Definition
An approach to artificial intelligence that represents knowledge and solves problems using explicit symbols, rules and reasoning operations.
The symbolic approach represents problems, goals and knowledge as structures that can be manipulated, then produces conclusions through reasoning or search. Newell and Simon's 1976 paper systematically discussed symbols and search. It synthesized methods and hypotheses; it does not mean that all symbolic programs first appeared that year.
This approach is well suited to expressing explicit rules, but acquiring knowledge and covering common sense in open environments are difficult. It can be combined with statistical and neural methods.
Principles and impact
Problems solved
- Turned explicitly expressible knowledge and reasoning steps into computational processes.
Principles
- Performs rule matching, logical deduction or state-space search over explicit representations.
Improvements
Not recorded
Significance
- Laid foundations for expert systems and various methods of automated reasoning.
History and applications
People
Allen Newell、Herbert A. Simon
Organizations
Not recorded
Events
Not recorded
Applications
Automated reasoning、Rule-based decisions
Limitations
- Achieving rule coverage and maintaining knowledge are costly; formal rules do not automatically guarantee that the premises are correct.
Upstream and downstream relations
Upstream
Sources: A Formal Basis for the Heuristic Determination of Minimum Cost PathsComputer Science as Empirical Inquiry: Symbols and Search
Sources: Computer Science as Empirical Inquiry: Symbols and SearchWhat is Artificial Intelligence? — Basic Questions
Downstream
Sources: DENDRAL — Original Papers and Software ArchiveComputer Science as Empirical Inquiry: Symbols and Search