Ontological Conjunctive Query Answering over Large Knowledge Bases

793 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
793
On SlideShare
0
From Embeds
0
Number of Embeds
5
Actions
Shares
0
Downloads
8
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Ontological Conjunctive Query Answering over Large Knowledge Bases

  1. 1. Research problem Encodings & Translations Current work Conclusion Questions Ontological Conjunctive Query Answering over Large Knowledge Bases Bruno Paiva Lima da Silva, Jean-Fran¸ois Baget, Madalina Croitoru c {bplsilva,baget,croitoru}@lirmm.fr Universit´ Montpellier 2 e April 16, 2011Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 1 / 26
  2. 2. Research problem Encodings & Translations Current work Conclusion Questions 1 Research problem 2 Encodings & Translations 3 Current work 4 Conclusion 5 QuestionsOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 2 / 26
  3. 3. Research problem Encodings & Translations Current work Conclusion QuestionsTable of Contents 1 Research problem 2 Encodings & Translations 3 Current work 4 Conclusion 5 QuestionsOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 3 / 26
  4. 4. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Decision problemOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  5. 5. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Knowledge base Decision problemOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  6. 6. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Factual knowledge (Very often a DB) Knowledge base Decision problemOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  7. 7. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Factual knowledge Ontology (Very often a DB) (Universal knowledge) Knowledge base Decision problemOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  8. 8. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Factual knowledge Ontology Query (Very often a DB) (Universal knowledge) (Conjunctive query) Knowledge base Decision problemOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  9. 9. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Factual knowledge Ontology Query (Very often a DB) (Universal knowledge) (Conjunctive query) Knowledge base Decision problem (1) “Is there an answer to the query in the knowledge base”?Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  10. 10. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Factual knowledge Ontology Query (Very often a DB) (Universal knowledge) (Conjunctive query) Knowledge base (2) Logical form Logical fact F Ontology O Query Q (Conjunction of atoms) (∀∃-rules) (Conjunctive query) Decision problem (1) “Is there an answer to the query in the knowledge base”?Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  11. 11. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem (1) Ontological conjunctive query answering Factual knowledge Ontology Query (Very often a DB) (Universal knowledge) (Conjunctive query) Knowledge base (2) Logical form Logical fact F Ontology O Query Q (Conjunction of atoms) (∀∃-rules) (Conjunctive query) Decision problem (1) “Is there an answer to the query in the knowledge base”? (2) {F, O} |= Q ?Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 4 / 26
  12. 12. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem F |= Q ... iff there is a substitution S associating every term of the query to a term in the facts. Problem: Finding substitutions (Also known as ENTAILMENT)Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 5 / 26
  13. 13. Research problem Encodings & Translations Current work Conclusion QuestionsResearch problem F |= Q ... iff there is a substitution S associating every term of the query to a term in the facts. Problem: Finding substitutions (Also known as ENTAILMENT) {F, O} |= Q ... iff after being enriched by O, there is a substitution S associating every term of the query to a term in the facts. Problem: Applying rules, Finding substitutions (Also known as RULE-ENTAILMENT)Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 5 / 26
  14. 14. Research problem Encodings & Translations Current work Conclusion QuestionsRules A rule contains two different parts: hypothesis and conclusion. ExampleOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 6 / 26
  15. 15. Research problem Encodings & Translations Current work Conclusion QuestionsRules A rule contains two different parts: hypothesis and conclusion. Example “If x and y are co-workers, and y and z are co-workers, then x and z are also co-workers” ∀x, y , z co-worker (x, y ) ∧ co-worker (y , z) → co-worker (x, z) Rules semantics are that anytime the hypothesis of a rule is found in the facts, its conclusion is then added to the KB as new information.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 6 / 26
