Alpha: Symbolization and Inference Bram van Heuveln Minds and Machines Lab RPI.

Slides:



Advertisements
Similar presentations
Copyright 2008, Scott Gray1 Propositional Logic 4) If.
Advertisements

The Game of Algebra or The Other Side of Arithmetic The Game of Algebra or The Other Side of Arithmetic © 2007 Herbert I. Gross by Herbert I. Gross & Richard.
Logic & Critical Reasoning
1 Logic Logic in general is a subfield of philosophy and its development is credited to ancient Greeks. Symbolic or mathematical logic is used in AI. In.
Logic Use mathematical deduction to derive new knowledge.
Propositional Logic CMSC 471 Chapter , 7.7 and Chuck Dyer
Deduction In addition to being able to represent facts, or real- world statements, as formulas, we want to be able to manipulate facts, e.g., derive new.
Logic.
Logic Concepts Lecture Module 11.
F22H1 Logic and Proof Week 7 Clausal Form and Resolution.
Truth Trees Intermediate Logic.
Logic 3 Tautological Implications and Tautological Equivalences
Automated Proof Generation for EG Bram van Heuveln Spring 2003 AI & Reasoning Lab, RPI.
Discussion #12 1/22 Discussion #12 Deduction, Proofs and Proof Techniques.
Formalizing Alpha: Soundness and Completeness Bram van Heuveln Dept. of Cognitive Science RPI.
Existential Graphs: Beta Introduction to Logic. Alpha Review: Symbolization ‘P’ ‘not P’ ‘P and Q’ ‘P or Q’ ‘if P then Q’ F EG PP PP P  Q P  Q P 
Syllabus Every Week: 2 Hourly Exams +Final - as noted on Syllabus
Existential Graphs and Davis-Putnam April 3, 2002 Bram van Heuveln Department of Cognitive Science.
PHIL 120: Jan 8 Basic notions of logic
Using the EG Applet Bram van Heuveln. Contents General Comments –Some general comments regarding the use of the applet Work Area –Instructions for creating.
Let remember from the previous lesson what is Knowledge representation
Introduction to Logic for Artificial Intelligence Lecture 1 Erik Sandewall 2010.
Proofs, Recursion and Analysis of Algorithms Mathematical Structures for Computer Science Chapter 2.1 Copyright © 2006 W.H. Freeman & Co.MSCS SlidesProofs,
Proof by Deduction. Deductions and Formal Proofs A deduction is a sequence of logic statements, each of which is known or assumed to be true A formal.
EE1J2 – Discrete Maths Lecture 5 Analysis of arguments (continued) More example proofs Formalisation of arguments in natural language Proof by contradiction.
Improvements and Extensions of the EG Interface Fall 2002.
Algebra Problems… Solutions Algebra Problems… Solutions © 2007 Herbert I. Gross Set 12 By Herbert I. Gross and Richard A. Medeiros next.
Discrete Mathematics and its Applications
Artificial Intelligence Reasoning. Reasoning is the process of deriving logical conclusions from given facts. Durkin defines reasoning as ‘the process.
Propositional Logic Review
Intro to Discrete Structures
A Java Implementation of Peirce’s Existential Graphs Bram van Heuveln Department of Philosophy State University College at Oneonta March 22, 2001.
