rule induction and reasoning in knowledge graphs
play

Rule Induction and Reasoning in Knowledge Graphs Daria Stepanova - PowerPoint PPT Presentation

Sep 04, 2023 •247 likes •1.53k views

Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Rule Induction and Reasoning in Knowledge Graphs Daria Stepanova Bosch Center for Artificial Intelligence, Renningen, Germany ODSC 2019,

  1. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Answer Set Programs Evaluation of ASP programs is model-based 1. Grounding: substitute all variables with constants in all possible ways 2. Solving: compute a minimal model (answer set) I satisfying all rules Answer set program (ASP) is a set of nonmonotonic rules (1) isMarriedTo ( mary , john ) (2) livesIn ( mary , ulm ) (3) livesIn ( john , ulm ) ← isMarriedTo ( mary , john ) , livesIn ( mary , ulm ) , not researcher ( john ) (4) researcher ( john ) researcher ( john ) I = { isMarriedTo ( mary , john ) , livesIn ( mary , ulm ) , ✭✭✭✭✭✭✭ ✭ livesIn ( john , ulm ) } Particularly suited for reasoning under incompleteness! 12 / 57

  2. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources 13 / 57

  3. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasoning with Incomplete Information Default Reasoning Assume normal state of affairs, unless there is evidence to the contrary By default married people live together. 14 / 57

  4. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasoning with Incomplete Information Default Reasoning Abduction Assume normal state of Choose between affairs, unless there is several explanations evidence to the contrary that explain an observation By default married people live together. John and Mary live together. They must be married. 14 / 57

  5. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasoning with Incomplete Information Default Reasoning Induction Abduction Generalize a number of Assume normal state of Choose between affairs, unless there is similar observations several explanations evidence to the contrary into a hypothesis that explain an observation By default married Given many examples people live together. of spouses living John and Mary live together generalize this together. They must be knowledge. married. 14 / 57

  6. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasoning with Incomplete Information Default Reasoning Induction Abduction Generalize a number of Assume normal state of Choose between affairs, unless there is similar observations several explanations evidence to the contrary into a hypothesis that explain an observation By default married Given many examples people live together. of spouses living John and Mary live together generalize this together. They must be knowledge. married. 14 / 57

  7. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources History of Inductive Learning • AI & Machine Learning 1960s-70s: Banerji, Plotkin, Vere, Michalski, ... • AI & Machine Learning 1980s: Shapiro, Sammut, Muggleton, ... • Inductive Logic Programming (ILP) 1990s: Muggleton, Quinlan, De Raedt, ... • Statistical Relational Learning 2000s: Getoor, Koller, Domingos, Sato, ... 15 / 57

  8. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Learning from Examples Inductive Learning from Examples [Muggleton, 1991] Given: • E + = { fatherOf ( john , mary ) , fatherOf ( david , steve ) } • E − = { fatherOf ( kathy , ellen ) , fatherOf ( john , steve ) } • T = { parentOf ( john , mary ) , male ( john ) , parentOf ( david , steeve ) , male ( david ) , parentOf ( kathy , ellen ) , female ( kathy ) } • Language bias: Horn rules with 2 body atoms 16 / 57

  9. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Learning from Examples Inductive Learning from Examples [Muggleton, 1991] Given: • E + = { fatherOf ( john , mary ) , fatherOf ( david , steve ) } • E − = { fatherOf ( kathy , ellen ) , fatherOf ( john , steve ) } • T = { parentOf ( john , mary ) , male ( john ) , parentOf ( david , steeve ) , male ( david ) , parentOf ( kathy , ellen ) , female ( kathy ) } • Language bias: Horn rules with 2 body atoms Possible hypothesis: • Hyp : fatherOf ( X , Y ) ← parentOf ( X , Y ) , male ( X ) 16 / 57

  10. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Learning from Interpretations Inductive Learning from Interpretations [Raedt and Dzeroski, 1994] Given: • I = { isMarriedTo ( mirka , roger ) , livesIn ( mirka , b ) , livesIn ( roger , b ) , bornIn ( mirka , b ) } • T = { isMarriedTo ( mirka , roger ); bornIn ( mirka , b ); livesIn ( X , Y ) ← bornIn ( X , Y ) } • Language bias: Horn rules with 2 body atoms 17 / 57