  16. 16. Research problem Encodings & Translations Current work Conclusion QuestionsFinding subsitutions Example Facts: Rules: works-for (Mark, LIRMM) ∧ ∀x, y , z works-for (x, z) ∧ works-for (y , z) → co-worker (x, y ) works-for (Travis, LIRMM) ∧ ∀x, y plays-for (x, y ) ∧ is-a(y , SquashClub) → plays(x, Squash) works-for (Tom, LIRMM) ∧ ∀x, y , z plays(x, z) ∧ plays(y , z) → same-sport(x, y ) plays-for (Mark, Team A) ∧ plays-for (Travis, Team B) ∧ plays-for (Tom, Team C ) ∧ is-a(Team A, SquashClub) ∧ is-a(Team B, RugbyClub) ∧ is-a(Team C , SquashClub) ∧ Q1: ∃x plays-for (x, Team B) Q2: ∃x, y co-worker (x, y ) ∧ same-sport(x, y )Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 7 / 26
  17. 17. Research problem Encodings & Translations Current work Conclusion QuestionsFinding subsitutions Example Facts: Rules: works-for (Mark, LIRMM) ∧ ∀x, y , z works-for (x, z) ∧ works-for (y , z) → co-worker (x, y ) works-for (Travis, LIRMM) ∧ ∀x, y plays-for (x, y ) ∧ is-a(y , SquashClub) → plays(x, Squash) works-for (Tom, LIRMM) ∧ ∀x, y , z plays(x, z) ∧ plays(y , z) → same-sport(x, y ) plays-for (Mark, Team A) ∧ plays-for (Travis, Team B) ∧ plays-for (Tom, Team C ) ∧ is-a(Team A, SquashClub) ∧ is-a(Team B, RugbyClub) ∧ is-a(Team C , SquashClub) ∧ Q1: ∃x plays-for (x, Team B) Answers: {(x,Travis)} Q2: ∃x, y co-worker (x, y ) ∧ same-sport(x, y )Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 7 / 26
  18. 18. Research problem Encodings & Translations Current work Conclusion QuestionsQueries and rule application Q2: ∃x, y co-worker (x, y ) ∧ same-sport(x, y ) R1 : ∀x, y , z works-for (x, z) ∧ works-for (y , z) → co-worker (x, y ) R2 : ∀x, y plays-for (x, y ) ∧ is-a(y , SquashClub) → plays(x, Squash) R3 : ∀x, y , z plays(x, z) ∧ plays(y , z) → same-sport(x, y ) Fact works-for (Mark, LIRMM) works-for (Travis, LIRMM) works-for (Tom, LIRMM) plays-for (Mark, Team A) plays-for (Travis, Team B) plays-for (Tom, Team C ) is-a(Team A, SquashClub) is-a(Team B, RugbyClub) is-a(Team C , SquashClub)Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 8 / 26
  19. 19. Research problem Encodings & Translations Current work Conclusion QuestionsQueries and rule application Q2: ∃x, y co-worker (x, y ) ∧ same-sport(x, y ) R1 : ∀x, y , z works-for (x, z) ∧ works-for (y , z) → co-worker (x, y ) R2 : ∀x, y plays-for (x, y ) ∧ is-a(y , SquashClub) → plays(x, Squash) R3 : ∀x, y , z plays(x, z) ∧ plays(y , z) → same-sport(x, y ) Fact works-for (Mark, LIRMM) co-worker (Mark, Travis) works-for (Travis, LIRMM) co-worker (Mark, Tom) works-for (Tom, LIRMM) co-worker (Travis, Mark) plays-for (Mark, Team A) co-worker (Travis, Tom) plays-for (Travis, Team B) co-worker (Tom, Mark) plays-for (Tom, Team C ) co-worker (Tom, Travis) is-a(Team A, SquashClub) is-a(Team B, RugbyClub) is-a(Team C , SquashClub)Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 8 / 26
  20. 20. Research problem Encodings & Translations Current work Conclusion QuestionsQueries and rule application Q2: ∃x, y co-worker (x, y ) ∧ same-sport(x, y ) R1 : ∀x, y , z works-for (x, z) ∧ works-for (y , z) → co-worker (x, y ) R2 : ∀x, y plays-for (x, y ) ∧ is-a(y , SquashClub) → plays(x, Squash) R3 : ∀x, y , z plays(x, z) ∧ plays(y , z) → same-sport(x, y ) Fact works-for (Mark, LIRMM) co-worker (Mark, Travis) works-for (Travis, LIRMM) co-worker (Mark, Tom) works-for (Tom, LIRMM) co-worker (Travis, Mark) plays-for (Mark, Team A) co-worker (Travis, Tom) plays-for (Travis, Team B) co-worker (Tom, Mark) plays-for (Tom, Team C ) co-worker (Tom, Travis) is-a(Team A, SquashClub) plays(Mark, Squash) is-a(Team B, RugbyClub) plays(Tom, Squash) is-a(Team C , SquashClub)Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 8 / 26
  21. 21. Research problem Encodings & Translations Current work Conclusion QuestionsQueries and rule application Q2: ∃x, y co-worker (x, y ) ∧ same-sport(x, y ) R1 : ∀x, y , z works-for (x, z) ∧ works-for (y , z) → co-worker (x, y ) R2 : ∀x, y plays-for (x, y ) ∧ is-a(y , SquashClub) → plays(x, Squash) R3 : ∀x, y , z plays(x, z) ∧ plays(y , z) → same-sport(x, y ) Fact works-for (Mark, LIRMM) co-worker (Mark, Travis) works-for (Travis, LIRMM) co-worker (Mark, Tom) works-for (Tom, LIRMM) co-worker (Travis, Mark) plays-for (Mark, Team A) co-worker (Travis, Tom) plays-for (Travis, Team B) co-worker (Tom, Mark) plays-for (Tom, Team C ) co-worker (Tom, Travis) is-a(Team A, SquashClub) plays(Mark, Squash) is-a(Team B, RugbyClub) plays(Tom, Squash) is-a(Team C , SquashClub) same-sport(Mark, Tom) same-sport(Tom, Mark)Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 8 / 26
