FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W143394815> ?p ?o ?g. }

• W143394815 startingPage "967" @default.
• W143394815 abstract "We are currently witnessing a trend toward an architectural separation of a knowledge base (KB) into an ontology and a set of rules. The ontology is a description of the concepts and relationships from the application domain; the rules are problem solving procedures expressed with the terms from the ontology. Moreover, terminological standardization taking place in more and more domains has led to the development of domain ontologies. These two developments raise the prospect of reusing existing ontologies when building a new knowledge based system. For instance, the Disciple approach for building a knowledge based agent relies on importing ontologies from existing repositories of knowl edge, and on teaching the agent how to perform various tasks, in a way that resembles how an expert would teach a human apprentice when solving problems in cooperation (Tecuci, 1998; Tecuci et al. 1999). In Disciple, the ontology serves as the generalization hierarchy for learning, an example being basically generalized to a rule by replacing its objects with more general objects from the ontology. However, the learning works well only if the ontology contains all the concepts needed to represent the application domain. We make the assumption that an ontology built from previously developed KBs will contain useful concepts, but it is incomplete and will not contain the more subtle distinctions needed for competent and efficient problem solving in a particular domain. These missing concepts will manifest themselves as exceptions to the learned problem solving rules. A negative exception of the rule is a negative example that is covered by the rule and any specialization (within the current ontology) of the rule that would uncover the exception would result also in uncovering of positive examples of the rule. Similarly, a positive exception of the rule is a positive example that is not covered by the rule and any generalization of the rule that would cover the exception would also result in covering of negative examples of the rule. We are enhancing Disciple by developing a mixedinitiative multistrategy approach to KB revision that will result in an extended and domain-adapted ontology, as well as a set of rules with fewer (if any) exceptions. We are developing two classes of KB revision methods, a class of local methods and a class of global methods. The local methods focus on one rule with its exceptions at a time, in conjunction with the current ontology. Some of the local methods use analogical transfer of discriminatin g features from some objects to other objects in the positive examples of a rule, by considering the similarities between the positive examples and their dissimilarities with the negative exceptions. Other local methods use explanationbased techniques, and similarities between the current rule and other rules, to discover or elicit from the expert discriminating features in the form of explanations of why a negative exception of a rule is not a correct problem solving episode. These local methods work well when the ontology already contains the definitions of the discriminating features, but the descriptions of some of the objects from the ontology are incomplete with respect to those features. The methods perform a local extension of the ontology that leads to a refinement of a rule and a removal of some of its exceptions. Other local methods do not immediately extend the ontology, but suggest characterizations of new concepts that would remove the exceptions. An example of such a characterization is the following: a concept that covers the maximum number of objects from a set of objects and the minimum number of objects from another set of objects . The global methods analyze the alternative concept characterizations suggested by the local methods and attempt to discover a reduced set of concept characterizations that would remove exceptions from a set of rules. In this process, they use various specialization and generalization operators that combine the concept characterizations into a reduced set. They then interact with the domain expert to identify which of the most useful characterizations correspond to meaningful concepts or features in the application domain. The global methods lead to the extension of the ontology with definitions of new objects and features. A significant part of our research effort is also devoted to the evaluation of the developed approach by measuring the effectiveness of the exception-driven discovery and learning with respect to the number of exceptions removed, the impact of the discovered knowledge on the remaining rules, the knowledge acquisition effort required from the domain expert, and the effect of the discovered knowledge on the agent's performance on a set of tasks. In conclusion, we are developing a suite of methods that continuously extend the ontology and revise the rules in the KB through discovery, learning and an interaction with a domain expert in order to achieve competent and efficient problem solving in a particular domain." @default.
• W143394815 created "2016-06-24" @default.
• W143394815 creator A5059429771 @default.
• W143394815 creator A5067046621 @default.
• W143394815 date "1999-07-18" @default.
• W143394815 modified "2023-09-24" @default.
• W143394815 title "Knowledge base revision through exception-driven discovery and learning" @default.
• W143394815 cites W2158015040 @default.
• W143394815 cites W648774283 @default.
• W143394815 hasPublicationYear "1999" @default.
• W143394815 type Work @default.
• W143394815 sameAs 143394815 @default.
• W143394815 citedByCount "0" @default.
• W143394815 crossrefType "proceedings-article" @default.
