural metadata describes format and relationships; administrative metadata covers rights, provenance, retention, and access controls.

SKOS

The Simple Knowledge Organization System, a Semantic Web standard for representing thesauri, taxonomies, and classification schemes using concepts, preferred and alternative labels, and broader, narrower, and related links.

Taxonomy governance workflow

A named term owner, a change-request and review process, version control, deprecation with redirects, and mapping tables when terms are merged or split; without it, vocabulary drifts from user language.

KM return on investment

Benefit minus cost, divided by cost; benefits such as avoided rework hours, faster onboarding, and reduced search time must be converted to currency with stated assumptions to be auditable.

Kirkpatrick evaluation levels

Four levels for evaluating learning and KM interventions: reaction, learning, behavior change, and results. Most KM reporting stops before behavior and results, overstating realized value.

Leading versus lagging indicators

Leading indicators such as contributions, reuse, and community activity predict future outcomes; lagging indicators such as cycle time, error rate, and time-to-competence prove whether value was realized.

Data classification tiers

Categories such as public, internal, confidential, and restricted that determine access control, encryption, and sharing rules; classification should be assigned when content is created, not retrofitted.

Legal hold

A directive suspending normal records disposition for identified custodians and systems because of anticipated litigation or investigation; destroying held content is spoliation and can draw sanctions.

Retrieval-augmented generation

A pipeline that chunks and embeds source content, retrieves top-k passages by similarity for a query, reranks them, and grounds a language model answer in those passages with citations.