Skip to main content
eScholarship
Open Access Publications from the University of California

UC Santa Barbara

UC Santa Barbara Electronic Theses and Dissertations bannerUC Santa Barbara

Management of Data and Collaboration for Business Processes

Abstract

A business process (BP) is a collection of activities and services assembled together to accomplish a business goal. Business process management (BPM) refers to the man- agement and support for a collection of inter-related business processes, which has been playing an essential role in all enterprises. Business practitioners today face enormous difficulties in managing data for BPs due to the fact that the data for BP execution is scattered across databases for enterprise, auxiliary data stores managed by the BPM sys- tems, and even file systems (e.g., definition of BP models). Moreover, current data and business process modeling approaches leave associations of persistent data in databases and data in BPs to the implementation level with little abstraction. Implementing busi- ness logic involves data access from and to database often demands high development efforts.

In the current study, we conceptualize the data used in BPs by capturing all needed information for a BP throughout its execution into a “universal artifact”. The concep- tualization provides a foundation for the separation of BP execution and BP data. With the new framework, the data analysis can be carried out without knowing the logic of BPs and the modification of the BP logics can be directly applied without understanding the data structure.

Even though universal artifacts provide convenient data access for processes, the data is yet stored in the underlying database and the relationship between data in artifacts and the one in database is still undefined. In general, a way to link the data of these two data sources is needed. we propose a data mapping language aiming to bridge BP data and enterprise database, so that the BP designers only need to focus on business data instead of how to manipulate data by accessing the database. We formulate syntactic conditions upon specified mapping in order that updates upon database or BP data can be properly propagated.

In database area, mapping database to a view has been widely studied In recently years, data exchange method extends the notion of database views to a target database (i.e., multiple views) by using a set of conjunctive queries called “tuple generating de- pendency” (tgd). Tgd is a language that is easy to understand/specify, expressive, and decidable for a wide range of properties, which is ideal as a mapping language. Naturally, if both enterprise database and artifacts are represented as relational database, we can take advantage of data exchange technology to bridge enterprise database and artifacts by using tgd as well. Therefore, we re-visit the mapping and update propagation problem under the relational setting.

In addition to the data management for a single BP, it is equivalently essential to un- derstand how messages and data should be exchanged among multiple collaborative BPs. With the introduction of artifacts, data is explicitly modeled that can be used in a collab- orative setting. Unfortunately, today’s BP collaboration languages (either orchestration or choreography) do not emphasize on how data is evolved during execution. More- over, the existing languages always assume each participant type has a single participant instance. Therefore, a declarative language is introduced to specify the collaboration among BPs with data and multiple instances concerned. The language adopts a subset of linear temporal logics (LTL) as constraints to restrict the behavior of the collaborative BPs.

As a follow-up study, we focus on the satisfiability problem of the declarative BP collaboration language, i.e., whether a given specification as a set of constraints allows at least one finite execution. Naturally, if a specification excludes every possible execution, it should be considered as an undesirable design. Therefore, we consider different combi- nation of the constraint types and for each combination, syntactic conditions are provided to decide whether the given constraints are satisfiable. The syntactic conditions automat- ically lead to polynomial testing methods (comparing to PSPACE-complete complexity of general LTL satisfiability testing).

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View