The accepted ontology
Browse accepted Systema Constructum entries: human-made tools, institutions, methods, practices, works, and their definitions.
100 entries in this page
Clustering is a human-made statistical method that partitions a set of objects into groups (clusters) such that objects within the same cluster are more similar to each other than to objects in other clusters. It operates through defined similarity or distanc…
A time series is a structured sequence of data points indexed in chronological order and recorded at successive, typically equally-spaced time intervals. It represents the human-made convention of organizing observations, measurements, or readings along a tem…
Principal Component Analysis (PCA) is a mathematical technique for dimensionality reduction that transforms a set of possibly correlated variables into a smaller set of uncorrelated variables called principal components. The first principal component captures…
Residual analysis is a diagnostic technique in statistics and machine learning where the residuals — the differences between observed values and the values predicted by a model — are examined to assess model adequacy. The procedure involves plotting residuals…
Partial correlation quantifies the linear relationship between two variables while controlling for the effect of one or more additional variables. It isolates the direct association between the pair by removing shared variance attributable to the conditioning…
Gradient boosting is a supervised learning algorithm that constructs an ensemble of weak predictive models (typically decision trees) in a stage-wise additive fashion. Each iteration fits a new model to the residual errors (negative gradients) of the current…
Logistic regression is a statistical method that models the probability of a binary outcome as a function of one or more predictor variables. It uses the logistic (sigmoid) function σ(z) = 1/(1+e^(-z)) to map a linear combination of inputs z = β₀ + β₁x₁ + ...…
A correlation matrix is a statistical data structure representing pairwise linear relationships among variables. It is a square symmetric matrix M where M[i,j] = r(Xi,Xj), the Pearson correlation coefficient between variables Xi and Xj, computed as the covari…
Kernel density estimation (KDE) is a non-parametric statistical technique for estimating the probability density function of a random variable from a finite data sample. It places a kernel function (typically a Gaussian bell curve) centered at each data point…
A precision-recall curve is a diagnostic visualization that plots precision against recall for a binary classifier across all possible decision thresholds. Each point on the curve corresponds to a specific threshold: recall (true positive rate) on the x-axis…
An empirical cumulative distribution function (ECDF) is a step function that estimates the cumulative distribution of a random variable from a finite sample of observed data points. For a sample of n ordered values x_(1) ≤ x_(2) ≤ ... ≤ x_(n), the ECDF at any…
A normal distribution is a continuous probability distribution characterized by a symmetric bell-shaped density function f(x) = (1/σ√(2π))·e^(-(x-μ)²/(2σ²)), fully specified by two parameters: the mean μ (center) and standard deviation σ (spread). It arises a…
Multiple testing correction is a family of statistical procedures that adjust significance thresholds or p-values when performing simultaneous hypothesis tests, controlling either the family-wise error rate or the false discovery rate. Its parameters are: the…
False discovery rate (FDR) is a statistical control method that limits the expected proportion of false positives among all rejected hypotheses in multiple testing. Its parameters are: the target FDR level (q), the set of simultaneous tests, the test statisti…
Resampling is a statistical procedure in which new datasets are generated by repeatedly drawing observations from an existing dataset or its empirical distribution. The core parameters are: (1) the source from which samples are drawn (the original dataset for…
A statistical method is a systematic procedure for collecting, analyzing, and interpreting data under conditions of uncertainty. It consists of a defined algorithmic sequence — assumptions, model specification, estimation or hypothesis testing, and validation…
A supervised learning paradigm is a machine learning approach in which a model is trained on a labeled dataset consisting of input-output pairs, learning a mapping function that generalizes to previously unseen inputs. The training process minimizes a loss fu…
An ensemble learning algorithm that constructs a collection of decision trees during training and outputs the average prediction across all trees. Each tree is trained on a bootstrapped sample of the data with random feature selection at each split, which dec…
A probability distribution is a mathematical function — discrete (mass function) or continuous (density function) — that assigns a non-negative value to each possible outcome in a sample space, subject to the constraint that the total sums to one (discrete) o…