• W143394815 hasAuthorship W143394815A5059429771 @default.
• W143394815 hasAuthorship W143394815A5067046621 @default.
• W143394815 hasConcept C111472728 @default.
• W143394815 hasConcept C115925183 @default.
• W143394815 hasConcept C134306372 @default.
• W143394815 hasConcept C137003198 @default.
• W143394815 hasConcept C138885662 @default.
• W143394815 hasConcept C154945302 @default.
• W143394815 hasConcept C162324750 @default.
• W143394815 hasConcept C177264268 @default.
• W143394815 hasConcept C199360897 @default.
• W143394815 hasConcept C207685749 @default.
• W143394815 hasConcept C2129575 @default.
• W143394815 hasConcept C22550185 @default.
• W143394815 hasConcept C25810664 @default.
• W143394815 hasConcept C2777002027 @default.
• W143394815 hasConcept C2779247653 @default.
• W143394815 hasConcept C31170391 @default.
• W143394815 hasConcept C33923547 @default.
• W143394815 hasConcept C34447519 @default.
• W143394815 hasConcept C36503486 @default.
• W143394815 hasConcept C41008148 @default.
• W143394815 hasConcept C4554734 @default.
• W143394815 hasConcept C50971890 @default.
• W143394815 hasConcept C78726541 @default.
• W143394815 hasConceptScore W143394815C111472728 @default.
• W143394815 hasConceptScore W143394815C115925183 @default.
• W143394815 hasConceptScore W143394815C134306372 @default.
• W143394815 hasConceptScore W143394815C137003198 @default.
• W143394815 hasConceptScore W143394815C138885662 @default.
• W143394815 hasConceptScore W143394815C154945302 @default.
• W143394815 hasConceptScore W143394815C162324750 @default.
• W143394815 hasConceptScore W143394815C177264268 @default.
• W143394815 hasConceptScore W143394815C199360897 @default.
• W143394815 hasConceptScore W143394815C207685749 @default.
• W143394815 hasConceptScore W143394815C2129575 @default.
• W143394815 hasConceptScore W143394815C22550185 @default.
• W143394815 hasConceptScore W143394815C25810664 @default.
• W143394815 hasConceptScore W143394815C2777002027 @default.
• W143394815 hasConceptScore W143394815C2779247653 @default.
• W143394815 hasConceptScore W143394815C31170391 @default.
• W143394815 hasConceptScore W143394815C33923547 @default.
• W143394815 hasConceptScore W143394815C34447519 @default.
• W143394815 hasConceptScore W143394815C36503486 @default.
• W143394815 hasConceptScore W143394815C41008148 @default.
• W143394815 hasConceptScore W143394815C4554734 @default.
• W143394815 hasConceptScore W143394815C50971890 @default.
• W143394815 hasConceptScore W143394815C78726541 @default.
• W143394815 hasLocation W1433948151 @default.
• W143394815 hasOpenAccess W143394815 @default.
• W143394815 hasPrimaryLocation W1433948151 @default.
• W143394815 hasRelatedWork W1011679765 @default.
• W143394815 hasRelatedWork W1494572628 @default.
• W143394815 hasRelatedWork W1552493609 @default.
• W143394815 hasRelatedWork W1560464553 @default.
• W143394815 hasRelatedWork W2009416503 @default.
• W143394815 hasRelatedWork W2020375154 @default.
• W143394815 hasRelatedWork W2034068590 @default.
• W143394815 hasRelatedWork W2034544412 @default.
• W143394815 hasRelatedWork W2051905507 @default.
• W143394815 hasRelatedWork W2138270831 @default.
• W143394815 hasRelatedWork W2149372408 @default.
• W143394815 hasRelatedWork W2570971010 @default.
• W143394815 hasRelatedWork W2761902018 @default.
• W143394815 hasRelatedWork W2788764253 @default.
• W143394815 hasRelatedWork W3036659126 @default.
• W143394815 hasRelatedWork W345048052 @default.
• W143394815 hasRelatedWork W44766521 @default.
• W143394815 hasRelatedWork W579827085 @default.
• W143394815 hasRelatedWork W58519581 @default.
• W143394815 hasRelatedWork W2187382322 @default.
• W143394815 isParatext "false" @default.
• W143394815 isRetracted "false" @default.
• W143394815 magId "143394815" @default.
• W143394815 workType "article" @default.