  22. 22. Research problem Encodings & Translations Current work Conclusion QuestionsQueries and rule application Q2: ∃x, y co-worker (x, y ) ∧ same-sport(x, y ) R1 : ∀x, y , z works-for (x, z) ∧ works-for (y , z) → co-worker (x, y ) R2 : ∀x, y plays-for (x, y ) ∧ is-a(y , SquashClub) → plays(x, Squash) R3 : ∀x, y , z plays(x, z) ∧ plays(y , z) → same-sport(x, y ) Fact works-for (Mark, LIRMM) co-worker (Mark, Travis) works-for (Travis, LIRMM) co-worker (Mark, Tom) works-for (Tom, LIRMM) co-worker (Travis, Mark) plays-for (Mark, Team A) co-worker (Travis, Tom) plays-for (Travis, Team B) co-worker (Tom, Mark) plays-for (Tom, Team C ) co-worker (Tom, Travis) is-a(Team A, SquashClub) plays(Mark, Squash) is-a(Team B, RugbyClub) plays(Tom, Squash) is-a(Team C , SquashClub) same-sport(Mark, Tom) same-sport(Tom, Mark) Answers: {(x,Mark),(y,Tom)} & {(x,Tom),(y,Mark)}Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 8 / 26
  23. 23. Research problem Encodings & Translations Current work Conclusion QuestionsGoals & Challenges We focus our work on finding substitutions between terms from a given query (constants or variables) and the terms from our facts. In order to do it, we use a BackTrack algorithm. Different methods for KR and manipulation by dedicated reasoning systems have been successfully studied in the past.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 9 / 26
  24. 24. Research problem Encodings & Translations Current work Conclusion QuestionsGoals & Challenges We focus our work on finding substitutions between terms from a given query (constants or variables) and the terms from our facts. In order to do it, we use a BackTrack algorithm. Different methods for KR and manipulation by dedicated reasoning systems have been successfully studied in the past. Large knowledge bases: New challenge F can be very large (see the Semantic Web) Large → Does not fit in main memory.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 9 / 26
  25. 25. Research problem Encodings & Translations Current work Conclusion QuestionsGoals & Challenges We focus our work on finding substitutions between terms from a given query (constants or variables) and the terms from our facts. In order to do it, we use a BackTrack algorithm. Different methods for KR and manipulation by dedicated reasoning systems have been successfully studied in the past. Large knowledge bases: New challenge F can be very large (see the Semantic Web) Large → Does not fit in main memory. “Can we have efficiently an answer to Q, when F is very large?”Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 9 / 26
  26. 26. Research problem Encodings & Translations Current work Conclusion QuestionsTable of Contents 1 Research problem 2 Encodings & Translations 3 Current work 4 Conclusion 5 QuestionsOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 10 / 26
  27. 27. Research problem Encodings & Translations Current work Conclusion QuestionsEncoding: Fact → Set Encoding the fact from our example:Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 11 / 26
  28. 28. Research problem Encodings & Translations Current work Conclusion QuestionsEncoding: Fact → Set Encoding the fact from our example: { works-for (Mark, LIRMM), works-for (Travis, LIRMM), works-for (Tom, LIRMM), plays-for (Mark, Team A), plays-for (Travis, Team B), plays-for (Tom, Team C ), is-a(Team A, SquashClub), is-a(Team B, RugbyClub), is-a(Team C , SquashClub) } Encoded yes, however totally unstructured. The complexity of every atomic operation depend on the size of the knowledge base in atoms.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 11 / 26
  29. 29. Research problem Encodings & Translations Current work Conclusion QuestionsEncoding: Fact → Tables Structuring our fact by the atoms predicates, we obtain tables:Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 12 / 26
  30. 30. Research problem Encodings & Translations Current work Conclusion QuestionsEncoding: Fact → Tables Structuring our fact by the atoms predicates, we obtain tables: works-for plays-for is-a 1 2 1 2 1 2 Mark LIRMM Mark Team A Team A SquashClub Travis LIRMM Travis Team B Team B RugbyClub Tom LIRMM Tom Team C Team C SquashClub This encoding can be directly stored in a Relational Database. Querying is then available either with BackTrack, either with a SQL interface.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 12 / 26
  31. 31. Research problem Encodings & Translations Current work Conclusion QuestionsEncoding: Fact → Graph Structuring the fact, this time by its terms, we obtain a graph:Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 13 / 26
  32. 32. Research problem Encodings & Translations Current work Conclusion QuestionsEncoding: Fact → Graph Structuring the fact, this time by its terms, we obtain a graph: Mark plays-for Team A is-a SquashClub works-for Tom Team C is-a works-for plays-for LIRMM works-for RugbyClub is-a plays-for Travis Team BOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 13 / 26
  33. 33. Research problem Encodings & Translations Current work Conclusion QuestionsAnalysis Encoding a fact without a structure is totally inappropriate for our problem.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 14 / 26