  11. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Learning from Interpretations Inductive Learning from Interpretations [Raedt and Dzeroski, 1994] Given: • I = { isMarriedTo ( mirka , roger ) , livesIn ( mirka , b ) , livesIn ( roger , b ) , bornIn ( mirka , b ) } • T = { isMarriedTo ( mirka , roger ); bornIn ( mirka , b ); livesIn ( X , Y ) ← bornIn ( X , Y ) } • Language bias: Horn rules with 2 body atoms Possible Hypothesis: • Hyp : livesIn ( Y , Z ) ← isMarriedTo ( X , Y ) , bornIn ( X , Z ) 17 / 57

  12. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b 18 / 57

  13. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } 18 / 57

  14. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D 18 / 57

  15. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , 18 / 57

  16. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } 18 / 57

  17. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 18 / 57

  18. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( mat ) ← researcher ( mat ) � �� � � �� � Hyp1 Hyp 2 18 / 57

  19. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( mat ) ← researcher ( mat ) � �� � � �� � Hyp1 Hyp 2 Hyp1 � Hyp2 18 / 57

  20. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( X ) ← researcher ( X ) , alive ( X ) � �� � � �� � Hyp1 Hyp2 18 / 57

  21. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( X ) ← researcher ( X ) , alive ( X ) � �� � � �� � Hyp1 Hyp2 Hyp1 � Hyp2 18 / 57

  22. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( X ) ← researcher ( X ) , alive ( X ) � �� � � �� � Hyp1 Hyp2 Hyp1 � Hyp2 • Relative entailment: Hyp1 � Hyp2 wrt T iff Hyp1 ∪ T | = Hyp2 18 / 57

  23. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( X ) ← researcher ( X ) , alive ( X ) � �� � � �� � Hyp1 Hyp2 Hyp1 � Hyp2 • Relative entailment: Hyp1 � Hyp2 wrt T iff Hyp1 ∪ T | = Hyp2 livesIn ( roger , bottmingen ) ? livesIn ( roger , switzerland ) 18 / 57

  24. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( X ) ← researcher ( X ) , alive ( X ) � �� � � �� � Hyp1 Hyp2 Hyp1 � Hyp2 • Relative entailment: Hyp1 � Hyp2 wrt T iff Hyp1 ∪ T | = Hyp2 livesIn ( roger , bottmingen ) ? livesIn ( roger , switzerland ) T : livesIn ( X , switzerland ) ← livesIn ( X , bottmingen ) 18 / 57

  25. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Generality ( � ): essential component of symbolic learning systems • Genaralization as θ -subsumption • Atoms: a � b iff a substitution θ exists such that a θ = b person ( X ) � person ( roger ) , θ = { X / roger } • Clause: C � D iff θ exists, s.t. C θ ⊆ D { worksAt ( X , Y ) } � { worksAt ( Z , bosch ) , researcher ( Z ) } , θ = { X / Z , Y / bosch } • Generalization as entailment • Logic program: Hyp1 � Hyp2 iff Hyp1 | = Hyp2 person ( X ) ← researcher ( X ) person ( X ) ← researcher ( X ) , alive ( X ) � �� � � �� � Hyp1 Hyp2 Hyp1 � Hyp2 • Relative entailment: Hyp1 � Hyp2 wrt T iff Hyp1 ∪ T | = Hyp2 livesIn ( roger , bottmingen ) � livesIn ( roger , switzerland ) T : livesIn ( X , switzerland ) ← livesIn ( X , bottmingen ) 18 / 57

  26. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Clause refinement [Shapiro, 1991]: e.g., MIS, FOIL, etc. • Explore clause search space from general to specific or vice versa to find a hypothesis that covers all examples. livesIn ( X , Y ) ← add atom unify variable to unify constant variables livesIn(X, Y) ← livesIn(U, V) livesIn(bob, Y) ← livesIn(X, X) ← 19 / 57

