Symbolic reasoning
Symbolic reasoning is the manipulation of explicit symbols, facts, and formal relations according to rules to derive conclusions, verify properties, or select actions.
Natural Language Processing in Action, Second Edition
Symbolic reasoning is an artificial-intelligence technique that derives conclusions by manipulating explicit symbols, facts, and logical rules.
- How it is used
- Grounds NLP pipelines by connecting language-model generation with explicit facts and logical inference.
- How deep it goes
- Contrasts symbolic and probabilistic reasoning, names first-order logic and GOFAI, and sketches dependency-tree-based grounding.
- Examples
- Uses knowledge graphs, natural-language dependency trees, compiler parse trees, and abstract syntax trees as concrete reference points.
- What you can do after
- Apply knowledge graphs to separate reasoning from language generation and make model explanations more fact-grounded.
Read how Natural Language Processing in Action, Second Edition explains symbolic reasoning →
Authorization in Action
Symbolic reasoning analyzes general formulas representing sets of possible inputs and behaviors instead of checking only concrete examples.
- How it is used
- Analyzes authorization policies for inconsistencies, redundancies, logical errors, and behavior changes during refactoring.
- How deep it goes
- Translates policies into formulas and checks full request spaces, stopping short of general-purpose programming flexibility.
- Examples
- Tests invoice approval, unmanaged-device deployment access, and permission preservation after policy refactoring.
- What you can do after
- Diagnose policy failures and verify broad authorization properties using Cedar analysis tools and solver-generated counterexamples.
Read how Authorization in Action explains symbolic reasoning →
Inside AI
Symbolic reasoning is reasoning that derives conclusions by manipulating explicit symbols according to formal rules and logic.
- How it is used
- Contrasts rule-based deduction and planning with pattern recognition and data-driven learning.
- How deep it goes
- Defines symbolic reasoning through one constrained puzzle and stops short of implementation details.
- Examples
- Uses the Farmer-Wolf-Goat-Cabbage riddle and medical diagnosis as contrasting applications.
- What you can do after
- Evaluate whether a task requires explicit rules and constraints beyond statistical pattern recognition.
Related concepts
- probabilistic reasoningcontrasts with · Probabilistic reasoning is drawing or selecting conclusions under uncertainty by representing and evaluating possibilities using probabilities.
- generationequivalent to · Generation is the computational production of outputs from supplied inputs or context, often by applying a model to produce predictions or new content.
- static analysisimplements · Static analysis is the examination of source code or other program artifacts without executing the program to infer properties or identify potential issues.
- tokenizationused by · Tokenization is the process of dividing text or other sequential input into discrete units called tokens for computational processing.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.