%0 Thesis %A Ratnaike, Arvind Viranga %D 2017 %T Lateral reference transformation for emergence %U https://bridges.monash.edu/articles/thesis/Lateral_reference_transformation_for_emergence/4546093 %R 10.4225/03/58781c1a92bd6 %2 https://ndownloader.figshare.com/files/7363231 %K Interaction of data %K Reference transformation framework %K Computational emergence %K Transformation by emergence %K Emergent semantics %K thesis(doctorate) %K ethesis-20100118-102024 %K 1959.1/165168 %K Reference Interpretation %K Open access %K monash:29168 %K 2009 %X This thesis investigates transformation of forms of indication; a transition from data in one form to a description in another form. The data input indicates a situation; the description output indicates a situation. Though both should, ideally, indicate the same situation, there is likely to be a drift in what situation is indicated. A theory for transformation by emergence is introduced, as is a framework for the transformation. The framework can transform an available known form of reference to a more accessible, destination form. This destination form is also the form of a domain knowledge source. This kind of reference transformation is a specialization of a more general prob- lem: that of improving indication; using possibly multiple source forms resulting in possibly multiple destination forms. The improvement should outweigh the cumulative error incurred during the act of improvement. In general, automatic interpretation of multimedia or sensory cues remains a challenge. Synthesis of interpretation is partly completed in the mind of the recipient, where association with prior experience, of both notions and indication thereof, takes place. The scope of the thesis is constrained to what a machine can manipulate. Lateral reference transformation should occur during early emergence. The transformation results in a collection of partial descriptions, which can be used to compose a greater situational description. The reference transformation framework is based on an underlying philosophy of emergence. Situational descriptions emerge from the interaction of domain knowl- edge and situational data. An assumption is made that data and knowledge elements can interact. The reference transformation is lateral, in the sense that the principal quality being increased by emergence is clarity rather than complexity. Domain knowledge is converted to a modifiable internal entity known as a bridg- ing entity. The framework is designed to be independent of media types. The framework adopts the destination form as the basis of the internal form of the bridging entity. It is the bridging entity that is gradually manipulated rather than the media of the original situational data indicators or references. Interaction of the bridging entity with situational data, creates a perturbed bridging entity. The bridging entity can be progressively perturbed through interaction with further data, resulting in an entity that is acceptable as a description of the situation. The final bridging entity is mapped back to an accessible knowledge form, based on an auxiliary entity that stores system experience. Confidence in the acceptability of transformation is based on how much of the situation is considered familiar. Context is considered from two perspectives. From the perspective of the data, it is the knowledge and the experience available for interpretation. This includes entities and concepts that aren’t directly observable in the data. From the perspec- tive of the description, context is whatever influences the emergence and subsequent interpretation, without being explicitly indicated by the description. The framework is investigated by considering its various aspects, in the contexts of disparate situational domains. The conclusions, based on this investigation, lead to modifications to the initial theory and framework. %I Monash University