  27. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Common Techniques in ILP • Clause refinement [Shapiro, 1991]: e.g., MIS, FOIL, etc. • Explore clause search space from general to specific or vice versa to find a hypothesis that covers all examples. livesIn ( X , Y ) ← add atom unify variable to unify constant variables livesIn(X, Y) ← livesIn(U, V) livesIn(bob, Y) ← livesIn(X, X) ← • Inverse entailment [Muggleton, 1995]: e.g., Progol, etc. • Properties of deduction to make hypothesis search space finite 19 / 57

  28. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Zoo of Other ILP Tasks ILP tasks can be classified along several dimensions: • type of the data source, e.g., positive/negative examples, interpretations, answer sets [Law et al. , 2015] • type of the output knowledge, e.g., rules, DL ontologies [Lehmann, 2009] • the way the data is given as input, e.g., all at once, incrementally [Katzouris et al. , 2015] • availability of an oracle, e.g., human in the loop • quality of the data source, e.g., noisy [Evans and Grefenstette, 2018] • data (in)completeness, e.g., OWA vs CWA... • background knowledge, e.g., DL ontology [d’Amato et al. , 2016], hybrid theories [Lisi, 2010] 20 / 57

  29. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Classical ILP for KGs ILP Goal ”The goal of ILP is to develop a correct (and complete) algorithm which efficiently computes hypotheses.” [Sakama, 2005] Knowledge Graphs But the world knowledge is complex, and this might not always be possible in the context of KGs due to several issues... 21 / 57

  30. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Specialities of KGs Open World Assumption : negative facts cannot be easily derived Maybe Roger Federer is a researcher and Albert Einstein was a ballet dancer? 21 / 57

  31. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Specialities of KGs Open World Assumption : negative facts cannot be easily derived Maybe Roger Federer is a researcher and Albert Einstein was a ballet dancer? 21 / 57

  32. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Challenges of Rule Induction from KGs Data bias : KGs are extracted from text, which typically mentions only popular entities and interesting facts about them. “Man bites dog phenomenon” 1 1 https://en.wikipedia.org/wiki/Man_bites_dog_(journalism) 22 / 57

  33. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Challenges of Rule Induction from KGs Huge size : Modern KGs contain billions of facts E.g., Google KG stores 70 billion facts 23 / 57

  34. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Challenges of Rule Induction from KGs World knowledge is complex , none of its “models” is perfect 24 / 57

  35. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Exploratory Data Analysis Question: How can we still learn rules from KGs, which do not perfectly fit the data, but still reflect interesting correlations that can predict sufficiently many correct facts? Answer: Relational association rule mining! Roots in classical datamining. 25 / 57

  36. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Association Rules • Classical data mining task: Given a transaction database, find out products (called itemsets) that are frequently bought together and form recommendation rules. Out of 4 people who bought apples, 3 also bought beer. 26 / 57

  37. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Some Rule Measures Support, confidence, lift 27 / 57

  38. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Some Rule Measures Support, confidence, lift 27 / 57

  39. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Some Rule Measures Support, confidence, lift 27 / 57

  40. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Frequent Itemset Mining • A=apple, B=beer... Frequent patterns are in green. • Monotonicity: any superset of an infrequent pattern is infrequent At the heart of Apriori algorithm 28 / 57

  41. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Relational Association Rule Learning • WARMER [Goethals and den Bussche, 2002] • Upgrade frequent itemsets to frequent conjunctive queries CQ: return all people with their spouses and living places q 1 ( X , Y , Z ) : − isMarriedTo ( X , Y ) ∧ livesIn ( X , Z ) Output: 6 tuples, i.e., supp ( q 1 ) = 6 CQ: return all people with their spouses and living places q 2 ( X , Y , Z ) : − isMarriedTo ( X , Y ) ∧ livesIn ( X , Z ) ∧ livesIn ( Y , Z ) Output: 3 tuples, i.e., supp ( q 2 ) = 3 29 / 57

  42. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Relational Association Rule Learning • WARMER [Goethals and den Bussche, 2002] • Upgrade frequent itemsets to frequent conjunctive queries • traverse the lattice • get frequent CQs based on user-specified value • split into body and head • rank based on a rule measure, e.g., confidence 29 / 57

