Constructive Reasoning for Semantic Wikis

摘要: One of the main design goals social software, such as wikis, is to support and facilitate interaction collaboration. This dissertation explores challenges that arise from extending software with advanced facilities reasoning semantic annotations and presents tools in form a conceptual model, structured tags, rule language, set novel forward chaining reason maintenance methods for processing rules help to overcome the challenges. Wikis wikis were usually developed an ad-hoc manner, without much thought about underlying concepts. A conceptual model suitable wiki takes advanced features such into account proposed. Moreover, so called tags are proposed semi-formal knowledge representation step between informal formal annotations. The focus rule languages Semantic Web has been predominantly on expert users on interplay languages and ontologies. KWRL, KiWi Rule Language, a rule language easily understandable for users it aware model it is inconsistency-tolerant, can be efficiently evaluated it builds upon Datalog concepts. The requirement fast response times interactive software translates our work bottom-up evaluation (materialization) of rules (views) ahead time – when or data change, not when they queried. Materialized views have updated when data change. While incremental view maintenance was intensively studied past literature subject abundant, the existing methods surprisingly many disadvantages they do not provide all information desirable explanation derived information, require possibly substantially larger Datalog programs with negation, recompute whole extension of predicate even if only small part affected by a change, adaptation handling general changes. A particular contribution this dissertation consists of forward simple declarative description efficient derive maintain information necessary explanation. The reasoning methods most described in terms extended immediate consequence operators the properties which proven classical logical programming framework. In contrast methods, evaluating original Datalog program do introduce negation present input program predicate’s extension is recomputed. Moreover, directly handle changes both data rules; change does need handled special case. A framework support graphs, structure inspired justification graphs maintenance, Support graphs enable unified description comparison the various define notion of derivation number derivations atom is always finite recursive case. A practical approach implementing reasoning, maintenance, and platform also investigated. It is shown how implementation may benefit using graph database instead along relational database.