Complex software systems evolve frequently, e.g., when introducing new features or fixing bugs during maintenance. However, understanding the impact of such changes on system behavior is often difficult. Many approaches have thus been proposed that analyze systems before and after changes, e.g., by comparing source code, model-based representations, or system execution logs. In this paper, we propose an approach for comparing run-time constraints, synthesized by a constraint mining algorithm, based on execution logs recorded before and after changes. Specifically, automatically mined constraints define the expected timing and order of recurring events and the values of data elements attached to events. Our approach presents the differences of the mined constraints to users, thereby providing a higher-level view on software evolution and supporting the analysis of the impact of changes on system behavior. We present a motivating example and a preliminary evaluation based on a cyber-physical system controlling unmanned aerial vehicles. The results of our preliminary evaluation show that our approach can help to analyze changed behavior and thus contributes to understanding software evolution.

Most software development organizations nowadays use issue-tracking tools to manage software processes throughout the life-cycle. Still, understanding development processes, keeping track of process execution, and reacting to deviations in projects remains challenging. In particular, the actual process usually differs from the process perceived by developers, making it hard to define the processes developers are expected to carry out. This is further challenged by frequently changing processes and process variations in different projects and teams. In this paper we describe an empirical study in which we applied a constraint mining approach from the field of software monitoring to automatically extract process definitions in the form of constraints. Specifically, we applied the approach to datasets extracted from four real-world projects (using the Jira issue-tracking tool) in a company developing a recreational activities platform. The mined constraints describe the boundaries of the actual processes and thus help to understand process behavior. Constraints can be frequently re-mined to understand process evolution. The mined constraints can also be used to monitor future processes to detect problems in the development process early on. We involved a domain expert to evaluate the usefulness of our results and investigated to what extent the mined constraints reflect the official development process of the company. We also report mining results for different issue types, across projects, and over different time windows.

Many requirements monitoring approaches have been proposed that check key properties of systems and their interactions at runtime. Some of these approaches also visualize monitoring results and provide details on requirements violations to end users. However, only few studies exist about the usefulness of requirements monitoring tools for practitioners, particularly regarding visualization. In this paper, we present a user study we have conducted with both industrial practitioners and researchers to assess the usefulness of visualization capabilities we have been developing for an event-based requirements monitoring tool. These capabilities allow users to monitor the status of the involved systems, to view trends and statistics, and to inspect the events and data that led to specific violations when diagnosing their root cause. We first performed a walkthrough of the tool using the cognitive dimensions of notations framework from the field of human-computer interaction. We then conducted a user study involving five software engineers of a large company from the automation software domain and four researchers. Using the tool's visualization capabilities all subjects succeeded in monitoring a real-world automation system and in diagnosing violations. Subjects regarded the visualization capabilities as essential for understanding the behavior of a complex system. Based on the study results we derive implications, opportunities, and risks of using visualization in requirements monitoring tools.

Online activities such as social networking, online shopping, and consuming multi-media create digital traces, which are often analyzed and used to improve user experience and increase revenue, e. g., through better-fitting recommendations and more targeted marketing. Analyses of digital traces typically aim to find user traits such as age, gender, and nationality to derive common preferences. We investigate to which extent the music listening habits of users of the social music platform Last.fm can be used to predict their age, gender, and nationality. We propose a feature modeling approach building on Term Frequency-Inverse Document Frequency (TF-IDF) for artist listening information and artist tags combined with additionally extracted features. We show that we can substantially outperform a baseline majority voting approach and can compete with existing approaches. Further, regarding prediction accuracy vs. available listening data we show that even one single listening event per user is enough to outperform the baseline in all prediction tasks. We also compare the performance of our algorithm for different user groups and discuss possible prediction errors and how to mitigate them. We conclude that personal information can be derived from music listening information, which indeed can help better tailoring recommendations, as we illustrate with the use case of a music recommender system that can directly utilize the user attributes predicted by our algorithm to increase the quality of it’s recommendations.