  43. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Rule Learning from KGs WARMER: confidence CWA: Whatever is not known is false. 30 / 57

  44. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Rule Learning from KGs WARMER: confidence CWA: Whatever is not known is false. | | + | | = 2 | | conf ( r ) = 4 r : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) 30 / 57

  45. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Rule Learning from KGs WARMER: confidence CWA: Whatever is not known is false. | | + | | = 2 | | conf ( r ) = 4 r : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) 30 / 57

  46. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Rule Learning from KGs AMIE [Galarraga et al. , 2015]: PCA confidence PCA: If at least 1 living place of Alice is known, then all are known. isMarriedTo Brad hasBrother isMarriedTo Ann John Kate livesIn livesIn livesIn livesIn Researcher Chicago Berlin livesIn IsA IsA livesIn isMarriedTo isMarriedTo Bob Alice Dave Clara livesIn | | + | | = 2 | | conf PCA ( r ) = 3 Amsterdam r : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) 30 / 57

  47. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources AMIE Refinement Operators livesIn ( X , Y ) ← add dangling atom add instantiated atom livesIn(X, Y) ← marriedTo(X, Z) livesIn(X, Y) ← isA(X, researcher) add closing atom livesIn(X, Y) ← marriedTo(X, Z), livesIn(Z, Y) https://www.mpi-inf.mpg.de/departments/databases-and-information-systems/research/yago-naga/amie/ 31 / 57

  48. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources 32 / 57

  49. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Nonmonotonic Rule Learning Nonmonotonic rule mining from KGs: OWA is a challenge! r : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , not researcher ( X ) 33 / 57

  50. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Theory Revision Quality-based Horn Theory Revision Given: • Available KG 34 / 57

  51. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Theory Revision Quality-based Horn Theory Revision Given: • Available KG • Horn rule set 34 / 57

  52. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Theory Revision Quality-based Horn Theory Revision Given: • Available KG • Horn rule set Find: • Nonmonotonic revision of Horn rule set 34 / 57

  53. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Theory Revision Quality-based Horn Theory Revision Given: • Available KG • Horn rule set Find: • Nonmonotonic revision of Horn rule set with better predictive quality 34 / 57

  54. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Avoid Data Overfitting How to distinguish exceptions from noise? r1 : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , not researcher ( X ) 35 / 57

  55. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Avoid Data Overfitting How to distinguish exceptions from noise? r1 : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , not researcher ( X ) not livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , researcher ( X ) 35 / 57

  56. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Avoid Data Overfitting How to distinguish exceptions from noise? r1 : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , not researcher ( X ) not livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , researcher ( X ) r2 : livesIn ( X , Z ) ← bornIn ( X , Z ) , not moved ( X ) not livesIn ( X , Z ) ← bornIn ( X , Z ) , moved ( X ) 35 / 57

  57. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Avoid Data Overfitting How to distinguish exceptions from noise? r1 : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , not researcher ( X ) not livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) , researcher ( X ) r2 : livesIn ( X , Z ) ← bornIn ( X , Z ) , not moved ( X ) not livesIn ( X , Z ) ← bornIn ( X , Z ) , moved ( X ) { livesIn ( c , d ) , not livesIn ( c , d ) } are conflicting predictions Intuition: Rules with good exceptions should make few conflicting predictions 35 / 57

  58. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Horn Theory Revision Quality-based Horn Theory Revision Given: • Available KG • Horn rule set Find: • Nonmonotonic revision of Horn rules, such that • number of conflicting predictions is minimal • average conviction is maximal M. Gad-Elrab, D. Stepanova, J. Urbani, G. Weikum. Exception-enriched Rule Learning from Knowledge Graphs. ISWC2016 D. Tran, D. Stepanova, M. Gad-Elrab, F. Lisi, G. Weikum. Towards Nonmonotonic Relational Learning from KGs. ILP2016 36 / 57

  59. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Exception Candidates � � not researcher ( X ) r : livesIn ( X , Z ) ← isMarriedTo ( Y , X ) , livesIn ( Y , Z ) not artist ( Y ) 37 / 57