Regression is a human-made family of statistical estimation methods that fit a mathematical model to observed data to quantify the relationship between a dependent (outcome) variable and one or more independent (predictor) variables, yielding fitted parameter…
A coxcomb chart (also called a polar area diagram or Nightingale rose) is a circular data visualization that divides the plane into sectors of equal angle, each with a radius proportional to the value it represents. It is designed for comparing quantities acr…
A ROC (Receiver Operating Characteristic) curve is a diagnostic plot used in binary classification evaluation. It maps each classification threshold to a point whose x-coordinate is the false positive rate (FPR = FP/(FP+TN)) and whose y-coordinate is the true…
A beeswarm plot is a human-made statistical visualization that draws every individual observation as a symbol placed at its true value along a primary axis while offsetting symbols along a perpendicular axis so that no two points overlap, rendering the empiri…
Gradient descent is an iterative optimization algorithm that minimizes a differentiable function by repeatedly computing the gradient (vector of partial derivatives) and stepping in the negative direction scaled by a step size (learning rate). Key parameters…
A slope graph is a data visualization composed of discrete categories arranged along an axis, with two (or more) time points or states represented as columns of values, and straight lines connecting the values within each category to show direction and magnit…
A confusion matrix is a contingency table that cross-tabulates predicted classification outcomes against actual ground-truth labels. Its rows enumerate the true classes and its columns the predicted classes; the four cells are true positives (TP), true negati…
Outlier: a data point that lies an abnormal distance from other values in a dataset. Detected via statistical methods (e.g., IQR rule: beyond 1.5×IQR from Q1/Q3; z-score: beyond ±3σ), visual inspection (boxplot, scatter), or robust estimation. Outliers flag m…
Concept drift is the phenomenon whereby the statistical properties of a target variable change over time relative to the variables used to predict it, degrading model performance. In practice it manifests as covariate shift (the input distribution P(X) change…
A grouped bar chart is a data visualization technique that organizes quantitative data into rectangular bars arranged in clusters, where each cluster corresponds to a category on the horizontal axis and each bar within the cluster represents a subcategory or…
A frequency distribution is a statistical representation — either tabular, graphical, or mathematical — that shows the number of observations falling within each of several non-overlapping categories or value ranges. Its parameters are: (1) a variable (contin…
Pseudocode is an informal, high-level description of the operating principle of a computer program or algorithm. It uses the structural conventions of programming languages and structured natural language mixed with math notation, omitting details unimportant…
A pivot table is a data summarization mechanism that reorganizes tabular data by grouping records along one or more dimension fields and computing aggregate functions (count, sum, average, min, max, etc.) on measure fields, producing a two-dimensional summary…
An embedding is a fixed-length numeric vector produced by a neural network (encoder) that maps discrete inputs — tokens, images, or structured records — into a continuous metric space. The mapping is trained to preserve relational structure: inputs sharing se…
A strip plot is a human-made visual encoding method that displays individual data points along a single axis (typically vertical), each positioned at its measured value to show the full distribution of a single variable. Overlapping points are jittered horizo…
A choropleth map is a human-made thematic visualization that divides geographic regions into discrete zones and color-codes each zone according to a statistical variable, enabling visual comparison of region-level data such as density, rate, or magnitude. The…
A parallel coordinates plot is a visual encoding method for multivariate data where each variable is represented as a vertical axis laid out in parallel, and each data item is drawn as a polyline connecting values across all axes. Parameters: a set of ordered…
A data pipeline is a human-made system of sequential processing stages that move, transform, and filter data from source to destination. Each stage performs a discrete operation — ingestion, cleaning, transformation, enrichment, or delivery — connected by exp…
Cross-validation is a statistical procedure for estimating how well a predictive model generalizes to unseen data. It partitions a dataset into k subsets (folds), iteratively trains the model on k-1 folds and validates on the remaining fold, then aggregates t…
A cross-check is a verification method that confirms the correctness of a result by comparing it with an independently obtained or computed value. Its parameters: (1) two or more independent computation paths — different methods, tools, or assumptions that ea…
A pedagogical cascade model in organizational safety where practitioners certified in safety protocols, hazard recognition, and risk management design and deliver instructional programs to their peers. The practice operates through certified trainer pools, st…