Understanding PML Paulo Pinheiro da Silva. PML PML is a provenance language (a language used to encode provenance knowledge) that has been proudly derived.
Marriage Problem Your the sovereign in a small kingdom. One of your jobs is to marry off the people in kingdom. There are three rules that apply.
Pattern-directed inference systems
Advanced Topics in Propositional Logic Chapter 17 Language, Proof and Logic.
1 CMSC 471 Fall 2002 Class #10/12–Wednesday, October 2 / Wednesday, October 9.
Existential Graphs Intermediate Logic. Existential Graphs A graphical logic system developed by C.S. Peirce almost 100 years ago. Peirce studied semiotics:
Algebra Problems… Solutions Algebra Problems… Solutions © 2007 Herbert I. Gross Set 10 By Herbert I. Gross and Richard A. Medeiros next.
Automated Theorem Proving: Resolution and Davis-Putnam Intermediate Logic.
INTRODUCTION TO LOGIC Jennifer Wang Fall 2009 Midterm Review Quiz Game.
1 CMSC 250 Discrete Structures CMSC 250 Lecture 1.
A Java implementation of Peirce’s Alpha Graphs Bram van Heuveln (RPI) Dennis Higgins (SUNY Oneonta) SUNY Oneonta undergraduate programmers: Elizabeth Hatfield,
A Java implementation of Peirce’s Existential Graphs* faculty advisors: Bram van Heuveln (Philosophy) Dennis Higgins (Math & Computer Science) student.
A Java implementation of Peirce’s Existential Graphs
Hazırlayan DISCRETE COMPUTATIONAL STRUCTURES Propositional Logic PROF. DR. YUSUF OYSAL.
Extra slides for Chapter 3: Propositional Calculus & Normal Forms Based on Prof. Lila Kari’s slides For CS2209A, 2009 By Dr. Charles Ling;
LOGIC AND QUANTIFIERS. INTRODUCTION Many algorithms and proofs use logical expressions such as: “IF p THEN q” or “If p1 AND p2, THEN q1 OR q2” Therefore.
Formal Proofs and Boolean Logic Chapter 6 Language, Proof and Logic.
Logic UNIT 1.
Week 4 - Friday.  What did we talk about last time?  Floor and ceiling  Proof by contradiction.
The Existential Graphs Project Rensselaer Reasoning Group September 12, 2001.
Foundations of Discrete Mathematics Chapter 1 By Dr. Dalia M. Gil, Ph.D.
1 Propositional Proofs 1. Problem 2 Deduction In deduction, the conclusion is true whenever the premises are true.  Premise: p Conclusion: (p ∨ q) 
Sound Arguments and Derivations. Topics Sound Arguments Derivations Proofs –Inference rules –Deduction.
March 23 rd. Four Additional Rules of Inference  Constructive Dilemma (CD): (p  q) (r  s) p v r q v s.
Chapter 7. Propositional and Predicate Logic
2. The Logic of Compound Statements Summary
Knowledge Representation and Reasoning
Introduction to Logic for Artificial Intelligence Lecture 1
7.1 Rules of Implication I Natural Deduction is a method for deriving the conclusion of valid arguments expressed in the symbolism of propositional logic.
Automated Proof Generation for EG
Truth Tables Hurley
Truth Trees.
TRUTH TABLES continued.
Back to “Serious” Topics…
The Method of Deduction
The Logic of Declarative Statements
Foundations of Discrete Mathematics
Propositional Logic CMSC 471 Chapter , 7.7 and Chuck Dyer
Presentation transcript:

