III: Small: Data-Centric Business Processes: Specification and Static Analysis

点击翻译按钮获取中文摘要

Businesses and other organizations increasingly rely on businessprocess management, and in particular the management of electronicworkflows underlying business processes. These workflows are oftencentered around a database. They are typically very complex and proneto costly bugs, which leads to a critical need for computer-aideddesign and static analysis tools. Such tools would result in enhancedfunctionality, as well as increased confidence in the robustness andcorrectness of complex business processes. They would potentiallybenefit a wide variety of applications ranging from e-commerce todigital government to healthcare and scientific applications. Classical software verification techniques applicable to static analysis include model checking and theorem proving. However, both have seriouslimitations: Model checking requires finite-state abstraction, whichresults in serious loss of semantics for both the business process andverified properties. Theorem proving is incomplete and requires expertuser feedback. This project proposes an alternative approach tostatic analysis. Instead of applying general-purpose techniques withonly partial guarantees of success, it aims to identify restricted butsufficiently expressive classes of business processes for which soundand complete static analysis can be performed in a fully automatic way.Moreover, the target of verification consists of semantically richdata-aware workflows, in contrast to the traditional process-centricworkflows. Data awareness is a feature of central importance inapplications such as the above, which has recently led to theemergence of a new approach to workflow specification in which data ispre-eminent. This approach, introduced by IBM, focuses on data records,known as "business artifacts" or simply "artifacts", that correspondto key business-relevant objects, their life cycles, and theservices (or tasks) that are invoked on the artifacts.IBM's artifact-centric approach has been demonstrated to yieldsubstantial improvements to the operations of medium- and large-sizedbusinesses. The main objective of the present project is to developnew tools for the high-level specification and verification of suchdata-centric business processes. The project investigates thetrade-offs between the expressiveness of the specification languageand the feasibility of the analysis tasks. It aims to establishtractability boundaries for static analysis and to develop practicalalgorithms and heuristics. Tools are to be developed for carrying outanalysis tasks, that will be integrated with Siena, IBM's prototypefor compiling artifact-centric business process specifications intoworkflow support code. The technical problems raised areintellectually quite challenging, and bring into play techniques fromlogic, automata theory, computational complexity, algorithms, andcomputer-aided verification. The resulting static analysis tools havehigh potential for becoming a transformational technology in the areaof business processes.For further information see the project web pagehttp://db.ucsd.edu/artifacts