A belief system is a structured set of propositions, values, and assumptions that a person or community accepts as true and uses to interpret experience and guide action. Its parameters: (1) internal coherence — beliefs are linked such that challenge to one p…
A safety-narrative is a deliberately constructed story or narrative frame used within an organization to communicate, interpret, and transmit safety-related meaning. It takes raw safety events — incidents, near-misses, hazard observations, compliance outcomes…
A data structure is a method of organizing information in a computational context such that the data can be accessed, traversed, and modified efficiently. It is defined by three aspects: (1) the logical arrangement of data elements and their interrelationship…
A solar chart is a circular diagram that maps the position of the sun across the sky for a given latitude, showing solar altitude and azimuth as concentric circles and radial lines respectively, with seasonal curves tracing the sun's path from dawn to dusk ac…
A stem-and-leaf plot is a data visualization method that displays the distribution of a quantitative dataset while preserving each individual data value. Each observation is split into a stem (one or more leading digits) and a leaf (the trailing digit); stems…
A pictogram is a symbolic representation that uses pictures, icons, or simplified drawings to convey information, instructions, or concepts without relying on written language. Parameters: (1) the symbol must be pictorial (recognizable as a picture of its ref…
A creed is a systematically formulated statement of core beliefs and principles, authored to define group identity and guide collective conduct. Parameters: (1) it must articulate shared convictions, not merely individual opinions; (2) it must serve as an exp…
A rug plot is a data visualization technique that encodes individual data observations as small tick marks along a single axis, preserving raw data density and distribution without binning or aggregation. It operates as a complement to binned displays (histog…
Safety-perception is the structured capacity of individuals and groups to detect, interpret, and prioritize safety-relevant information within their environment. Parameters: (1) detection threshold — the minimum signal strength or novelty required for a hazar…
A structured taxonomy enumerating the distinct mechanisms by which a system, process, or component can fail. Parameters: (1) failure mode — the specific way in which a component ceases to perform its intended function; (2) failure mechanism — the physical or…
A fire drill is a structured, simulated emergency exercise designed to train occupants in evacuation procedures and emergency response. Its parameters include a trigger (scheduled or unscheduled), predefined evacuation routes, assigned roles (monitor, marshal…
A structured approach to presenting information that deliberately selects and emphasizes certain aspects of a situation while omitting or downplaying others, thereby shaping how recipients interpret, evaluate, and respond to it. Parameters: (1) a source frame…
A visual communication artifact that combines graphic design, data visualization, and concise text to convey information, knowledge, or understanding efficiently. It carves a bounded form: intentional layout of visual elements (icons, charts, typography) arra…
A cognitive framework is a structured set of concepts, rules, and mental models that humans create to interpret, organize, and process information. Parameters: (1) framework — a coherent structure of interrelated concepts and the rules for their application;…
A human-made practice in which an organization systematically evaluates its own safety posture by comparing actual safety performance, controls, and compliance against established standards, objectives, and regulatory requirements. Parameters: (1) internal ev…
A systematic organizational practice for identifying potential threats to an operation and building capabilities — documented procedures, resource inventories, and recovery protocols — to ensure that critical functions continue during and after a disruption.…
Safety routine is a scheduled, recurring activity designed to maintain or improve safety performance. It is characterized by: (1) fixed frequency — performed at regular intervals (daily, weekly, monthly); (2) standardized procedure — following a documented ch…
Safety incident reporting is the structured practice of documenting, escalating, and communicating information about accidents, near-misses, hazards, and safety violations within an organization. It operates through defined channels (forms, hotlines, digital…
Organizational resilience is the structured capacity of an institution to absorb disruption, adapt under stress, and recover while maintaining core functions. It comprises four mechanisms: (1) redundancy — spare capacity built into critical systems; (2) modul…
A just culture is an organizational framework for incident response that classifies behaviors into three categories: human error (unintentional mistakes, addressed by improving systems), at-risk behavior (choices where benefits outweigh perceived risks, addre…
