Skip to content

Concept Glossary: C

This page lists English glossary entries for this letter. Entry bodies are assembled from term source files.

caching

  • Meaning: Caching stores intermediate inputs, lookup results, or computed outputs so they can be reused later instead of recomputed from scratch. It is an operational choice between calculating again and reusing a result that is already available.
  • Why it matters: Caching can reduce both cost and latency when the same prompt fragments, retrieval results, or computations recur. It also creates a freshness risk, so it must be read as a balance between performance and recency rather than as a simple speed trick.
  • Related concepts: cost, latency, batch
  • Core Section: P1-14.6

calibration

  • Meaning: Calibration checks whether a model probability score matches real observed frequency. If cases scored around 0.70 are correct about 70 percent of the time, the score is better calibrated.
  • Why it matters: A high score is not automatically a reliable probability. Calibration matters when scores are turned into human review rules, automatic approval thresholds, or risk alerts because the operating decision depends on whether the number can be trusted as a probability.
  • Related concepts: probability, classification, evaluation
  • Core Section: P1-6.2
  • Appears in: P1-6.3, P1-7.3, P4-6.4, P4-11.1, P4-15.3

Chain-of-thought, CoT

  • Meaning: Chain-of-thought is a prompting strategy that asks a model to expose intermediate reasoning steps instead of only giving the final answer. It tries to make the path to the conclusion easier to inspect.
  • Why it matters: CoT can help with multi-condition reasoning tasks, but visible reasoning does not guarantee factual correctness. The concept helps separate checking an answer path from verifying external evidence.
  • Related concepts: prompt engineering, self-consistency, alignment
  • Core Section: P6-10.3

chunk

  • Meaning: A chunk is a smaller passage or text unit cut from a longer document for search, retrieval, or comparison. It repackages a document into units that can be matched more directly with a question.
  • Why it matters: RAG quality depends not only on the embedding model but also on how documents are chunked. Chunks that are too small can lose context, while chunks that are too large can mix in noise and make relevant passages harder to retrieve.
  • Related concepts: embedding, document, retrieval-augmented generation, RAG
  • Core Section: P1-13.1
  • Appears in: P1-13.2, P1-13.3, P1-14.2, P6-12.1, P6-12.2

class

  • Meaning: A class is a blueprint that defines what attributes and methods a kind of object should have. Individual objects are concrete instances made from that definition.
  • Why it matters: Classes help readers understand why similar objects share behavior and structure. This makes library objects such as DataFrame, Tokenizer, and Model easier to read as instances of a defined type rather than as isolated names.
  • Related concepts: object, method, attribute
  • Core Section: P2-8.6
  • Appears in: P2-9.1, P2-12.3

classification

  • Meaning: Classification is a problem type where an input case is assigned to one of a predefined set of categories or labels. It chooses from known options instead of generating a new free-form output.
  • Why it matters: Many practical AI tasks, such as spam detection, anomaly alerts, and approval decisions, become classification questions. The concept also separates categorical prediction from regression, clustering, generation, and threshold-based operating decisions.
  • Related concepts: clustering, prediction, label, regression, threshold
  • Core Section: P1-8.1
  • Appears in: P1-1.2, P5-15.1

client

  • Meaning: A client is the component that sends requests to a server or service and receives responses. A browser, API caller, or desktop app can act as the requesting side of a system.
  • Why it matters: The client/server distinction clarifies who starts a request and who handles it. In MCP-style systems, it also helps separate the user-facing app, the connection-opening client, and the server that provides tools or resources.
  • Related concepts: host, server, Model Context Protocol, MCP, application
  • Core Section: P1-14.4
  • Appears in: P1-14.6, P6-11.1

cluster label

  • Meaning: A cluster label is the number or marker assigned by a clustering algorithm to distinguish one discovered group from another. The label is usually an identifier, not a human meaning.
  • Why it matters: Cluster labels are easy to confuse with supervised-learning labels. Readers need to interpret each group after clustering before treating a numeric label as a meaningful category name.
  • Related concepts: clustering, label, unsupervised learning
  • Core Section: P1-8.2

clustering

  • Meaning: Clustering is an unsupervised problem type that finds groups of similar items in data without predefined answer labels. It asks how the data seems to separate by its own patterns.
  • Why it matters: Clustering shows that learning can be about discovering structure rather than matching known answers. The resulting groups still require human interpretation, because the algorithmic group number does not by itself explain the group meaning.
  • Related concepts: classification, prediction, dataset
  • Core Section: P4-17.1
  • Appears in: P1-1.2, P1-8.2, P4-2.2, P4-17.2, P4-18.2

