The popularization and democratization of cartography is inevitable with support technology maturing and lots of web mapping platform launched. A new study filed of Cartography called on-demand mapping is emerging which making mapping ubiquitous and pervasive. However, although personalized maps are increasingly being generated, mostly by non-experts, and published in order to meet the diverse requirement, such maps are commonly obscure, cluttered and ineffective in communicating spatial information.The main reason for this lack of quality and legibility in personalized maps is due to the absence of a quality control procedure which was traditionally done by cartographers. Obviously, it is not reasonable to ask cartographers to do this manually in this highly dynamic, real-time and ubiquitous mapping context. A better approach is to formalize their cartographic knowledge into the mapping tools which in turn could help improve the quality of the user-generated maps.This thesis focuses on map generalization, a key component in map production that requires a high level of cartographic knowledge, and investigates principles and methods in support of implementing map generalization tools that have sophisticated functionalities and are also widely accessible by internet users. The contributions of this thesis are described as follows:1) SaaS service is used to providing map generalization aimed at stand-alone generalization software which has several flaws such as high threshold, low universal, confined to cartographers and overlooking characteristics of on-demand mapping. This research summarizes the differences between stand-alone generalization systems and distributed web generalization service, such as novices and professionals coexisting, demand-driven specific generalization taking over supply-driven common generalization, user’s explicit or implicit needs instead of scale becoming control factor etc. Hence five principles of map generalization service are designed:①interfaces shield complex knowledge of map generalization;②task of generalization must correspond to users’requirements and preferences;③automated generalization process control;④real-time responses;⑤a quick-response based evaluation of generalization results.2) A conceptual framework is designed for on-demand generalization. The framework driven by user requirement employs data and processing services hosted on the cloud to establish a smart map generalization service. The core of this framework is an engine with four kinds of function:encapsulation for adaptive parameter adjustment, manipulation for adaptive algorithm selection, composition for adaptive operator decision and optimization for the revision of generalization knowledge.3) This thesis analyzes the characteristics of generalization task in on-demand mapping context and extracts a multi-factor model including scale control and quality control. Scale control drives degree of generalization positively and quality control helps to choose operator, algorithm and threshold according to users’emphasis on characteristic, relationship and pattern of spatial data. This model can export four kinds of scale-sensitive data:single scale point, scale series, continuous scale range and variable-scale region with interfaces such as parameter, scale, degree, symbol and data size etc.4) This thesis discusses input, output, pre-condition and post-condition of generalization algorithms on the basis of idea of semantic web and four kinds of semantic information suitable for describing generalization services are summarized and classified to four categories, capabilities description, contextual information, instruction and requirements, which are used to extend WPS OGC specification. The semantic information in this study is concise and accurate able to improve the semantic interoperability of map generalization service meanwhile laying the groundwork for service composition. Introduce the concept of recommender system and Q&A Community to improve the user experience of map generalization. Map generalization service can adapt to users’background, mapping task and context owing to recommender systems including knowledge-based and collaborative filtering recommendation.5) A web map generalization prototype is implemented. This system is equipped with highly efficient algorithms suitable for online mapping and implements several operators such as simplification, smooth, aggregation, dissolve, collapse, enhancement and displacement for vector data. The system can visualize and evaluate the results instantly with controlling the degree of generalization precisely.This thesis extends the scope of common geospatial service to support on-demand mapping by combines map generalization on behalf of specialized geoprocessing process with spatial information service. Starting from the differences between stand-alone generalization and online web generalization, we improve the classic conceptual framework of generalization and introduce several innovative ideas such as user profile, semantic description of generalization algorithms, recommender system arid publish map generalization functionality to SAAS service. The user-friendly generalization service shields complex algorithm parameter information and has several advantages over stand-alone software, such as cloud computing, out of the box, on-demand service, continuous update. There are two kinds of interface for different kinds of requirement:Graphic User Interface and Application Programming Interface. The prototype system demonstrates the generalization SAAS service is feasible and has a promising prospect. |