Psychological safety is the constructed measure of shared group belief regarding interpersonal risk-taking — specifically, whether members believe they can speak up, admit errors, propose novel ideas, or ask questions without fear of humiliation, rejection, o…
A safety-risk-assessment is a human-made systematic process that identifies potential hazards, evaluates the likelihood and severity of harm they could cause to people, property, or the environment, and determines appropriate controls to mitigate those risks.…
A safety-strategy is a human-made strategic plan that establishes the direction, priorities, resource allocation, and accountability framework for achieving and maintaining safety within an organization or system. It persists through documented strategy state…
A safety-barometer is a human-made measure or gauge designed to indicate the relative state or level of safety within a system, organization, or domain. It operates by aggregating discrete safety signals — indicators, metrics, and observations — into a compos…
Safety-practice is a specific action or method performed or sustained to advance safety outcomes. Parameters: (1) the actor intends or is trained to perform it for safety purposes, (2) it has repeatable observable behavioral components, and (3) it is sustaine…
A safety-promotion is a human-made organized effort designed to improve safety awareness, culture, and practices within an organization. It comprises planned activities such as safety campaigns, communication channels, recognition programs, leadership messagi…
Safety-maintenance is a human-made system of planned and reactive activities designed to preserve and restore the operational readiness of safety-critical equipment, procedures, and organizational safety structures. Its parameters: (1) planned maintenance on…
A safety-audit-report is a formal document produced through a systematic safety audit that records findings, identified non-conformances, observations of good practice, and recommended corrective or preventive actions. Its parameters are: (1) scope — the syst…
A safety-hierarchy is a structured ranking of control methods ordered by their demonstrated capacity to reduce or eliminate hazard exposure. Its parameters are five tiers: (1) elimination — physically removing the hazard, (2) substitution — replacing the haza…
A safety-suggestion is a formalized channel through which frontline participants in an organization can submit observations, concerns, or proposals related to safety conditions and practices. It operates as a bottom-up feedback mechanism that converts individ…
Safety-knowledge is the body of information, practices, and understanding about safety that an organization collects, maintains, and applies. It encompasses documented procedures, incident reports, lessons learned, regulatory requirements, best practices, and…
A chain of thought is a structured reasoning method where an agent or system generates intermediate reasoning steps before producing a final answer. It persists through explicit step-by-step articulation of the reasoning process, enabling verification, debugg…
Model interpretability is a systematic method for understanding and explaining how artificial intelligence models produce their outputs. It comprises a family of techniques—such as feature attribution, surrogate modeling, activation analysis, and attention vi…
A control tower is a centralized governance function that integrates real-time data collection, performance monitoring, and directive authority over distributed operations. Its parameters are: (1) a dedicated monitoring function that aggregates operational da…
A best-practice is a method, technique, or process that has been demonstrated through repeated application to produce superior, reliable results, and is formally recognized and adopted as the standard within a discipline or organization. Its parameters are: (…
A safety-eval is the structured practice of assessing safety management systems, controls, and operational performance against defined standards and benchmarks. It operates through scheduled audits, inspections, and reviews that produce documented findings an…
Safety-forecasting is a systematic practice of using data, models, and leading indicators to predict future safety performance trends and potential incidents before they occur. It extends safety monitoring by projecting forward: trend analysis, predictive mod…
Feedforward is a control mechanism in which a system anticipates disturbances and adjusts its output before those disturbances affect the controlled variable. It operates by measuring the input or an external signal that predicts a disturbance, then applying…
Safety-lagging-indicator is a retrospective safety metric that quantifies past organizational safety performance by measuring recorded incidents, injuries, near-misses, and system failures after they occur. It operates through four mechanisms: (1) incident re…
Algorithmic audit is a structured examination procedure that systematically evaluates automated systems for compliance with legal, ethical, and organizational standards. It operates through four mechanisms: (1) impact assessment documenting system purpose and…
