Data Brain Modeling And Its Applications Based On Brain Informatics

It is a popular interdisciplinary cooperation mode to support brain science research by information technologies. Various applications of information technologies, such as brain databases, mining tools of brain data, have been a powerful impetus to accelerate the data centric brain science study. Owing to the characteristics of thinking centric study, the key of information technology supporting in Brain Information (BI) is to integrate various experimental studies and computing studies. Its basic is the integration of data, information and knowledge. In brain science, the studies about information technologies mainly focus on how to provide effective supports for specific research activities. There isn't an effective means which can realize the integration of data, information and knowledge. Thus, the BI researchers need to study new information technologies for the integration of key data, information and knowledge in BI data cycle, as well as the development of new interdisciplinary cooperation modes.In this thesis, we start from a real world problem on constructing the BI data cycle system and grasp the key issue—the integration of data, information and knowledge. Because BI study has a data centric process and conceptual data modeling is a key technology for the development of information systems, the conceptual modeling of brain data is regarded as a core issue. Based on the exist work on conceptual data model, ontology development, brain database, and domain-driven data mining, we proposed a Data-Brain based framework to integrate data, information and knowledge. In this framework, the Data-Brain is a new-style and domain-driven conceptual data model, which is used to model BI methodology. By the Data-Brain and various Brain Informatics provenances, key data, information and knowledge of BI study can be integrated to support the systematic BI study. Taking BI as the domain background, this thesis introduces the core theories and technologies of Data-Brain based data-information-knowledge integration framework, involved with Data-Brain modeling, Data-Brain based systematic brain data management and Data-Brain driven systematic brain data analysis.Data-Brain modeling is a core issue of developing the BI data-information-knowledge integration framework. In order to model BI methodology, the Data-Brain is different from the traditional conceptual schemas of brain databases, brain models and brain related ontologies. Thus, firstly, starting from the research motivation of Data-Brain modeling, this thesis gives the definition of Data-Brain and designs a multi-view and multi-dimension Data-Brain framework based on systematic BI methodology. Secondly, we confirm the ontological modeling process and the ontological type of Data-Brain by analyzing the characteristics of Data-Brain. According the basic theories of ontological modeling, a meta-model of Data-Brain is designed to support the Data-Brain modeling. Finally, extending the traditional ontological modeling methodology, we add the characteristics of domain into modeling process to design a BI methodology based Data-Brain modeling process. Using the meta-models of Data-Brain, the whole modeling process is introduced in detail, from dimension construction, the construction of relationships among dimensions, to conceptual view extraction. The formal definition of Data-Brain is also given.As the bridge connecting experimental studies with computing studies in BI, systematic brain data management is the basic function of BI data-information- knowledge integration framework. It needs to realize not only data storage and data sharing oriented management, but also systematic analysis oriented management. Firstly, based on the analysis of existing brain databases, especially fMRI and EEG/ERP databases, this thesis generalizes the systematic brain data management and points out, the core issue of systematic brain data management is to collect and store the data information coming from different aspects of systematic BI studies, and effectively organize and utilize them for multi-aspect and multi-level data information analyses which are coming from human or intelligent agents during systematic brain data analysis and simulation. Secondly, in order to solve this core issue, we propose a new meta-data of brain data, i.e., BI provenances, and discuss its concept, classification, formal definition and construction process. The provenance cube and its basic operations are also defined. These studies provide academic and technical base to realize systematic brain data management. Finally, we introduce an induction centric case study, in which BI provenances and the provenance cube are used to meet the requirements of multi-aspect and multi-level data information analysis, to demonstrate the validity of Data-Brain based approach of systematic brain data management.Systematic brain data analysis is an important issue of BI methodology. A core function of data-information-knowledge integration framework is to support systematic brain data analysis. After analyzing the existing multi-aspect brain data analysis, this thesis proposes a Data-Brain driven multi-aspect brain data analysis to offset the deficiencies of existing expert-driven approach. This Data-Brain driven approach models the process of multi-aspect brain data analysis as an organized society of autonomous knowledge discovery agents to organize data and analytical methods into various multi-aspect analysis workflows utilizing various information technologies, such as search, reasoning, semantic matching, dynamic service planning, etc. In this society, the Data-Brain and Brain Informatics provenances are regarded as the knowledge base to support workflow planning. Utilizing the obtained analysis workflows, a four phases of analytical methodology is proposed to realize a Data-Brain driven multi-aspect brain data analysis. Related theories and technologies are realized in a prototype system. Based on this prototype system, the whole Data-Brain driven process of multi-aspect brain data analysis is introduced by a numerical induction oriented case study.From the theoretical point of view, this thesis has the following contribution: (1) A Data-Brain based data-information-knowledge integration framework is proposed to realize a new brain research mode with the supporting of information technologies; (2) Aiming at domain driven conceptual data modeling, related theories, technologies, concrete modeling process and strategy are studied; (3) A Data-Brain based approach is proposed to realize the systematic human data management; (4) Data-Brain driven approach of systematic brain data analysis is proposed and related theoretical and technical explorations are completed. In addition, this thesis also aims to contribute from the practical point of view: (1) Based on the characteristics of thinking centric BI study, core technical explorations and practices of data-information-knowledge integration framework are completed to transform systematic BI methodology from abstract concepts to a concrete and practical research route. This is the key to popularize BI study all over the world. (2) This thesis represents a case study on the data cycle system of hyper world. Although it aims to BI, the proposed theories, technologies, approaches and strategies can be extended to other domains for accelerating the study on the data cycle system of hyper world.