  34. 34. Research problem Encodings & Translations Current work Conclusion QuestionsAnalysis Encoding a fact without a structure is totally inappropriate for our problem. Relational Databases handle very well knowledge located in secondary memory, however: Atomic operations of the BackTrack use SQL operations which complexity also depend on the size of the tables. Using SQL instead may also not be the best solution: Joins become very costly as the number of predicates increases.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 14 / 26
  35. 35. Research problem Encodings & Translations Current work Conclusion QuestionsAnalysis Encoding a fact without a structure is totally inappropriate for our problem. Relational Databases handle very well knowledge located in secondary memory, however: Atomic operations of the BackTrack use SQL operations which complexity also depend on the size of the tables. Using SQL instead may also not be the best solution: Joins become very costly as the number of predicates increases. Running the BackTrack algorithm with a graph works very well when the graph is stored in main memory. Unfortunately, it does not scale very well.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 14 / 26
  36. 36. Research problem Encodings & Translations Current work Conclusion QuestionsTable of Contents 1 Research problem 2 Encodings & Translations 3 Current work 4 Conclusion 5 QuestionsOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 15 / 26
  37. 37. Research problem Encodings & Translations Current work Conclusion QuestionsCurrent challenges In order to be able to perform reasoning over very large knowledge bases, we started searching for storage systems:Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 16 / 26
  38. 38. Research problem Encodings & Translations Current work Conclusion QuestionsCurrent challenges In order to be able to perform reasoning over very large knowledge bases, we started searching for storage systems: that have the ability to support very large knowledge bases stored in secondary memory.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 16 / 26
  39. 39. Research problem Encodings & Translations Current work Conclusion QuestionsCurrent challenges In order to be able to perform reasoning over very large knowledge bases, we started searching for storage systems: that have the ability to support very large knowledge bases stored in secondary memory. efficient on homomorphism elementar operations, such as: computing & retrieving the neighbourhood of a term and to be able to iterate over this structure. checking whether there is a given relation between two given nodes or not.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 16 / 26
  40. 40. Research problem Encodings & Translations Current work Conclusion QuestionsCurrent challenges In order to be able to perform reasoning over very large knowledge bases, we started searching for storage systems: that have the ability to support very large knowledge bases stored in secondary memory. efficient on homomorphism elementar operations, such as: computing & retrieving the neighbourhood of a term and to be able to iterate over this structure. checking whether there is a given relation between two given nodes or not. in which the complexity (time) of the insertion of a new atom does not depend on the size of the KB.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 16 / 26
  41. 41. Research problem Encodings & Translations Current work Conclusion QuestionsAlaska project Alaska Project: Abstract Logic-based Architecture for Storage systems & Knowledge bases Analysis Implementation of classes and interfaces that ensure that all the storage systems plugged in will answer to the same methods using a common type of data. Written in JAVA: Very easy to plug several pieces of code in, however, with a significant loss in speed and efficiency.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 17 / 26
  42. 42. Research problem Encodings & Translations Current work Conclusion QuestionsAlaska: Architecture Knowledge Base < interface > < interface > < interface > IFact IAtom ITerm Common Fact Atom Graph Impls. RDB Impls. RDF Impls. Predicate Term Figure: Class diagram for the architecture.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 18 / 26
  43. 43. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #1 Comparing storage systems between themselves: F |= Q Abstract Architecture Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 19 / 26
  44. 44. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #1 Comparing storage systems between themselves: F |= Q Abstract Architecture Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 19 / 26
  45. 45. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #1 Comparing storage systems between themselves: F |= Q Abstract Architecture Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 19 / 26