  60. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Experiments • Approximated ideal KG: original KG • Available KG: for every relation randomly remove 20% of facts from approximated ideal KG • Horn rules: h ( X , Y ) ← p ( X , Z ) , q ( Z , Y ) • Exceptions: e 1 ( X ) , e 2 ( Y ) , e 3 ( X , Y ) • Predictions are computed using answer set solver DLV https://github.com/htran010589/nonmonotonic-rule-mining.git 38 / 57

  61. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Experiments • Approximated ideal KG: original KG • Available KG: for every relation randomly remove 20% of facts from approximated ideal KG • Horn rules: h ( X , Y ) ← p ( X , Z ) , q ( Z , Y ) • Exceptions: e 1 ( X ) , e 2 ( Y ) , e 3 ( X , Y ) • Predictions are computed using answer set solver DLV Examples of revised rules: Plots of films in a sequel are written by the same writer, unless a film is American r 1 : writtenBy ( X , Z ) ← hasPredecessor ( X , Y ) , writtenBy ( Y , Z ) , not american film ( X ) Spouses of film directors appear on the cast, unless they are silent film actors r 2 : actedIn ( X , Z ) ← isMarriedTo ( X , Y ) , directed ( Y , Z ) , not silent film actor ( X ) https://github.com/htran010589/nonmonotonic-rule-mining.git 38 / 57

  62. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources 39 / 57

  63. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasonable Rules 40 / 57

  64. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasonable Rules 40 / 57

  65. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasonable Rules People with the same parents are likely siblings r 1 : hasSibling ( X , Z ) ← hasParent ( X , Y ) , hasChild ( Y , Z ) 40 / 57

  66. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasonable Rules People with the same parents are likely siblings r 1 : hasSibling ( X , Z ) ← hasParent ( X , Y ) , hasChild ( Y , Z ) 40 / 57

  67. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasonable Rules People with the same parents are likely siblings | | + | | = 2 | | conf ( r 1 ) = 4 r 1 : hasSibling ( X , Z ) ← hasParent ( X , Y ) , hasChild ( Y , Z ) 40 / 57

  68. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Reasonable Rules People with the same parents are likely siblings | | + | | = 2 | | conf ( r 1 ) = 4 | | | hasSibling ( X , ) ∈G}| = 2 conf pca ( r 1 ) = 2 |{ r 1 : hasSibling ( X , Z ) ← hasParent ( X , Y ) , hasChild ( Y , Z ) 40 / 57

  69. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Erroneous Rules due to Data Bias 41 / 57

  70. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Erroneous Rules due to Data Bias 41 / 57

  71. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Erroneous Rules due to Data Bias × If one is studying in a university where you teach, he/she is your child r 2 : hasChild ( X , Z ) ← worksAt ( X , Y ) , educatedAt ( Z , Y ) 41 / 57

  72. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Erroneous Rules due to Data Bias × If one is studying in a university where you teach, he/she is your child r 2 : hasChild ( X , Z ) ← worksAt ( X , Y ) , educatedAt ( Z , Y ) 41 / 57

  73. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Erroneous Rules due to Data Bias × If one is studying in a university where you teach, he/she is your child | | + | | = 2 | | conf ( r 2 ) = 4 | hasChild ( X , ) ∈G}| = 2 | | conf pca ( r 2 ) = 2 |{ r 2 : hasChild ( X , Z ) ← worksAt ( X , Y ) , educatedAt ( Z , Y ) 41 / 57

  74. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Exploiting Meta-data in Rule Learning Goal: make use of cardinality constraints on edges of the KG to improve rule learning. T. Pellissier-Tanon, D. Stepanova, S. Razniewski, P . Mirza, G. Weikum. Completeness-aware rule learning from KGs. ISWC2017. 42 / 57

  75. Motivation Preliminaries Rule Learning Exception-awareness Incompleteness Rules from Hybrid Sources Cardinality Statements • num ( p , s ) : Number of outgoing p -edges from s in the ideal KG • miss ( p , s ) : Number of missing p -edges from s in the available KG • If miss ( p , s ) = 0, then complete ( p , s ) , otherwise incomplete ( p , s ) num ( hasChild , john ) = 3 miss ( hasChild , john ) = 1 incomplete ( hasChild , john ) 43 / 57