CNN, convolutional neural network

  • Meaning: A convolutional neural network is a neural-network structure that repeatedly reads local visual patterns and combines them into higher-level representations. Early layers can detect simple edges or textures, while later layers combine larger parts and object-level cues.
  • Why it matters: CNNs are a central example of learned hierarchical representation in image recognition. They explain why local pattern reading, convolution, and pooling became powerful alternatives to treating an image as one flat fully connected input.
  • Related concepts: image recognition, learned representation, AlexNet
  • Core Section: P1-9.1
  • Appears in: P5-11.1, P5-11.2, P5-11.3

code cell

  • Meaning: A code cell is the executable cell in a notebook. It may look like a document block, but running it can change variables, memory, outputs, and runtime state.
  • Why it matters: Code cells explain why notebooks differ from plain scripts. The visible order of cells may not match the actual execution order, so reproducible notebook work requires tracking which cells ran and what state they created.
  • Related concepts: notebook, markdown cell, output
  • Core Section: P2-10.1
  • Appears in: P2-10.2, P2-10.3, P2-11.2

Colab

  • Meaning: Colab is Google's hosted Jupyter notebook environment that runs notebooks in a browser. It lowers setup effort, but the runtime resources live in a temporary session outside the local computer.
  • Why it matters: Colab makes practice easier to start, while also introducing constraints around runtime resets, file persistence, paths, and package state. Readers need this distinction to explain why code can behave differently in hosted and local environments.
  • Related concepts: Jupyter, runtime, reproducibility
  • Core Section: P2-10.2
  • Appears in: P2-3.5, P2-7.1, P2-10.1

column

  • Meaning: A column is the vertical set of values in a table that represents one variable, feature, attribute, state, or possible target. In a DataFrame, columns are the main units selected, transformed, compared, and aggregated.
  • Why it matters: Many tabular-data decisions begin by asking what each column means. Understanding columns helps separate input features, metadata, target candidates, and grouping variables inside the same table.
  • Related concepts: row, DataFrame, feature
  • Core Section: P2-12.1
  • Appears in: P2-12.2, P2-12.3, P2-15.2

combinatorial explosion

  • Meaning: Combinatorial explosion is the rapid growth of possible choices or paths until checking all cases becomes impractical. The difficulty comes from the structure of the search space, not just from slow implementation.
  • Why it matters: This concept explains why exhaustive checking often breaks down and why heuristics, pruning, and approximate strategies matter. A small increase in problem size can make exact enumeration unrealistic within available time and resources.
  • Related concepts: search, heuristic, optimization
  • Core Section: P1-6.1
  • Appears in: P1-7.1

commit

  • Meaning: A commit is a recorded unit of meaningful change saved to repository history with a message. It captures what changed and why at a particular point in the project.
  • Why it matters: Saving files alone does not explain intent or scope. Commits make later review, comparison, rollback, and collaboration possible by turning work into inspectable change units.
  • Related concepts: Git, staging area, repository
  • Core Section: P2-14.1
  • Appears in: P2-14.2

comparison report

  • Meaning: A comparison report is an output structure that summarizes differences between a recent state and a baseline so a person can decide what to inspect first. It is a review-oriented output, not an automatic final decision.
  • Why it matters: A comparison report is useful when data or labels are not ready for a direct prediction task. It helps separate “let the model decide” from “let a person read the differences first.”
  • Related concepts: baseline, review, target, output structure
  • Core Section: P3-9.2
  • Appears in: P3-index, P3-1.1, P3-3.2, P3-8.2, P3-9.1, P3-9.3, P3-9.4, P3-9.5, P3-9.6, P3-summary

comparison result

  • Meaning: A comparison result is the actual difference observed when a baseline and current value, baseline model and improved model, or old and new setting are placed under the same comparison conditions.
  • Why it matters: Without a comparison result, “better” can remain an impression. The concept connects scores, samples, tables, and execution summaries into a record of what changed and where.
  • Related concepts: comparison table, baseline, execution summary, next question, evaluation
  • Core Section: P7-1.2
  • Appears in: P7-index, P7-2.2, P7-summary

comparison table

  • Meaning: A comparison table places two or more targets, such as models, settings, samples, or periods, on shared columns so their differences can be read. It records what is being compared and along which axes.
  • Why it matters: Improvement claims only make sense relative to a baseline and comparison unit. A comparison table keeps those units visible, so retrospectives and next experiments can be based on concrete differences rather than impressions.
  • Related concepts: baseline, error case, retrospective, review, evaluation metric
  • Core Section: P7-2.1
  • Appears in: P7-index, P7-1.1, P7-1.3, P7-2.2, P7-5.1, P7-summary