Alpha: Symbolization and Inference Bram van Heuveln Minds and Machines Lab RPI

Alpha Alpha is the part of Existential Graphs (EG) corresponding to propositional (or truth- functional) logic (PL). This presentation covers: –symbolization from PL to EG from EG to PL –inference rules strategies

Symbolization Sheet of Assertion To assert some statement in EG, you put the symbolization  of that statement on a sheet of paper, called the ‘Sheet of Assertion’ (SA). Thus, to assert the truth of some statement p, draw:  where  is the symbolization of p SA

Symbolization Location is irrelevant The location of the symbolization on the SA does not matter: as long as it is somewhere on the SA, it is being asserted. Thus:   states the same as: In fact, the above two graphs are regarded as completely identical.

Symbolization Juxtaposition and Conjunction By drawing the symbolization of two statements on the SA, you are asserting the truth of both statements at once. Hence, the mere juxtaposition of two symbolizations on the SA can be interpreted as the assertion of a single conjunction. Thus:   can be seen as the assertion of both  and , but also as the assertion of   .

Symbolization Generalized Conjunction Since any number of symbolizations can be juxtaposed on the SA, juxtaposition becomes a kind of generalized conjunction that can have any number of conjuncts. Moreover, since the location of each of the symbolizations on the SA does not matter, no particular order on these conjuncts is imposed. This coincides with our abstract understanding of conjunction, and it is here that EG has an important advantage over the linear notation of traditional PL. An example will help:

Symbolization Generalized Conjunction: Example The top-right graph can be interpreted in any of the following ways in PL: –the assertion of 3 statements: P, Q, and R –the assertion of 2 statements: P and Q  R (or of R and P  Q, or of P and R  Q, etc.) –the assertion of a single statement: P  (Q  R) (or of (P  Q)  R, or of (Q  R)  P, or of P  (R  Q), etc.!) However, our abstract understanding is in each case the same: P, Q, and R are all, and at the same time, true. Hence, a single symbolization should suffice, and this is exactly what EG can offer us. P R Q

Symbolization Cut and Negation You assert the negation of some statement by drawing a cut (circle, oval, rectangle, or any other enclosing figure) around the symbolization of that statement. Thus:  asserts that  is false. (from now on, the SA will no longer be drawn)

Symbolization Empty Graph and Tautology Any blank piece of paper can be seen as an ‘empty graph’. Thus, any graph can be seen as the juxtaposition of that graph with an empty graph. However, since this juxtaposition should express the same as the original graph, any empty graph expresses a tautology. Another way of looking at this is to view any tautology as an ‘empty claim’ since, it being a tautology, it effectively doesn’t claim anything at all.

Symbolization Empty Cut and Contradiction A cut without any contents is called an ‘empty cut’. Since an empty cut negates an empty graph, any empty cut expresses a contradiction (  ).

Symbolization Expressive Completeness Using juxtaposition for conjunction, and cuts for negation (and letters for simple, atomic statements), any compound, truth- functional statement can be symbolized in EG. That is, since conjunction and negation form an expressively complete set of operators, EG is expressively complete as well (and EG does not need parentheses!)

Symbolization From PL to EG P ~~                P      Symbolization in EG Expression in PL

Symbolization From EG to PL P  Q or Q  P or P and Q ~(~P  ~Q) or ~P  Q or P  Q or ~(~Q  ~P) or ~Q  P or Q  P ~(P  ~Q) or ~(~Q  P) or ~P  Q or Q  ~P or P  Q Q QP P QP QP ~(P  Q) or ~(Q  P) or ~P  ~Q or ~Q  ~P Possible Readings

Inference Inference Rules Alpha has four inference rules: –2 rules of inference: Insertion Erasure –2 rules of equivalence: Double Cut Iteration/Deiteration To understand these inference rules, one first has to grasp the concepts of subgraph, double cut, level, and nested level.

Inference Subgraph The notion of subgraph is best illustrated with an example: R Q The graph on the left has the following subgraphs: QRR R Q R Q,,,, In other words, a subgraph is any part of the graph, as long as cuts keep all of their contents. Any graph is a subgraph of itself, and empty graphs can be considered subgraphs as well.

Inference Double Cut A Double Cut is any pair of cuts where one is inside the other and where there is only the empty graph in between. Thus: P RQ RQ,, and contain double cuts, butdoes not.

Inference Level The level of any subgraph is the number of cuts around it. Thus, in the following graph: R Q Q R R R Q R Q is at level 2 (the graph itself) is at level 0,,, andare at level 1, and

Inference Nested Level A subgraph  is said to exist at a nested level in relation to some other subgraph  if and only if one can go from  to  by going inside zero or more cuts, and without going outside of any cuts. E.g. in the graph below: P R Q R exists at a nested level in relation to Q, but not in relation to P. Also: Q andR exist at a nested level in relation to each other.

Inference Double Cut The Double Cut rule of equivalence allows one to draw or erase a double cut around any subgraph. Obviously, this rule corresponds exactly with Double Negation from PL.    