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Attributed Description Logics: Reasoning on Knowledge Graphs Markus Krtzsch 1 Maximilian Marx 1

Attributed Description Logics: Reasoning on Knowledge Graphs Markus Krtzsch 1 Maximilian Marx 1

Attributed Description Logics: Reasoning on Knowledge Graphs Markus Krtzsch 1 Maximilian Marx 1 Ana Ozaki 2 Veronika Thost 3 1 Knowledge-Based Systems Group, TU Dresden 2 KRDB Research Centre for Knowledge and Data, Free University of Bolzano 3

383 views • 22 slides
CHAPTER-4 1 LOGIC AND REASONING ! Knowledge and ! Reasoning in Knowledge- Reasoning Based

CHAPTER-4 1 LOGIC AND REASONING ! Knowledge and ! Reasoning in Knowledge- Reasoning Based

CHAPTER-4 1 LOGIC AND REASONING ! Knowledge and ! Reasoning in Knowledge- Reasoning Based Systems ! syntax and semantics ! shallow and deep reasoning ! validity and satisfiability ! forward and backward ! logic languages chaining ! alternative

616 views • 23 slides
Rule Learning from Knowledge Graphs Guided by Embedding Models Vinh Thinh Ho 1 , Daria Stepanova 1

Rule Learning from Knowledge Graphs Guided by Embedding Models Vinh Thinh Ho 1 , Daria Stepanova 1

Motivation Our Approach Rule Construction Rule Prunning Evaluation Conclusion Rule Learning from Knowledge Graphs Guided by Embedding Models Vinh Thinh Ho 1 , Daria Stepanova 1 , Mohamed Gad-Elrab 1 , Evgeny Kharlamov 2 , Gerhard Weikum 1 1

700 views • 31 slides
Automated Reasoning Induction Alan Bundy Automated Reasoning Induction Lecture 14, page 1 A

Automated Reasoning Induction Alan Bundy Automated Reasoning Induction Lecture 14, page 1 A

Automated Reasoning Induction Alan Bundy Automated Reasoning Induction Lecture 14, page 1 A Summation Problem What is 1 + 2 + 3 + .... + 999 + 1000? Is there a formula for the general solution? Can we prove this? Automatically?

429 views • 27 slides
Uncertain Knowledge and Reasoning 9 AI Slides (6e) c Lin Zuoquan@PKU 1998-2020 9 1 9

Uncertain Knowledge and Reasoning 9 AI Slides (6e) c Lin Zuoquan@PKU 1998-2020 9 1 9

Uncertain Knowledge and Reasoning 9 AI Slides (6e) c Lin Zuoquan@PKU 1998-2020 9 1 9 Uncertain Knowledge and Reasoning 9.1 Uncertainty 9.2 Probability Syntax and semantics Inference Independence Bayes rule 9.3 Bayesian

1.18k views • 104 slides
http://cs224w.stanford.edu 1. Introduction to Knowledge Graphs 2. Knowledge Graph completion 3.

http://cs224w.stanford.edu 1. Introduction to Knowledge Graphs 2. Knowledge Graph completion 3.

CS224W: Machine Learning with Graphs Jure Leskovec, Hongyu Ren, Stanford University http://cs224w.stanford.edu 1. Introduction to Knowledge Graphs 2. Knowledge Graph completion 3. Path Queries 4. Conjunctive Queries 5. Query2Box: Reasoning with Box

895 views • 86 slides
The Ramify Rule of Separation Logic Compositional Reasoning for Sharing Jules Villard 1 Joint work

The Ramify Rule of Separation Logic Compositional Reasoning for Sharing Jules Villard 1 Joint work

The Ramify Rule of Separation Logic Compositional Reasoning for Sharing Jules Villard 1 Joint work with Aquinas Hobor 2 1 University College London 2 National University of Singapore Programs with Sharing in the Wild Graphs Acyclic graphs (DAGs)

889 views • 70 slides
CSE 421: Introduction to Algorithms Induction)* Graphs Shayan&Oveis&Gharan 1 Graphs