computation graph

  • Meaning: A computation graph represents operations and value dependencies as connected nodes, making forward and backward calculation paths visible. It breaks a large formula into smaller calculation blocks and their connections.
  • Why it matters: Computation graphs make backpropagation and automatic differentiation easier to understand. They show which intermediate values depend on which earlier operations and how gradient signals can flow backward through the graph.
  • Related concepts: backpropagation, gradient, automatic differentiation
  • Core Section: P5-5.2
  • Appears in: P5-6.1

computational limit

  • Meaning: A computational limit is the point where the candidate space grows too large to inspect every case within realistic time and resources. The issue is not merely that the computer is slow; the search space itself becomes too large.
  • Why it matters: Computational limits explain why complete search can be possible in theory but unusable in practice. They motivate heuristics, pruning, priority strategies, and approximate solutions in real systems.
  • Related concepts: search, search space, exhaustive search
  • Core Section: P1-7.1
  • Appears in: P1-index, P1-6.1, P1-7.2, P1-7.4, P1-8.1

confabulation

  • Meaning: Confabulation is a generation failure where a model confidently constructs content that sounds factual but lacks support. It is fluent fabrication rather than a simple typo or broken sentence.
  • Why it matters: The concept helps distinguish natural-sounding text from grounded text. It explains why source checking and human review remain necessary even when an answer is smooth and even when retrieval has provided some candidate evidence.
  • Related concepts: hallucination, evidence, factuality, review
  • Core Section: P1-10.3
  • Appears in: P1-12.3, P1-13.2, P7-5.2

confidential information

  • Meaning: Confidential information is internal information that may not be legal personal data but can harm an organization, customer, contract, or operation if exposed. Draft contracts, unreleased code, customer lists, secrets, and launch plans are examples.
  • Why it matters: Generative AI inputs and logs can copy or expose organizational information as well as personal data. The concept prevents the mistake of treating non-personal internal material as safe to paste into any tool.
  • Related concepts: sensitive information, security, privacy, log, permission
  • Core Section: P1-15.3
  • Appears in: P1-14.5, P1-15.2, P1-16.2

consistency

  • Meaning: Consistency is the property that outputs keep similar judgment standards and expression direction under similar conditions. It is about stability of criteria, not just the tone of a single answer.
  • Why it matters: One good-looking output is not enough to judge quality. Consistency asks whether similar inputs receive decisions based on the same standard, which is especially important for generative AI evaluation and repeated use.
  • Related concepts: evaluation, reproducibility, limit of prompting, quality, human evaluation
  • Core Section: P1-12.3

generalization

  • Meaning: Generalization is the state in which a model captures a relationship that still works reasonably well on data it did not see during training. It means the model has learned a broader pattern instead of merely memorizing the training examples.
  • Why it matters: Generalization is the key difference between learning and memorization. It explains why validation and test data are needed even when training scores look good, and why overfitting and underfitting are both forms of generalization failure. Model performance is not just about fitting known examples; it is about holding up in similar new situations.
  • Related concepts: overfitting, underfitting
  • Core Section: P1-3.2
  • Appears in: P5-8.1, P5-8.2

constraint

  • Meaning: A constraint is a limit a result must follow, such as length, range, required elements, or forbidden content. It defines the boundary of allowed output in addition to the task request.
  • Why it matters: Constraints turn prompts into clearer work boundaries and review criteria. They help control output length, scope, required evidence, and prohibited behavior, making the result easier to judge and reuse.
  • Related concepts: prompt, output format, instruction
  • Core Section: P1-12.1

context

  • Meaning: Context is the background information, document material, and situation provided so a task or response can be interpreted correctly. It changes how the same instruction should be read.
  • Why it matters: The same request can produce different useful answers depending on the audience, source material, and scope included as context. Context selection is therefore a design choice in prompting, RAG, and long-document work.
  • Related concepts: prompt, instruction, example
  • Core Section: P1-12.1

context window

  • Meaning: A context window is the maximum token range a model can keep available during one input-output computation. It is the size of the working memory for the current response, not unlimited memory.
  • Why it matters: Long documents, long conversations, and retrieved passages must be selected or summarized to fit the context window. A larger window helps capacity but does not remove the need to choose what matters most.
  • Related concepts: token, tokenization, retrieval-augmented generation, RAG
  • Core Section: P6-4.2
  • Appears in: P6-2.1

contextual representation

  • Meaning: A contextual representation is an internal vector representation that changes with surrounding context even for the same word or token. The same surface form can receive different representations in different sentences.
  • Why it matters: This concept marks the move beyond fixed word meanings. It helps readers understand why models such as BERT and LLMs can represent a word according to its current role in a sentence.
  • Related concepts: embedding, BERT, encoder
  • Core Section: P1-11.3
  • Appears in: P6-20.1