Inference Insertion The Insertion rule allows one to insert any graph at any odd level.   12k+1  1

Inference Erasure The Erasure rule of inference allows one to erase any graph from any even level.   12k  1

Inference Iteration/Deiteration The Iteration/Deiteration rule of equivalence allows one to place or erase a copy of any subgraph at any nested level in relation to that subgraph. 

Inference Formal Proofs A formal proof in EG consists in the successive application of inference rules to transform one graph into another. Formal proofs in EG are used just as in PL: –To show that an argument is valid, transform the graph of the premises into the graph of the conclusion. –To show that a set of statements is inconsistent transform the graph of the statements into an empty cut. –To show that two statements are equivalent, transform the one into the other, and vice versa. –To show that a statement is a tautology, transform an empty graph into the graph of that statement.

Inference Transforming rather than Rewriting An interesting difference between doing formal proofs in EG and doing formal proofs in traditional systems is that in the former one transforms (by adding or deleting) a single graph, whereas in the latter one deals with multiple sentences, and has to do a lot of rewriting. Example: Suppose we want to infer Q  (R  S) from P and P  [Q  (R  S)]. In PL, we would use Modus Ponens to go from two separate statements to a third, having to rewrite all of Q  (R  S) on a separate line. In EG, we have a single graph being the juxtaposition of the symbolizations of P and P  [Q  (R  S)], after which the second P gets deleted by deiteration and the desired result is obtained through the simple elimination of a double cut.

Inference Proofs as Movies Because graphs are being transformed rather than rewritten, proofs in EG are going to look quite different from proofs in PL. Proofs become like videos that one can play, rewind, fast-forward, etc. It would be interesting to see if this dynamic character of proofs has any further conceptual consequences as far as people are able to do proofs and think about proofs.

Inference Subproofs Another interesting difference between doing formal proofs in EG and PL is that in EG there is no need for doing subproofs. Of course, one could define subproofs in EG, but one should notice that at that point one is no longer dealing with a single graph that is being transformed: extra formal machinery is needed to deal with subproofs, just as in PL. The interesting fact is that the 4 inference rules of EG are both sound and complete, even though they don’t use subproofs (see “Alpha: Soundness and Completeness”).

Inference Simulating Subproofs Subproofs aren’t needed in EG since subproofs can be simulated using the rules of EG in the following general manner: –1. Draw an empty double cut on level 0. –2. Insert the assumption of the subproof within the outer cut (i.e. on level 1). –3. Iterate the original graph within the inner cut, as well as the extra assumption. –4. Manipulate the graphs on level 2 as usual. –5. Use obtained result appropriately (see next slides) Subproofs within subproofs can be done at levels 2, 4, etc.

Inference Conditional Proofs Simulating Conditional Proof: –The assumption is the antecedent (  ) of the desired conditional –After iterating the original graph (  ) and the assumption on the even level, one tries to derive the consequent (  ). –The result is the desired conditional.      DC IN IT(2x) 

Inference Indirect Proof Simulating Indirect Proof: –The assumption is the negation of the desired goal (  ) –After iterating the original graph (  ) and the assumption on the even level, one tries to derive an empty cut. –Once the empty cut has been obtained, the desired goal can be obtained through double cut elimination.         DC IN IT(2x) DC

Inference Deriving Empty Cuts Deriving an empty cut often merely requires the application of Erasure, Deiteration, and erasing double cuts. In other words, one often merely has to eliminate parts of the graph in order to derive a contradiction.

Inference Efficiency of Proofs In traditional PL systems, there is a trade-off between the number of inference rules and the number of steps of a formal proof: if one wants a formal proof to require fewer steps, one has to introduce more inference rules, and if one wants fewer inference rules, formal proofs will require more steps. While EG has fewer rules (4) than traditional PL systems (10 to 20), proofs in EG usually require fewer steps!

Inference Ease of Proofs Although hard empirical data needs to be gathered, doing proofs in EG seems to be easier than doing proofs in PL: –Graphical representation –Transforming rather than rewriting –Subproofs –Fewer rules, fewer steps. –Ease of deriving empty cut.