CSE 421: Introduction to Algorithms Induction)* Graphs Shayan&Oveis&Gharan 1 Graphs

CSE 421: Introduction to Algorithms Induction)* Graphs Shayan&Oveis&Gharan 1 Graphs 2 Undirected)Graphs)G=(V,E) A 2 10 3 Disconnected)graph 11 12 4 8 Multi)edges 13 B 9 Isolated)vertices 6 Self)loop 7 3

230 views • 19 slides
Natural Deduction and Rule Induction Dr. Liam OConnor University of Edinburgh LFCS UNSW, Term

Natural Deduction and Rule Induction Dr. Liam OConnor University of Edinburgh LFCS UNSW, Term

Natural Deduction Rule Induction Ambiguity Simultaneous Induction Natural Deduction and Rule Induction Dr. Liam OConnor University of Edinburgh LFCS UNSW, Term 3 2020 1 Natural Deduction Rule Induction Ambiguity Simultaneous Induction

610 views • 59 slides
Association rule mining Association rule induction: Originally designed for market basket analysis

Association rule mining Association rule induction: Originally designed for market basket analysis

Association rule mining Association rule induction: Originally designed for market basket analysis . Aims at finding patterns in the shopping behavior of customers of supermarkets, mail-order companies, on-line shops etc. More specifically: Find

378 views • 34 slides
Program for Today Rule induction CSEP 546: Data Mining Propositional First-order

Program for Today Rule induction CSEP 546: Data Mining Propositional First-order

Program for Today Rule induction CSEP 546: Data Mining Propositional First-order First project Instructor: Pedro Domingos Rule Induction 1 2 3 4 5 First Project: Clickstream Mining Overview The Gazelle Site

232 views • 10 slides
VU @ D2.1.1 Part 1: Approximation Reasoning method Knowledge Knowledge base Base

VU @ D2.1.1 Part 1: Approximation Reasoning method Knowledge Knowledge base Base

VU @ D2.1.1 Part 1: Approximation Reasoning method Knowledge Knowledge base Base A-Box (UoM) and T-Box Reasoning Input Output Method Input 1 Guidelines for Reasoning Methods Semantically well-founded providing a

388 views • 6 slides
Knowledge Graphs Large ge and complex plex graphs capturing millions of entities and

Knowledge Graphs Large ge and complex plex graphs capturing millions of entities and

Knowledge Graphs Large ge and complex plex graphs capturing millions of entities and relationships between them! Entity Relationship Ubiqu quitous itous toda day: y: Linking Open Data Freebase DBpedia YAGO How to Query Knowledge

512 views • 16 slides
Mechanically Certifying Formula-based Noetherian Induction Reasoning Sorin Stratulat Universit

Mechanically Certifying Formula-based Noetherian Induction Reasoning Sorin Stratulat Universit

Mechanically Certifying Formula-based Noetherian Induction Reasoning Sorin Stratulat Universit de Lorraine, LITA 1 Formula-based Noetherian Induction Noetherian induction principles Noetherian induction: let ( E , < ) be a well-founded

695 views • 55 slides
CSC2542 Topics in Knowledge Representation & Reasoning: Automated Planning & Reasoning

CSC2542 Topics in Knowledge Representation & Reasoning: Automated Planning & Reasoning

CSC2542 Topics in Knowledge Representation & Reasoning: Automated Planning & Reasoning About Action Fall 2010 General Information URL: http://www.cs.toronto.edu/~sheila/2542/f10 Lectures: Thursday 2:00 4:00 PM, BA3116 (for now)

167 views • 15 slides
Principles of Knowledge Representation and Reasoning May 20 & 23, 2008 Nonmonotonic

Principles of Knowledge Representation and Reasoning May 20 & 23, 2008 Nonmonotonic

Principles of Knowledge Representation and Reasoning May 20 & 23, 2008 Nonmonotonic Reasoning Introduction Motivation Different Forms of Reasoning Principles of Knowledge Representation and Reasoning Different Formalizations

228 views • 10 slides
Ontology Engineering Lecture 2: First Order Logic Maria Keet email: mkeet@cs.uct.ac.za home:

