In recent years, architecture decision records (ADRs) have emerged as a lightweight way to capture information about software architecture, decisions, context, and consequences. ADRs have a simple structure, which makes them easy to use even for software architects with little experience. A promising research direction is the reuse of the recorded architecture knowledge on a cross-project basis. The rapidly growing number of design decisions in different technology stacks, however, makes manual analysis arduous, which is why AI-based methods are required to facilitate efficient identification of relevant knowledge. In this paper, we investigate the suitability of knowledge graphs for analyzing ADRs for providing easy access to domain knowl-edge and enabling the discovery of additional insights. The main challenges are to model the ontology (i.e., the knowledge graph schema) from ADRs and to find a mapping to Wikidata concepts to enrich the knowledge graph with public knowledge. We explain and verify our approach to create a semantic model of existing data sources within a case study, for which we derived over 4,000 pre-filtered open-source ADRs from GitHub, and present initial analysis results. We believe that this approach could impact software engineering practice by assisting software architects in the deeper analysis of current ADR practices and the creation of new ADRs by proposing best practices in specific domains.

Since its release as an IEC standard series, the OPC Unified Architecture (OPC UA) has become a widely adopted platform for both vertical and horizontal communications as well as data exchange in cyber-physical systems (CPS). A key benefit of OPC UA is its extensible object model, which offers the possibility to define custom information models, which may represent a specific production domain. As these generally represent interfaces for other applications, comprehensible documentation is an essential requirement in most cases.In this paper, we describe an approach to document, maintain and compare versions of OPC UA information models generated from existing code artifacts in the domain of serial machinery engineering. Base for the contribution of this work is the prototype of a documentation and model comparison tool (NodeDoc). We will describe its usage context and requirements as well as its main features. In addition, we discuss encountered challenges and provide insights into proposed solutions along with relevant lessons learned during the development of NodeDoc.

Accurate and up-to-date software documentation is an important factor in the maintenance and evolution of software systems. Especially with legacy software, documentation is often outdated or missing entirely and manual redocumentation is not feasible. In recent years, automatic code summaries based on artificial neural network (ANN) models have been proposed to address this problem, and metric-based evaluations suggest promising quality of the generated summaries. To evaluate the applicability of state-of-the-art code summarization in an industry context, we conduct an expert evaluation to assess the quality of the generated summaries for JPA program comprehension. We then compare the level of quality perceived by human experts for both predicted and reference summaries and discuss how these results are influenced by industry-specific requirements and how they correlate with automatically computed source code summary metrics. The results show that the quality of predicted summaries is predominantly (about 80%) poor in terms of accuracy and completeness. Moreover, the results support the generally increasing consensus that the widely used BLEU or ROUGE-L score is not a suitable means of evaluating the quality of code summarization. While these metrics are an adequate means of comparison with existing related work, they cannot reflect the human-perceived level of quality in practice.

Digitalization of knowledge work in communication-intensive domains such as intellectual property protection poses great challenges but also opportunities to improve today’s working environments. The legal domain is strongly characterized by knowledge work, whereby, despite a common legal framework, creativity of individual experts is decisive. This knowledge-intensive work deals with a great amount of data objects, not only as a working basis, but also as a result. While experts heavily follow individual working styles, they still rely on a vast amount of administrative tasks, which are carried out by the supporting staff. These tasks are expected to be performed regularly, reliably and without errors, despite necessary adjustments to the current case and the changing legal framework. Today, knowledge work and administrative tasks are typically supported by different tools that are hardly integrated. Therefore, the tracing of continuous work processes based on exchanged data objects is a great challenge. This traceability is crucial, not only for legal security reasons, but also to enable mining and learning of applicable knowledge about processes. In this paper, we propose a bottom-up approach, which applies a continuously evolving graph of integrated data objects and tasks to model and store static and dynamic aspects of administrative as well as knowledge work, and test the approach in a real-world setting in the domain of intellectual property. We further present initial results of a novel dependency-based mining approach to learn data-dependent task sequences in the graph-based model and discuss several methods for enabling privacy-preserving sharing and mining.

Abstract Source code comments contain key information about the underlying software system. Many redocumentation approaches, however, cannot exploit this valuable source of information. This is mainly due to the fact that not all comments have the same goals and target audience and can therefore only be used selectively for redocumentation. Performing a required classification manually, for example, in the form of heuristics, is usually time‐consuming and error‐prone and strongly dependent on programming languages and guidelines of concrete software systems. By leveraging machine learning (ML), it should be possible to classify comments and thus transfer valuable information from the source code into documentation with less effort but the same quality. We applied classical ML techniques but also deep learning (DL) approaches to legacy systems by transferring source code comments into meaningful representations using, for example, word embeddings but also novel approaches using quick response codes or a special character‐to‐image encoding. The results were compared with industry‐strength heuristic classification. As a result, we found that ML outperforms the heuristics in number of errors and less effort, that is, we finally achieve an accuracy of more than 95% for an image‐based DL network and even over 96% for a traditional approach using a random forest classifier.

Digitalisation of knowledge work, especially in communication-intensive domains is one of the greatest challenges, but also one of the greatest opportunities to improve today's working environments. This demands for a flexible system that supports both knowledge intensive creative work and highly individual processes. Smooth integration is hindered by the lack of the task context in knowledge management systems so far. Furthermore, a model to define and handle mental concepts, which are typically evolving during daily work, is missing, to allow for targeted use of appropriate knowledge in process tasks. In this paper, we propose a bottom-up approach to model and store the static and dynamic aspects of knowledge in terms of data objects and tasks that are connected with each other. The proposed solution leverages the flexibility of a graph-based model to enable open and continuously evolving user-centred processes for knowledge work, but also predefined administrative processes. Besides our approach, we show results from testing a prototypical implementation in a real-life setting in the domain of intellectual property management applications.