control

  • Meaning: Control is a problem type where the system chooses the next action or adjustment based on the current state and goal, then applies it to the environment. It moves from estimating what is true to deciding what to do.
  • Why it matters: Control separates information outputs from actions that change the world. It explains why safety and responsibility become more sensitive when AI output directly affects movement, speed, timing, or system state.
  • Related concepts: planning, action, impact
  • Core Section: P1-1.2

convergence

  • Meaning: Convergence is the process where values or a sequence get closer to a particular value or stable state over repeated steps. It does not require reaching the exact value immediately; the key is whether changes shrink over time.
  • Why it matters: Convergence helps readers interpret iterative computation and training curves. It separates movement toward a stable point from divergence, oscillation, or unstable learning behavior.
  • Related concepts: limit, rate of change, optimization
  • Core Section: P2-2.3
  • Appears in: P2-6.3

conversational LLM

  • Meaning: A conversational LLM is the user-facing form created when instruction following, dialogue formatting, safety tuning, and interface layers are added on top of an LLM. It is a model-plus-product layer, not only the base model.
  • Why it matters: The concept separates a language model from a chatbot experience. System prompts, safety rules, tools, and UI flow can change how the same base model feels and behaves to users.
  • Related concepts: GPT, pretraining, instruction tuning
  • Core Section: P6-5.2

convolution

  • Meaning: Convolution is an operation that moves a small filter across an input and computes local pattern responses at each position. It repeatedly asks whether a particular pattern appears in each local window.
  • Why it matters: Convolution is the core operation that lets CNNs read local visual patterns without mixing the whole image at once. It helps explain feature maps and why the same detector can be applied across many image locations.
  • Related concepts: CNN, convolutional neural network, pooling, feature map
  • Core Section: P5-11.2

copy

  • Meaning: A copy is a new data object separated from the original so later changes do not directly modify the source. The key question is whether a new object was actually made, not whether a new name was assigned.
  • Why it matters: In NumPy and data preprocessing, slicing, fancy indexing, and boolean masks can relate to the original data differently. Understanding copies prevents accidental source modification while also making memory cost visible.
  • Related concepts: shared underlying object, boolean mask, fancy indexing
  • Core Section: P2-11.4
  • Appears in: P2-8.7, P2-12.1

  • Meaning: Copyright is the legal right that protects creative expression such as text, images, charts, code, and other concrete works. It protects expression rather than the abstract idea alone.
  • Why it matters: Copyright matters when using books, articles, images, code, or diagrams in AI inputs and public drafts. Public availability does not automatically mean the expression can be reused freely.
  • Related concepts: quotation, license, training data
  • Core Section: P1-15.2
  • Appears in: P1-10.3

corpus

  • Meaning: A corpus is a collection of text gathered for training, analyzing, or evaluating language models and embeddings. It represents selected language use, not just a random pile of documents.
  • Why it matters: Language models learn patterns from the distribution of text they see. Corpus domain, time period, style, cleaning, bias, and coverage strongly shape model strengths and limitations.
  • Related concepts: language model, data, word2vec
  • Core Section: P1-11.1
  • Appears in: P1-11.3, P6-2.5, P6-19.1

cosine similarity

  • Meaning: Cosine similarity measures how similarly two vectors point in direction. It focuses on the angle between vectors more than their absolute lengths.
  • Why it matters: In text embedding search, direction can matter more than magnitude. Cosine similarity explains how semantic similarity can be read as closeness of direction, while also reminding readers that magnitude-sensitive tasks may need other measures.
  • Related concepts: similarity, distance, similarity search
  • Core Section: P1-13.2
  • Appears in: P1-13.1, P2-3.2, P2-3.4

cost

  • Meaning: Cost is the resource burden of running a service, including model calls, tool execution, storage, network use, evaluation, time, infrastructure, and human review. It asks whether a path remains sustainable when repeated.
  • Why it matters: AI service quality is not only about whether something works once. Cost determines whether a feature can be operated repeatedly, and it must be judged alongside latency, throughput, quality, and review burden.
  • Related concepts: latency, throughput, operation, quality, automatic evaluation
  • Core Section: P1-14.6
  • Appears in: P1-7.1, P1-14.5, P1-16.3, P6-16.2, P6-17.1

credential

  • Meaning: A credential is a proof value such as a password, API key, token, or cookie used to verify identity, access, or request validity. It can directly grant access to accounts, sessions, or external services.
  • Why it matters: If credentials appear in AI inputs, logs, or test outputs, exposure can lead to account compromise or service misuse. This concept connects masking, least privilege, and input restrictions to concrete operational risk.
  • Related concepts: sensitive information, confidential information, security, log, permission
  • Core Section: P1-15.3
  • Appears in: P1-14.5, P1-16.2