Ontology Engineering Lecture 2: First Order Logic Maria Keet email: mkeet@cs.uct.ac.za home:

Introduction Syntax Semantics Reasoning Summary Ontology Engineering Lecture 2: First Order Logic Maria Keet email: mkeet@cs.uct.ac.za home: http://www.meteck.org Department of Computer Science University of Cape Town, South Africa

789 views • 75 slides
1 options for perception (e.g., enlarged text, physical models) options for language,

1 options for perception (e.g., enlarged text, physical models) options for language,

Powerful Ways to Support All Students in the Diverse Classroom Paula Kluth, Ph.D. www.paulakluth.com @paulakluth 2019 1 action & expression engagement representation presenting info & content in different ways 2 NOT a

143 views • 11 slides
Introd u ction to Te x t Encoding FE ATU R E E N G IN E E R IN G FOR MAC H IN E L E AR N IN G

Introd u ction to Te x t Encoding FE ATU R E E N G IN E E R IN G FOR MAC H IN E L E AR N IN G

Introd u ction to Te x t Encoding FE ATU R E E N G IN E E R IN G FOR MAC H IN E L E AR N IN G IN P YTH ON Robert O ' Callaghan Director of Data Science , Ordergroo v e Standardi z ing y o u r te x t E x ample of free te x t : Fello w- Citi z

808 views • 46 slides
WELCOME, MR. FITZGERALD! March 13, 2020 SERVING PEOPLE SERVING ANIMALS ANIMAL CARE

WELCOME, MR. FITZGERALD! March 13, 2020 SERVING PEOPLE SERVING ANIMALS ANIMAL CARE

WELCOME, MR. FITZGERALD! March 13, 2020 SERVING PEOPLE SERVING ANIMALS ANIMAL CARE Cleaning Feeding Daily observations Enrichment Adoption visitations Lost pet reunions Humane euthanasia CLIENT SERVICES

517 views • 9 slides
Weaponised Children: Wars, Ghosts and the adults the children will become. Sarah Calvert.

Weaponised Children: Wars, Ghosts and the adults the children will become. Sarah Calvert.

Weaponised Children: Wars, Ghosts and the adults the children will become. Sarah Calvert. PhD. Clinical Psychologist Distinguished Scholar, Waikato University. The Adult the Child will Become We are a social animal Families are an

309 views • 16 slides
FALL 2018 ABET UPDATE ADVISORY BOARD MEETING NOV. 29 2018 QUICK OUTLINE ABET Accreditation

FALL 2018 ABET UPDATE ADVISORY BOARD MEETING NOV. 29 2018 QUICK OUTLINE ABET Accreditation

FALL 2018 ABET UPDATE ADVISORY BOARD MEETING NOV. 29 2018 QUICK OUTLINE ABET Accreditation Committee ABET Timeline Milestones Completed Advisory Board Review and Feedback ABET ACCREDITATION COMMITTEE Michael Soltys Jason

115 views • 7 slides
UNDERGRADUATE PROGRAMS IN INFORMATICS: ACM/IEEE RECOMMENDATIONS & NEW ORGANIZATIONAL MODELS

UNDERGRADUATE PROGRAMS IN INFORMATICS: ACM/IEEE RECOMMENDATIONS & NEW ORGANIZATIONAL MODELS

UNDERGRADUATE PROGRAMS IN INFORMATICS: ACM/IEEE RECOMMENDATIONS & NEW ORGANIZATIONAL MODELS FOR TURKEY Ali Yaz c TOBB University of Economics & Technology, Computer Engineering Department, 06560, Ankara, Turkey

194 views • 18 slides
Announcements Hand in Homework 8 now Pick up all old homework now CSE 311 Foundations of

Announcements Hand in Homework 8 now Pick up all old homework now CSE 311 Foundations of

Announcements Hand in Homework 8 now Pick up all old homework now CSE 311 Foundations of Review session Computing I Sunday, June 9, 4 pm, EEB 125 List of Final Exam Topics and sampling of some Lecture 28 typical kinds of

230 views • 21 slides

More recommend