استدلال در دنیای واقعی: به سوی استنتاج فضایی-زمانی مقیاس‌پذیر و نامشخص (ماشین‌های تفکر آتلانتیس، جلد 2)

The general problem addressed in this book is a large and important one: how to usefully deal with huge storehouses of complex information about real-world situations. Every one of the major modes of interacting with such storehouses – querying, data mining, data analysis – is addressed by current technologies only in very limited and unsatisfactory ways. The impact of a solution to this problem would be huge and pervasive, as the domains of human pursuit to which such storehouses are acutely relevant is numerous and rapidly growing. Finally, we give a more detailed treatment of one potential solution with this class, based on our prior work with the Probabilistic Logic Networks (PLN) formalism. We show how PLN can be used to carry out realworld reasoning, by means of a number of practical examples of reasoning regarding human activities inreal-world situations. Cover......Page 1 Atlantis Thinking Machines......Page 3 Real-World Reasoning: Toward Scalable, Uncertain Spatiotemporal, Contextual and Causal Inference......Page 4 Print: 9789491216107......Page 5 Contents......Page 6 Acknowledgements......Page 10 1 Introduction......Page 12 1.1 The Advantages of a Logical Approach......Page 13 1.2 Main High-Level Conclusions......Page 14 1.3.1 Part I: Representations and Rules for Real-World Reasoning......Page 15 1.3.2 Part II: Acquiring, Storing and Mining Logical Knowledge......Page 20 1.3.3 Part III: Real World Reasoning Using Probabilistic Logic Networks......Page 23 PART I: Representations and Rules for Real-World Reasoning......Page 26 2 Knowledge Representation Using Formal Logic......Page 28 2.1 Basic Concepts of Term and Predicate Logic......Page 29 2.2 Review of Propositional Logic......Page 30 2.2.1 Deduction in Propositional Logic......Page 32 2.3 Review of Predicate Logic......Page 33 2.3.1 Deduction in First-Order Logic......Page 35 2.3.2 First-order Theories......Page 37 2.3.3 Forward and Backward Chaining......Page 38 2.3.4 Decidability and Decision Procedures......Page 40 2.4.1 Example of Kings and Queens......Page 41 2.4.2 Example of Minesweeper......Page 42 2.4.3 Example of Socrates......Page 44 2.5 Modal logic......Page 45 2.6 Deontic logic......Page 47 2.6.1 Fuzzy deontic logic......Page 48 2.7 The frame problem......Page 49 2.7.1 Review of the Frame Problem......Page 50 2.7.2 Working around the Frame Problem......Page 51 3.1 Fuzzy logic......Page 54 3.2 Possibility theory......Page 55 3.3.2 Bayesian Networks......Page 56 3.3.4 Bayesian Inference......Page 60 3.3.5 Markov Logic Networks......Page 62 3.4 Imprecise and indefinite probability......Page 65 4 Representing Temporal Knowledge......Page 66 4.1 Approaches to Quantifying Time......Page 67 4.2 Allen’s Interval Algebra......Page 69 4.2.1 Allen Algebra in the Twitter Domain......Page 72 4.3 Uncertain Interval Algebra......Page 76 5 Temporal Reasoning......Page 80 5.1 The Challenge Time Presents to Classical Logic......Page 81 5.1.1 First order logic: temporal arguments approach......Page 82 5.1.2 Reified temporal logic......Page 83 5.1.3 Modal temporal logic......Page 85 5.1.4 Integration of deontic and temporal logic......Page 88 5.2 Inference systems for temporal logic......Page 90 5.2.1 Inference in the Simple First Order Logic Approach......Page 91 5.2.2 Reified temporal logic......Page 92 5.2.3 Modal temporal logic......Page 93 5.2.4 Computational Tree Logic......Page 94 5.3 Examples of Temporal Inference in the Twitter Domain......Page 96 6 Representing and Reasoning