Towards large-scale information integration

Abstract

Software engineers confront many challenges during software development. One challenge is managing the relationships that exist between software artifacts. We refer to this task as information integration, since establishing a relationship between documents typically implies that an engineer must integrate information from each of the documents to perform a development task. In the past, we have applied open hypermedia techniques and technology to address this challenge. We now extend this work with the development of an information integration environment. We present the design of our environment along with details of its first prototype implementation. Furthermore, we describe our efforts to evaluate the utility of our approach. Our first experiment involves the discovery of keyword relationships between text-based software artifacts. Our second experiment examines the code of an open source project and generates a report on how its module relationships have evolved over time. Finally, our third experiment develops the capability to link code claiming to implement W3C standards with the XHTML representation of the standards themselves. These experiments combine to demonstrate the promise of our approach. We conclude by asserting that the process of software development can be significantly enhanced if more tools made their relationships available for integration.

An end-to-end industrial software traceability tool

“…The sheer number of artifacts produced in a project, the differing levels of formality and specificity between various artifact types, and the complex interrelationships between artifacts [4,5,27] form the heart of the traceability problem.…”

“…The heterogeneous tools used in supporting the development process, with no data exchange capability between tools make it impossible to enter information once. This lack of interoperability between tools have been identified as a problem by [5,13,24] Entering information once not only applies to having an artifact stored in different tools, but it also applies to information entered in one phase being carried over to another phase. For instance, in the planning phase, provisional UCs are created without a unique identifier.…”

“…Having redundant data is one of the main problems of traceability [27], causing additional overhead of reconciling data [5]. Another method of enabling entering information once is through the information integration approach [5], which supports traceability between heterogeneous artifacts by translating them to a homogeneous representation.…”

“…Another method of enabling entering information once is through the information integration approach [5], which supports traceability between heterogeneous artifacts by translating them to a homogeneous representation. Once inside the homogeneous environment, traces can be automatically generated using a set of rules.…”

WikiWiki weaving heterogeneous software artifacts

“…An evidence of this, is the diversity of work applying XML specifically to software documentation, which include the following, to mention only a few of the more relevant to this paper: representations of source code (JavaML [2,5], cppML [21], srcML [10]), UML [14], mechanisms to integrate heterogeneous documents from different sources [16,15,4], and literate programming systems [31,9].…”