  46. 46. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #1 Comparing storage systems between themselves: F |= Q Abstract Architecture Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 19 / 26
  47. 47. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #1 Comparing storage systems between themselves: F |= Q Test results Abstract Architecture Name KB size Querying time RDB ... Mb ... ms Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 19 / 26
  48. 48. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #1 Comparing storage systems between themselves: F |= Q Test results Abstract Architecture Name KB size Querying time RDB ... Mb ... ms GDB ... Mb ... ms Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 19 / 26
  49. 49. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #2 Comparing differrent querying interfaces for a same storage system: F |= Q Abstract Architecture Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 20 / 26
  50. 50. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #2 Comparing differrent querying interfaces for a same storage system: F |= Q Abstract Architecture Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 20 / 26
  51. 51. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #2 Comparing differrent querying interfaces for a same storage system: F |= Q Test results − Query size Querying time BT ... terms ... ms Abstract Architecture Test results − Query size Querying time BT ... terms ... ms Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 20 / 26
  52. 52. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #2 Comparing differrent querying interfaces for a same storage system: F |= Q Test results − Query size Querying time BT ... terms ... ms SQL ... terms ... ms Abstract Q → SQL Architecture Test results − Query size Querying time BT ... terms ... ms Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 20 / 26
  53. 53. Research problem Encodings & Translations Current work Conclusion QuestionsApplication #2 Comparing differrent querying interfaces for a same storage system: F |= Q Test results − Query size Querying time BT ... terms ... ms SQL ... terms ... ms Abstract Q → SQL Q → ... Architecture Test results − Query size Querying time BT ... terms ... ms Graph ... terms ... ms Relational DB Graph DBOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 20 / 26
  54. 54. Research problem Encodings & Translations Current work Conclusion QuestionsImplementations Implementations currently supported by the Alaska project. Abstract Architecture Relational Databases Graph Implementations Next step: Which kind of data to use?Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 21 / 26
  55. 55. Research problem Encodings & Translations Current work Conclusion QuestionsImplementations Implementations currently supported by the Alaska project. Abstract Architecture Relational Databases Graph Implementations Next step: Which kind of data to use?Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 21 / 26
  56. 56. Research problem Encodings & Translations Current work Conclusion QuestionsTable of Contents 1 Research problem 2 Encodings & Translations 3 Current work 4 Conclusion 5 QuestionsOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 22 / 26
  57. 57. Research problem Encodings & Translations Current work Conclusion QuestionsFuture work As the execution performance also came into play in our research problem, our future work will consist in: finding and plugging more pertinent storage systems into our system. identifying any other problems that might have an influence when querying over large knowledge bases. running tests against several large knowledge bases available throughout the web. identifying the storage methods that answer best our problem, and where improvements can be made.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 23 / 26
  58. 58. Research problem Encodings & Translations Current work Conclusion QuestionsThen after... We will also consider working on: implementing some kind of knowledge generator that would generate unbiased facts, which we could test against real data. optimizing our BackTrack algorithm in order to enhance the performance of our system. perhaps implementing a rule application system in order to tackle the RULE-ENTAILMENT problem.Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 24 / 26
  59. 59. Research problem Encodings & Translations Current work Conclusion QuestionsTable of Contents 1 Research problem 2 Encodings & Translations 3 Current work 4 Conclusion 5 QuestionsOntological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 25 / 26
  60. 60. Research problem Encodings & Translations Current work Conclusion QuestionsQuestions Thank you! Questions & comments...Ontological Conjunctive Query Answering over Large Knowledge BasesPAIVA LIMA DA SILVA Bruno (Universit´ Montpellier 2) e 26 / 26

×