Notes

Outline

"DQL" DQL A Query-Answering Language for DAML+OIL

Roles For Reasoning In The Semantic Web Ontology developers Classification Inconsistency detection Evolution Debugging and refinement Maintenance Integration Web site developers Recognition Inconsistency detection Evolution Debugging and refinement For users seeking services Service composition Planning Execution monitoring For users seeking information Query-answering

A DAML+OIL Reasoning Working Group Promote interaction and collaboration among DAML contractors working on reasoning E-mail distribution list Web site Design consensus DAML query language? Design a consensus language for querying DAML+OIL knowledge bases Design consensus DAML justification language? Do this in conjunction with and after design of DAML rule language.

Objectives for DQL Standard for querying DAML+OIL KBs Language for: Queries Answers to queries Justifications for answers Protocol for: Agent-agent interaction during query-answering Support development of: DAML+OIL query-answering reasoning methods Interoperating DAML+OIL query-answering agents Provide basis for Semantic Web standard For full DAML and OWL languages

DQL Design Issues The formal properties of queries and answers What is a query? What is an answer and what does it mean? Inferring answers may be expensive Impractical to always try to compute all answers Answers may only be known to exist Equivalence of multiple answers There may be an infinite number of answers Is a query to a specific knowledge base? Justifications for answers What are they and when should they be computed? Are queries and answers represented in DAML+OIL?

What is a DQL Query? A DML query necessarily includes: Pattern – collection of DAML+OIL sentences in which literals and/or resources have been replaced by variables E.g. (owns ?p ?c)  (type ?c Car)  (color ?c Red) “Who owns a red car?” Distinguished Variables in the pattern whose bindings are sought E.g., {?p}

What is a DQL Query? A DML query necessarily includes: Pattern Distinguished Variables A DML query optionally includes: DAML+OIL knowledge base with respect to which the query is posed All queries are answered with respect to a knowledge base The answering agent may select the knowledge base …

What is a DQL Query? A DML query necessarily includes: Pattern Distinguished Variables A DML query optionally includes: DAML+OIL knowledge base Premise – additional DAML+OIL sentences to facilitate if-then queries E.g., “If C1 is a Seafood Course and W1 is a drink of C1, what color is W1?” Premise: (type C1 Seafood-Course)  (drink C1 W1) Pattern:  (color W1 ?x) Justification request

What is a DQL Answer? A DQL answer necessarily contains: Bindings for some or all of the distinguished variables E.g., “Who is a parent of who?”       Pattern:  (parent-Of ?p ?c)       Distinguished variables:  {?p ?c}       Bindings:  {(?p Earl)  (?c Richard)}              or if the child is not known:  {(?p Earl)} The sentences produced by applying the bindings to the pattern and treating the remaining variables as existentially quantified are entailed by the KB E.g., (exists  ?c  (parent-Of Earl ?c)) The bindings include all the bindings for distinguished variables that the answering agent has A conformance requirement

What is a DQL Answer? A DQL answer necessarily contains: Bindings for some or all of the distinguished variables The sentences produced by applying the bindings to the pattern and treating the remaining variables as existentially quantified are entailed by the KB The bindings include all the bindings for distinguished variables that the answering agent has Each binding is to a literal or URI in the vocabulary of the KB Not allowed: Arbitrary term expressions (descriptions)      E.g., { … (?c “A Person all of whose uncles are lawyers”)} Skolem constants

What is a DQL Answer? A DQL answer necessarily contains: Bindings for some or all of the query pattern’s distinguished variables Query to which this is an answer Answering Agent that produced the answer Knowledge base used to answer the query A DQL answer optionally contains: Justification if requested

DQL Queries DQL is for answering queries of two forms When there are distinguished variables: “What resources or literals from the KB’s vocabulary denote objects that make the query pattern true?” E.g., “Who are the parents of Richard?” When where are no distinguished variables “Is the query pattern true in the query KB?” E.g., “Does Earl have any children?”

How Many Answers? Answers are generated one at a time Response to a query is one of the following: Answer and Continuation – The continuation can be used to request a next answer from the answering agent “no” – no further answers are entailed by the KB “unknown” – no further answers can be produced by the answering agent (but more may be entailed) Every answer contains a distinct set of bindings But two sets of bindings may denote the same objects More capabilities can be built using continuations E.g., 4Find all answers          4Each answer denotes a distinct set of objects

Example Query-Answering Dialogue “child” is an inverse of “parent” Every “Person” has 2 “parent”s Every “parent” of every “Person” is a “Person”

DQL Design Issues The formal properties of queries and answers What is a query? What is an answer and what does it mean? Inferring answers may be expensive Answers may only be known to exist Equivalence of multiple answers There may be an infinite number of answers Is a query to a specific knowledge base? Justifications for answers E.g., proofs, premises, sources, certifications, etc. Not yet designed Are queries and answers represented in DAML+OIL? Yes, using a DAML-Q ontology(?)

Feedback Welcome Birds of a feather session on Thursday 11:00 to noon During this meeting, talk with Pat & myself Ongoing, send e-mail to the joint committee