User Guide

Step-by-step guides for each module of the library.

• Knowledge Structures and Relations
  • Surmise relations
  • Knowledge structures
  • Surmise functions
  • Exact property checks and closures
• Combining structures, skills and assessment
  • Neighbourhoods and paths
  • Maximal mesh
  • Inverting a skill multimap
  • Multiplicative assessment
• Attributions, families and operations on spaces
  • From an attribution to a surmise function
  • Families and closures
  • Combining compact representations
  • Refining equivalent items
• Structural reports and paths
  • Item heights and compact neighbourhoods
  • Paths, required states, and matrices
  • Changes involving several items
  • Irredundant generators and the empty-operation convention
  • Reference differences
• Querying Experts
  • How it works
  • Basic usage
  • Controlling the algorithm
  • Inspecting results
• Competence-Based KST (CbKST)
  • The model
  • Skill multimaps: alternative competencies
  • Usage
  • Skill-to-item conversion
  • Alternative: surmise functions
  • Deriving structures from data: IITA
  • Effective fringes and observable learning
  • ITA and IITA inference
• Problem functions on arbitrary competence families
  • Keep rows, column order and identities
  • Exact preimages and positive coverage
  • Explicit endpoints and CbKST correspondence
  • Enumeration limits
• Course-dependent skill structures
  • Build and inspect a course
  • What is derived
  • Static compatibility and executable lessons
  • Interchange and computational limits
• Adaptive Assessment
  • Deterministic half-split assessment
  • The BLIM model
  • One-shot assessment
  • Adaptive assessment
  • Low-level control
  • The canonical informative rule
  • Complete sessions and multiplicative updating
  • Evaluate observed responses versus simulate latent states
• Parameter Estimation
  • Predicting states after fitting
  • What it estimates
  • High-level API
  • Low-level API
  • How it works
  • Estimation methods: ML, MD, MDML
  • Fixed and shared parameters
  • Simulating data from a BLIM
  • Identifiability diagnostics
  • Classic datasets
  • Random restarts
  • Degenerate items
  • Goodness of fit and model selection
  • Additional classical estimators
• Random starts and reproducible refits
  • Initialization laws
  • Incomplete BLIM search
  • Bootstrap policy
  • Verification scope
• Bootstrap statistics and observation designs
  • Refit policy and unresolved fits
  • Weighted empirical independent masks
  • Pearson with missing responses: a conditional claim
• Statistical reports, discrepancy and expected errors
  • Marginal mastery and full-test errors
  • Minimum discrepancy distribution
  • Training quantities and other data
  • AICc convention and limits
  • Trade-offs and exports
• SLM, ITA and IITA inference
  • Simple learning model
  • Random SLM starts and search diagnostics
  • Discrepancy inclusion options
  • Threshold item tree analysis
  • Quasi-ordinal simulation and supplied IITA candidates
  • Population IITA and fixed-relation inference
  • References
• Observed frequencies and equivalent items
  • Frequency reports
  • Empirical structure at a frequency threshold
  • Reducing a quasi-order
• Incomplete responses and observed-data estimation
  • Explicit data, explicit assumptions
  • EM updates and boundaries
  • Prediction, saturated likelihood and goodness of fit
  • Simulation and bootstrap
  • Probability individual data
• Empirical validation and BLIM diagnostics
  • Distances and prerequisite validation
  • Complete response-table residuals
  • Bootstrap parameter replicates
  • Parameter-family trade-offs
  • Compare probabilistic models on common data
• Prediction and reliability
  • Entropy measures (2024)
  • Modal classification measures (2025)
  • Boundaries, cost and interpretation
• Bundled Datasets
  • Doignon & Falmagne (1999, chapter 7)
  • Using a dataset
  • Probability: Anselmi & Wickelmaier
  • Individual probability records and pairing
  • PISA 2003: DAKS response data
  • Selection from the comparison packages
• Import / Export
  • CSV formats
  • JSON
  • Validation
  • Skill multimaps
  • General clauses and families
  • KST and SRBT legacy text
  • Course assignments
• Patterns, matrices and named sets
  • Frequencies and expansion
  • Inclusion, paths and fringes
  • Legacy KST compatibility
• Reference workflows through explicit compositions
  • Complete-data ML with parameter-change stopping
  • Smallest quasiordinal completion
  • Raw curriculum relations before closure
• Spreadsheet and table interchange
  • Read and write KST objects
  • Frequencies and missing responses
  • Multiple objects and curriculum workbooks
  • Formula results and limits
• Comparing Structures
  • Distance between structures
  • Agreement on prerequisites
  • Domain validation
• Visualization
  • Hasse diagram of a structure
  • Paths, neighbourhoods and numerical graph export
  • Hasse diagram of a surmise relation
  • Clauses and atomic items
  • Coefficients and bootstrap distributions
• Command Line Interface
  • ks inspect — examine a KST file
  • ks query — derive a structure from expert knowledge
  • ks assess — adaptive student assessment
• MCP Server
  • What the server provides
  • Installation
  • Configuration
  • Usage examples
  • Architecture
• Scope, assumptions and computational limits
  • Different kinds of limitation
  • Structural conventions
  • Statistical conventions
  • Evidence and extension policy