20-09-2014, 10:31 AM
A Unified Data and Service Integration Approach for Dynamic Business
Collaboration
A Unified Data and Service.pdf (Size: 761.33 KB / Downloads: 7)
Abstract—
In many collaborations across multiple
organizations, both data integration and service integration are
equally important. Most existing information system
integration approaches focus only on one aspect, resulting
incomplete intergration results. In this paper, we propose a
businss object model where data and its services are correlated,
and a corresponding unified approach in which the modeling,
composition and interaction of both data and service can be
achieved coherently. This approach can help dynamic businss
collaboration by on demand and automatic updates of both
data integration results and service integration results. The
feasiblity and advantages of the approach is validated via a use
case and a preliminary implementation.
INTRODUCTION
Business collaboration is cooperation between multiple
enterprises or organizations working together to achieve a
business goal [1]. In recent years, as a new emerging
paradigm for distributed computing, Service Oriented
Architecture (SOA) has been widely adopted, which can
provide more flexible and dynamic collaboration pattern by
on demand integration of existing services cross
organizational boundaries.
However, most of current service integration approaches,
such as workflow modeling or service compositions [2][3][4],
mainly focus on the integration of service interfaces. A
complete collaboration process needs to be clearly
predefined before its execution. In these approaches, the
business data are hidden behind the control logic of
processes. Users cannot understand the data on the whole.
They can only observe the data from the inputs and outputs
of a service. It is hard for users to quickly capture data
changes and timely adjust processes, especially when
handling emergencies.
The process of a natural disaster relief, such as
earthquake rescue, is a typical example, which involves the
collaboration of multi-departments, e.g., government,
transportation and hospital, and their information systems.
Along with the updates of the disaster situations, decision
makers should constantly adjust their relief plans. For
example, new material shortage reports have to be responded
promptly by updating the original relief plan.
THE UDSI APPROACH
The main idea of the UDSI approach is to mix up the
border of data model and service model. Data and service
should both be regarded as the first class elements for
business collaboration. So we propose a business object
model where business data, modeled as data object, and its
handling operations, modeled as data services, are two
equally important elements. Data objects are used to model
the business data to be shared across the organizations. They
are represented with the nested relational model, which
provide an intuitive and visualized way for business users to
handling complex data. A data service related to a data
object encapsulates a meaningful operation to handle the data
object. By combining data and its services, business object
model provides a coherent way to share, use and integrate
data objects and their services
IMPLEMENTATION
We developed a prototype to support and validate the
UDSI approach, which is shown in Fig. 5. Fig. 5(a) shows
the tool to support the creation of an atomic data object. With
this tool, users can first create the head of a nested table to
describe the metadata of a data object. Then required data
will be extracted from the data sources and transformed into
a set of nested relations. Our current implementation can
support several common data sources, including the
relational database, XML, json and HTML. This process has
two key issues. The first one is how to transform a given data
source (such as HTML, XML, etc.) to the nested relations.
The second one is how to let users decide what data should
be transformed and how to realize it. The corresponding
techniques can be found in our previous papers [8
CONLUSIONS
In real application scenarios, data and services are often
coherent ingredients for information systems. Many services
are designed to handle business data and can be used to
realize data integration. Further, data changes will influence
the service invocations, and vice versa. This paper explores a
unified data and service integration approach for dynamic
collaboration, where the modeling, composition and
interaction of both data and service can be achieved
coherently.
In this paper, a business object model is proposed to
model business data as data object and its handling services
as two equally important elements. Based on the visual
nested table operators, on demand composition of multiple
business objects can be easily realized. We also propose KPI
on the top of business objects to monitor data changes. KPI
can trigger service invocations through pre-defined ECA
rules so that the events of data changes can be dynamically
responded. Further, data changes caused by the service
invocations can be reflected in existing business object and
KPIs. Our preliminary prototype and application demonstrate
the feasibility and advantages of our approach.
For future work, we plan to validate and refine the
approach through more applications. Several aspects of the
approach can be improved. First, we will study how to
facilitate the composition of heterogeneous business objects
via (semi-)automatic semantic matching. Secondly, we will
investigate how to realize the role-based business object
sharing and presentation since different business object
might be shared based on user roles