Source code comments contain key information about the underlying software system. Many redocumentation approaches, however, cannot exploit this valuable source of information. This is mainly due to the fact that not all comments have the same goals and target audience and can therefore only be used selectively for redocumentation. Performing a required classification manually, e.g. in the form of heuristic rules, is usually time-consuming and error-prone and strongly dependent on programming languages and guidelines of concrete software systems. By leveraging machine learning, it should be possible to classify comments and thus transfer valuable information from the source code into documentation with less effort but the same quality. We applied different machine learning techniques to a COBOL legacy system and compared the results with industry-strength heuristic classification. As a result, we found that machine learning outperforms the heuristics in number of errors and less effort.

The OMG standard Systems Modeling Language (SysML) has been on the market for about thirteen years. This standard is an extended subset of UML providing a graphical modeling language for designing complex systems by considering software as well as hardware parts. Over the period of thirteen years, many publications have covered various aspects of SysML in different research fields. The aim of this paper is to conduct a systematic mapping study about SysML to identify the different categories of papers, (i) to get an overview of existing research topics and groups, (ii) to identify whether there are any publication trends, and (iii) to uncover possible missing links. We followed the guidelines for conducting a systematic mapping study by Petersen et al. (Inf Softw Technol 64:1–18, 2015 ) to analyze SysML publications from 2005 to 2017. Our analysis revealed the following main findings: (i) there is a growing scientific interest in SysML in the last years particularly in the research field of Software Engineering, (ii) SysML is mostly used in the design or validation phase, rather than in the implementation phase, (iii) the most commonly used diagram types are the SysML-specific requirement diagram, parametric diagram, and block diagram, together with the activity diagram and state machine diagram known from UML, (iv) SysML is a specific UML profile mostly used in systems engineering; however, the language has to be customized to accommodate domain-specific aspects, (v) related to collaborations for SysML research over the world, there are more individual research groups than large international networks. This study provides a solid basis for classifying existing approaches for SysML. Researchers can use our results (i) for identifying open research issues, (ii) for a better understanding of the state of the art, and (iii) as a reference for finding specific approaches about SysML.

Abstract Business process management (BPM) provides many benefits for a company including productivity, efficiency, compliance, risk management, consistency, repeatability, and measurability. Many of these aspects also ensure and improve functional safety, privacy, and security in process execution. However, managed business processes lack flexibility; ie, changing a business process requires more effort than ad hoc execution and adaptation. Thus, especially in small companies, the acceptance rate of managed business processes is low. So we claim that in current BPM approaches, the structuredness ensuring functional safety and security concepts contradicts with the objective for flexibility. The main goal of the AdaBPM project is to address this contradiction by providing a formal approach to handle advanced adaptations and exception handling in business processes. The technical objectives of the project include (1) the definition of a basic framework being capable of handling different levels of adaptivity and considering security and safety requirements at the same time, (2) a rigorous process specification language and model refinement methods, (3) static business process adaptations, and (4) dynamic (or ad hoc) adaptations. The result of our research is a general approach for flexible BPM combined with the possibility to nevertheless manage the process and define safety and security restrictions.

The possibility to react to unexpected situations in business process execution is restricted since all possible process flows must be specified at design-time. Thus, there is need for a flexible approach that reflects the way in which human actors would handle discrepancies between real-life activities and their representation in business process definitions. In this paper, we propose a novel approach that supports dynamic business processes and is based on a framework comprising a process pattern library with domain-specific patterns and execution logs for mining related process instances. Given a running business process and an unexpected situation, the proposed approach provides a largely automatic adaptation of the business process by replacing failed activities with fitting process alternatives identified by exploring existing process knowledge. The feasibility of the approach is demonstrated by applying the main steps to a business scenario taken from the industry domain.

Business process variability is an active research area in the field of business process management and deals with variations and commonalities among processes of a given process family. Many theoretical approaches have been suggested in the last years; however, practical implementations are rare and limited in their functionality. In this paper, we propose a new approach for business process variability based on well-known graph transformation techniques and with focus on practical aspects like definition of variation points, linking and propagation of changes, as well as visual highlighting of differences in process variants. The suggested concepts are discussed within a case study comprising two graph transformation systems for generating process variants; one supports variability by restriction, the other supports variability by restriction and by extension. Both graph transformation systems are proven to be globally deterministic, but differ regarding their complexity. The overall approach is being implemented in the BPM suite of our partner company.

The different views and modeling techniques of both the business analysts and software developers are a common problem in business process modeling. Various modeling approaches result in communication problems, as well as redundancies and inconsistencies in system documentation. Thus, when modeling process-oriented enterprise applications seamless support of both expert groups is necessary. However, current business process management and workflow technologies are not fully integrated with user interaction, nor do they offer an appropriate data model. Based on the requirements of two industrial projects, we developed an integrated framework that combines the best practices from process-oriented and form-based approaches to overcome these shortcomings. The described framework supports the submit/response-style interaction paradigm and is independent of modeling languages and tools. In this work, we will present a detailed description of the proposed framework, its application to an industrial project and a discussion of related work.

In this paper we propose a typed approach to business process specification based on typed workflow charts. These can be exploited as a domain-specific programming language and facilitate tight integration between workflow definition and system dialogue programming. The approach also supports the integration of business process modeling and business process automation. We discuss two ways of exploiting this potential for integration, one is the design of an integrated business process management suite and the other is a software artifact tracker based on a view-based, multi-dimensional software modeling tool.