Technical Vocabulary Quick Reference（专业词汇速查）

Organised by module. Each entry: English term → 中文 → spelling/usage notes.

Module A — Symbolic Logic

Term	中文	Watch Out
Propositional Logic	命题逻辑	NOT “proportional”
First-Order Logic (FOL)	一阶逻辑	Also: predicate logic
Connective	逻辑联结词	∧ AND, ∨ OR, → implies, ¬ NOT
Modus Ponens	肯定前件	P, P→Q ⊢ Q
Modus Tollens	否定后件	¬Q, P→Q ⊢ ¬P
Resolution	归结	NOT “revolution”
CNF (Conjunctive Normal Form)	合取范式	AND of ORs
Quantifier	量词	∀ universal, ∃ existential
Vacuous truth	空真	P→Q is TRUE when P is FALSE
Predicate	谓词	Function returning T/F: Fly(x)
Inference	推理	NOT “reference”

Module B — LNN

Term	中文	Watch Out
Logic Neural Network (LNN)	逻辑神经网络	“Logic Neural” not “Logical Neural”
T-norm	三角范数	Generalises AND to [0,1]
Łukasiewicz	卢卡西维茨	Hard to spell; max(0, a+b-1)
Differentiable	可微的	Enables gradient-based learning
Truth bounds	真值上下界	[L, U] interval
Conjunction / Disjunction	合取 / 析取	AND / OR
Bidirectional inference	双向推理	Upward + downward pass

Module C — Knowledge Representation

Term	中文	Watch Out
Expert System	专家系统	Rule-based, mimics human experts
Knowledge Base	知识库	NOT just “database”
Inference Engine	推理引擎	Applies rules to derive conclusions
Ontology	本体论	Formal concepts + relationships
OWL	网络本体语言	Web Ontology Language
RDF	资源描述框架	(subject, predicate, object) triples
Knowledge Graph	知识图谱	Entity-relation-entity graph
RAG	检索增强生成	Retrieval-Augmented Generation

Module D — Knowledge Graphs

Term	中文	Watch Out
Entity/Relation Embedding	实体/关系嵌入	Dense, learned vectors
TransE	TransE模型	h + r ≈ t
Link Prediction	链接预测	(h, r, ?) or (?, r, t)
Negative Sampling	负采样	Corrupt h or t for training
L1 norm / Manhattan distance	L1范数	Σ\|x_i - y_i\|

Module E — MYCIN

Term	中文	Watch Out
Backward Chaining	反向链接	Goal-driven
Forward Chaining	正向链接	Data-driven
Confidence Factor (CF)	确信因子	Range [-1, +1]; NOT a probability
Production Rule	产生式规则	IF-THEN format
E-MYCIN	基本MYCIN	Domain-independent shell
Knowledge Acquisition Bottleneck	知识获取瓶颈	Hard to extract expert knowledge

Module F — Decision Trees & Ensembles

Term	中文	Watch Out
Decision Tree	决策树	NOT “decision three”
Entropy	熵	H(X) = -Σ p(x) log₂ p(x)
Information Gain	信息增益	IG = H(Y) - H(Y\|X)
Gini Index	基尼指数	1 - Σ p²(i)
Bagging	袋装法	Bootstrap Aggregating — parallel
Boosting	提升法	Sequential error correction
Random Forest	随机森林	Bagging + feature bagging
AdaBoost	自适应提升	Adaptive Boosting
XGBoost	极端梯度提升	eXtreme Gradient Boosting
Weak Learner	弱学习器	Slightly better than random
Decision Stump	决策桩	One-split tree
Bootstrap	自助采样	Sample WITH replacement

Module G — Soft Computing

Term	中文	Watch Out
Vagueness	模糊性	Blurry boundaries → fuzzy logic
Uncertainty	不确定性	Unknown state → Bayesian
Fuzzy Logic	模糊逻辑	Degrees of truth [0, 1]
Membership Function	隶属函数	μ_A(x) — NOT a probability
Bayesian Reasoning	贝叶斯推理	P(H\|e) via Bayes’ theorem
Prior / Posterior	先验 / 后验	Before / after seeing evidence
Likelihood	似然	P(evidence \| hypothesis)
Naive Bayes	朴素贝叶斯	Assumes feature independence

Top Confused Pairs（最易混淆）

Pair	Key Difference
vagueness vs uncertainty	Blurry concept vs unknown fact
bagging vs boosting	Parallel/variance vs sequential/bias
forward vs backward chaining	Data-driven vs goal-driven
inference vs reference	Reasoning vs citing
parameter vs hyperparameter	Learned vs manually set
embedding vs encoding	Dense learned vs any representation
ontology vs knowledge graph	Schema/vocabulary vs data instances
entropy vs information gain	Impurity measure vs impurity reduction
precision vs accuracy	TP/(TP+FP) vs correct/total