First order bottom-up defeasible logic reasoner

Mahdi Rohaninezhad (P64762)

Please use this identifier to cite or link to this item: https://ptsldigital.ukm.my/jspui/handle/123456789/513281

Full metadata record

DC Field	Value	Language
dc.contributor.advisor	Shereena Arif, Dr.
dc.contributor.author	Mahdi Rohaninezhad (P64762)
dc.date.accessioned	2023-10-16T04:35:13Z	-
dc.date.available	2023-10-16T04:35:13Z	-
dc.date.issued	2016-12-28
dc.identifier.other	ukmvital:97589
dc.identifier.uri	https://ptsldigital.ukm.my/jspui/handle/123456789/513281	-
dc.description	Defeasible logic (DL) has attracted a significant attention in non-monotonic reasoning and found many applications in knowledge-based systems and semantic web technologies. This is due to DL is able to reason with incomplete and contradictory information with linear computational complexity. Although top-down (query-based) DL reasoners have reached to a mature stage where are able to support first order language and being fully utilized in knowledge-based systems, bottom-up implementations of the logic commence the first steps. This research concerns applicability of bottom-up DL in knowledge-based systems and aims twofold. The first is to propose an efficient and scalable grounder system for bottom-up DL which makes it applicable in knowledge-based systems. And the second goal is to apply DL in semantic web technology to show its expressive power and reasoning capabilities in dealing with incomplete and contradictory information. To achieve the first objective, SPINdle DL reasoner is chosen since it is the most efficient and plenary bottom-up implementation of the logic. To extend SPINdle towards first order language, firstly a first order defeasible language (prolog-like language) compiler was designed and developed for the reasoner accompanied with RuleML convertor to address semantic web. The inclusion of first order language in bottom-up implementation of the logic arises the need of addressing a NP-complete decision problem called instantiation. A backtracking instantiation algorithm was proposed in the heart of the grounder system. Then, indexing algorithms and greedy ordering were suggested to increase efficiency of backtracking instantiation algorithm. Finally, parallel instantiation and inference algorithms were proposed based on divide-conquer and producer-consumer strategies to increase the system scalability. To achieve desired speedup in parallelization, load balancing and granularity control were taken into account. The system performance (time and memory complexity) was tested with the variety of large scale datasets and rulesets. The experimental results showed that the system performs efficiently and is able to deal with tens of millions of facts (compliant with web data model) in several minutes. A significant gain in the system throughput was experienced by executing benchmarks in parallel mode. To address the second goal, a semantic web framework on policies called open digital right language (ODRL) was enriched by modal DL. Some ontology vocabularies were proposed to embrace modal DL in order to represent and reason with incomplete, contradictory and normative policies of online social network use cases.,Certification of Master's/Doctoral Thesis" is not available
dc.language.iso	eng
dc.publisher	UKM, Bangi
dc.relation	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat
dc.rights	UKM
dc.subject	Defeasible logic
dc.subject	Reasoner
dc.subject	Logic programming
dc.title	First order bottom-up defeasible logic reasoner
dc.type	Theses
dc.format.pages	152
dc.identifier.callno	QA76.63 R637 2016 3 tesis
dc.identifier.barcode	002957(2017)
Appears in Collections:	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Files in This Item:

File	Description	Size	Format
ukmvital_97589+SOURCE1+SOURCE1.0.PDF Restricted Access		575.2 kB	Adobe PDF	View/Open

Show simple item record Recommend this item