Multimodal Construction Of Mashup Applications Based On Spreadsheet And Data Flow

As Web has become a vast repository of online information resources and applications, it is challenging to meet the needs of information integration during Web-scale computing environments. On the other hand, people would like to develop valued-added applications by using online information resources to satisfy their individual requirements (called Mashup Applications). However, existing information integration applications, such as search engines simplely list relevant results from multiple sites without filtration. Data Warehouse and query-based information integration middlewares need tremendous modeling and development previously by professional IT staff. Workflow system is still too cumbersome while improving the flexibility of information integration. The complexity in accessing variable Web information resources and expressing business logic of Mashup Applications, together with lacking of professional programming knowledge lead to the main challenge of user-driven development of Mashup applications.To solve the problems in end-user-steered development of flexible and individual Mashup applications, we bring forward a exploratory and multimodal Mashup building paradigm based on nested Spreadsheet and data flow. The dissertation focuses on usability and dynamicity enhancement of user-steered development of Mashup applications. Main works and contributions are as follows:(1) To handle the complexity in expressing business logic of Mashup application, we propose a spreadsheet-like information integration mode which includes a basic data structure based on nested table algebra called nested spreadsheet, on which a nested spreadsheet recursive algebra is built. At the same time, a finite sequence of carefully chosen recursive Mashup operators are defined to express Mashup business logic. Data operations work in a set basis and loop control structures are hidden from users.(2) To handle the flexibility and dynamicity, we design and implement a demonstrated exploratory programming environment——ClientBrowser which is based on nestd relation model and user-steered exploratory service composition. It combines the advantage of both spreadsheet and data flow, and offers a multimodal composed paradigm. Data operations are implemented as visual actions on nested spreadsheets by which Mashups can be built without user's control of data flow and control flow programming. At the same time, Mashup building processes are recorded synchronously by data flow. Furthermore, we provide a modification mechanism based on business logic and data dependencies in flow path to support exploratory and humanized functions such as quickly locating application changes, undoing operation sequence, breakpoint debugging, service recommendation to enable services can be dynamically integrated in real-time, self-configure to meet user's updating business needs. The achievement can not only reduce the complexity of composing Mashup applications, but also improve the usability and flexibility of it.