The goal of this project is to provide an efficient platform for browser-based, data-driven web application computation. The platform enables cost-effective development of fast-responding applications that adjust well to accesses from mobile clients (e.g. smart phones, tablets). The project achieves its goal using the following approaches: (1) designing novel high level, location-transparent, declarative, data-driven languages that require much lower coding effort than direct HTML5 coding in order to specify the business process and data access of the applications; (2) developing an optimizer for low latency query execution plans that utilize browser-based storage and asynchronous computation; (3) developing an action scheduler that optimizes the location and execution order of the actions described in the declarative language; (4) developing a user/action concurrency control theory and a dependency analysis algorithm so that the user can view and act while prior actions are still computed; (5) prototyping an application-enabling platform that encompasses the developed languages, algorithms and optimizations; and (6) evaluating the effectiveness of the platform in two aspects: how much it reduces latency and how much it reduces the coding effort.
The project's research will have great impacts on mobile-accessible, data-driven web applications, which, by being written in the proposed automatically optimized, declarative languages, will enjoy both low latency and low development cost. The project supports graduate and undergraduate students. Lectures on the research results will be incorporated into PI's undergraduate-level course on web application development. Publications, software, an online service and experimental data from this research will be disseminated via the project web site (www.db.ucsd.edu/browserbasedforward).