Organizational-learning is the practice by which an organization systematically converts individual and collective experience into retained knowledge that alters future behavior. It operates through four mechanisms: (1) knowledge capture — structured recordin…
Algorithmic accountability is the governance practice of assigning responsibility for the outcomes of automated systems to the organizations and individuals that design, deploy, and operate those systems. It operates through three parameters: (1) traceability…
Algorithmic transparency is the practice of making the design, operation, and outcomes of automated systems understandable and accessible to affected parties and overseers. It requires that system behavior be explainable through documentation, audit trails, a…
A prompt template is a pre-structured text format containing placeholders, variable markers, and usage instructions for generating AI system inputs. Parameters: (1) a fixed skeleton with semantic slots (e.g. [role], [task], [context], [output-format]), (2) pl…
Automated-decision-making is the structured practice of delegating decision authority to algorithmic systems — software, rulesets, statistical models, and machine learning outputs — that process inputs according to predefined logic and produce decisions witho…
whistleblower-protection is a set of legal, organizational, and social mechanisms designed to safeguard individuals who report wrongdoing, illegality, or danger within an institution. It specifies parameters: protected reporters (employees, contractors, obser…
A safety-improvement IS a specific kind of systematic change process — an organized effort that identifies gaps in safety performance, designs and implements interventions, and sustains gains through measurement and feedback. Defined by four phases: assessmen…
A safety feedback loop IS a specific kind of information mechanism — a structured pathway that routes safety-relevant data from detection points (incidents, near-misses, audit findings) back into the design, training, and procedural systems that prevent futur…
A safety bottleneck IS a specific kind of constraining factor — a limited resource, procedural gap, or organizational weakness that caps overall safety performance regardless of strengths elsewhere. Defined by identifying the minimum-capacity link in the safe…
data-ethics is the structured practice of applying moral principles to the collection, processing, storage, and use of data. It establishes parameters for what constitutes fair, transparent, accountable, and privacy-respecting data practices — distinguishing…
A control-framework is a human-made institutional construct consisting of a structured set of guidelines, standards, procedures, and controls that govern and direct an organization's operations. Parameters: (1) scope — the domain it covers (security, finance,…
A human-made fiduciary role and practice in which individuals or groups assume responsibility to care for, maintain, and improve the safety conditions of a workplace, system, or community — extending beyond their own immediate duties to look after the collect…
A human-made organizational practice that grants workers authority, resources, and obligation to stop work when unsafe, address identified hazards, and influence safety decisions without fear of reprisal. Parameters: (1) the grant of authority comes from orga…
A lead-indicator is a human-made measure: a quantitative or qualitative signal extracted from proactive activities, processes, or conditions that predict future outcomes or performance trends. Parameters: (1) the measured activity or condition (e.g., training…
A business model is a human-made construct that describes how an organization creates, delivers, and captures value. It specifies the customer segments served, the products or services offered, the revenue and cost structure, and the key activities, resources…
Open-source is a human-made model of collaborative creation governed by explicit permission grants. Parameters: (1) source material (software, design, content) made publicly accessible, (2) legal licenses granting rights to use, modify, and redistribute, (3)…
Safety-engagement is a human-made organizational practice: the active, voluntary participation of workers and stakeholders in safety-related activities beyond baseline compliance. It encompasses volunteering safety ideas, attending safety meetings, participat…
A smart contract is a human-made protocol for automated contractual execution: a self-enforcing agreement encoded as deterministic software on a distributed ledger, where predefined conditions trigger automatic performance of obligations without intermediary…
A safety-observance is the ongoing practice of conforming to established safety rules, standards, and procedures within an organization or system. It carves a behavioral norm — distinct from safety-compliance (the enforcement/verification mechanisms) and safe…
A safety-commission is a formally constituted regulatory body with statutory or delegated authority to develop safety standards, conduct oversight of safety practices, investigate safety failures, and enforce compliance across regulated domains. Its parameter…