On SpatialKnowledge......Page 100 6.2 Topological Representation......Page 101 6.3 Directional Reasoning......Page 109 6.4 Occupancy Grids: Putting It All Together......Page 113 6.5 Handling Change......Page 119 6.6 Spatial Logic......Page 120 6.6.1 Extending RCC into a Topological Logic......Page 122 6.6.2 Combining Spatial and Temporal Logic......Page 123 7 Representing and Reasoning on ContextualKnowledge......Page 124 7.1.1 The Divide-and-conquer approach......Page 126 7.1.2 Compose-and-conquer Approaches......Page 127 7.1.3 Compatibility constraints......Page 128 7.2 Other Approaches to Contextual Knowledge Representation......Page 129 7.3 Contextual Knowledge in Probabilistic Logic Networks......Page 130 7.4.1 User Modeling in Information Retrieval Systems......Page 132 7.4.2 User modeling from the cognitive perspective......Page 134 7.4.4 Contextual Logic for User Modeling......Page 135 7.5 General Considerations Regarding Contextual Inference......Page 139 7.6 A Detailed Example Requiring Contextual Inference......Page 141 8 Causal Reasoning......Page 148 8.1 Correlation does not imply causation......Page 149 8.2 Other Challenges in Causal Reasoning......Page 150 8.3 Mill’s Methods......Page 151 8.4 Hill’s Criteria......Page 154 8.6 Potential-outcomes (counterfactual) models......Page 155 8.7 Structural-equation models......Page 157 8.8 Probabilistic causation......Page 158 PART II: Acquiring, Storing and Mining Logical Knowledge......Page 160 9 Extracting Logical Knowledge from Raw Data......Page 162 9.2 Extracting Logical Knowledge from Graphs, Drawings, Maps and Tables......Page 164 9.3 Extracting Logical Knowledge from Natural Language Text......Page 166 10.1 Comparison of Available Storage Technologies......Page 170 10.2 Transforming Logical Relationship-Sets into Graphs......Page 174 11 Mining Patterns from Large Spatiotemporal Logical Knowledge Stores......Page 180 11.1 Mining Frequent Subgraphs of Very Large Graphs......Page 181 11.2 Learning Partial Causal Networks......Page 183 11.3 Scalable Techniques for Causal Network Discovery......Page 185 PART III: Probabilistic Logic Networks for Real-World Reasoning......Page 188 12.1 Motivations Underlying PLN......Page 190 12.2 Term and Predicate Logic in PLN......Page 192 12.3 Knowledge Representation in PLN......Page 194 12.4 PLN Truth Values and Formulas......Page 196 12.5 Some Relevant PLN Relationship Types and Inference Rules......Page 198 12.5.1 SatisfyingSet and Member......Page 199 12.5.2 Equivalence and Implication......Page 200 12.5.3 Quantifiers, Average and ThereExists......Page 201 12.5.5 Intensional Inheritance......Page 202 12.6 Applying PLN......Page 203 12.7 Deploying PLN in the OpenCog System......Page 204 13.1 Temporal relationship types......Page 206 13.2 PLN Temporal Inference in Action......Page 208 13.3 PLN Causal Relationship Types......Page 212 13.4 PLN Contextual Inference in Action......Page 213 14 Inferring the Causes of Observed Changes......Page 222 14.1.1 Axioms......Page 224 14.1.2 Inference Trails......Page 228 15 Adaptive Inference Control......Page 248 15.1 Specific Examples Requiring Adaptive Inference Control......Page 249 15.1.1 Using Commonsense Knowledge about Space in Inference Control......Page 252 15.1.2 Using Commonsense Knowledge about Time in Inference Control......Page 255 15.2 General Issues Raised by the Above Examples......Page 257 15.3.1 Activation Spreading and Inference Control in OpenCog......Page 258 15.3.2 Working around the Frame Problem via Integrative AGI......Page 260 16 Conclusion......Page 266 References......Page 268