Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/33319
Title: Incremental query answering over semantic contextual information
Authors: MOHAMMAD OLIYA
Keywords: reasoning, context, incremental, query, rule, ontology
Issue Date: 11-Jan-2012
Source: MOHAMMAD OLIYA (2012-01-11). Incremental query answering over semantic contextual information. ScholarBank@NUS Repository.
Abstract: In the vision of ubiquitous computing, users should get the right information, at the right time, at the right situation. Such provision of appropriate information will assist users in performing their daily tasks in a natural and transparent way. \emph{Context-aware} systems are more flexible, adaptable, and autonomous. </br> </br> Formal representation of context information is gaining an ever increasing number of advocates in the literature. It fosters interoperability among heterogeneous context sources and eases the development of context-aware applications. Thanks to the representation and reasoning power of their underlying logics, it is also possible to describe complex context data, share and integrate context between heterogeneous entities, deduce abstract or hidden knowledge, and deal with the inconsistency of the data. The Web Ontology Language (OWL) is the standard way for representing the semantics of information in the web, and is the main formal and practical method for modeling context. </br> </br> One major issue with regards to the application of OWL is the overhead of query answering when changes occur in the observed facts. Traditionally, reasoning on an updated knowledge base is performed from the scratch. As the query answering mechanisms are based on available reasoning techniques, this also results in the re-evaluation of the query from the beginning. </br> </br> In this dissertation, a novel incremental query answering technique for semantic (OWL-based) contextual information is proposed. The aim is to avoid redundant computations and alleviate the cost of reasoning from scratch. Our method can be applied to the fragments of OWL which can be axiomatized as a set of rules, including Resource Description Format Schema (RDFS) and Description Horn Logics. We consider \emph{instance retrieval queries} which ask for instances of the contextual situations predefined in the ontology. In addition, we only consider changes in the facts (ABox) and not the changes in the definition of knowledge structure (TBox). </br> </br> Our technique consists of first translating the ontology schema as well as queries into rules. By targeting the Description Horn Logic fragment of OWL, we are able to represent the ontology schema as a set of definite Horn rules, i.e. rules with only one literal at the head. These rules are then used to build the \emph{Rete} network which incrementally maintains the query results as changes occur in the observed data. As the evaluation of the rules can be computationally expensive, we further introduce an optimization to prune the rules which do not affect query results. The empirical results suggest the practicality of our method for the perceived application domains. </br> </br> To the best of our knowledge, we are the first to address the problem for context aware systems. The novelty of our method lies in the identification of the proper tools and language fragments that can work in tandem for the expected results. Our method does not need alteration of existing OWL-based reasoners, enabling context aware systems to achieve incremental query answering functionality with minimal changes. In addition, our method supports \emph{hybrid} inference method where application-specific rules can be used in conjunction with pure OWL based reasoning.
URI: http://scholarbank.nus.edu.sg/handle/10635/33319
Appears in Collections:Master's Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
OliyaM.pdf7.81 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

192
checked on Dec 11, 2017

Download(s)

159
checked on Dec 11, 2017

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.