Unmanned aerial vehicles (UAVs) are becoming increasingly pervasive in everyday life, supporting diverse use cases such as aerial photography, delivery of goods, or disaster reconnaissance and management. UAVs are cyber-physical systems (CPS): they integrate computation (embedded software and control systems) with physical components (the UAVs flying in the physical world). UAVs in particular and CPS in general require monitoring capabilities to detect and possibly mitigate erroneous and safety-critical behavior at runtime. Existing monitoring approaches mostly do not adequately address UAV CPS characteristics such as the high number of dynamically instantiated components, the tight int elements, and the massive amounts of data that need to be processed. In this paper we report results of a case study on monitoring in UAVs. We discuss CPS-specific monitoring challenges and present a prototype we implemented by extending \reminds, a framework for software monitoring so far mainly used in the domain of metallurgical plants. Additionally, we demonstrate the applicability and scalability of our approach by monitoring a real control and management system for UAVs in simulations with up to 30 drones flying in an urban area.

The behavior of complex software-intensive systems of systems often only fully emerges during operation, when all systems interact with each other and with their environment. Runtime monitoring approaches are thus used to detect deviations from the expected behavior, which is commonly defined by engineers, e.g., using temporal logic or domain-specific languages. However, the deep domain knowledge required to specify constraints is often not available during the development of systems of systems with multiple teams independently working on heterogeneous components. In this paper, we thus describe our ongoing PhD research to automatically mine constraints for runtime monitoring from recorded events. Our approach mines constraints on event occurrence, timing, data, and combinations of these properties. The approach further presents the mined constraints to users offering multiple ranking strategies and can also be used to support users in system evolution scenarios.

Industrial software systems are often systems of systems (SoS) whose full behavior only emerges during operation. They therefore require monitoring techniques to observe systems and detect deviations from their requirements. The focus of existing monitoring approaches, however, is mainly on detecting violations of expected behavior, while support for diagnosing violations is typically limited or even neglected. Diagnosis is particularly challenging in SoS due to their technological heterogeneity and the diversity of development tools in use. Uncovering the root cause of a violation typically requires developers to trace violations to artifacts such as source code or requirements documents, which is difficult without detailed domain knowledge. In this paper we describe our experiences of developing a tool-supported approach facilitating the diagnosis of requirements violations in SoS. We describe how we complemented a requirements monitoring model with a system artifact model relating SoS artifacts needed for diagnosis with monitored events. We customized our approach to an industrial SoS and conducted a scenario-based walkthrough with engineers developing the SoS and engineers and researchers unfamiliar with it. The results of our evaluation have shown that our approach can significantly ease diagnosing violations in a real-world SoS.

The full behavior of software-intensive systems of systems (SoS) emerges during operation only. Runtime monitoring approaches have thus been proposed to detect deviations from the expected behavior. They commonly rely on temporal logic or domain-specific languages to formally define requirements, which are then checked by analyzing the stream of monitored events and event data. Some approaches also allow developers to generate constraints from declarative specifications of the expected behavior. However, independent of the approach, deep domain knowledge is required to specify the desired behavior. This knowledge is often not accessible in SoS environments with multiple development teams independently working on different, heterogeneous systems. In this New Ideas Paper we thus describe an approach that automatically mines constraints for runtime monitoring from event logs recorded in SoS. Our approach builds on ideas from specification mining, process mining, and machine learning to mine different types of constraints on event occurrence, event timing, and event data. The approach further presents the mined constraints to users in an existing constraint language and it ranks the constraints using different criteria. We demonstrate the feasibility of our approach by applying it to event logs from a real-world industrial SoS.

Industrial software systems are often systems of systems (SoS) whose full behavior only emerges at runtime. The systems and their interactions thus need to be continuously monitored and checked during operation to determine compliance with requirements. Many requirements monitoring approaches have been proposed. However, only few of these come with tools that present and visualize monitoring results and details on requirements violations to end users such as industrial engineers. In this tool demo paper we present visualization capabilities we have been developing motivated by industrial scenarios. Our tool complements ReMinds, an existing requirements monitoring framework, which supports collecting, aggregating, and analyzing events and event data in architecturally heterogeneous SoS. Our visualizations support a `drill-down' scenario for monitoring and diagnosis: starting from a graphical status overview of the monitored systems and their relations, engineers can view trends and statistics about performed analyses and diagnose the root cause of problems by inspecting the events and event data that led to a specific violation. Initial industry feedback we received confirms the usefulness of our tool support. Demo video: https://youtu.be/iv7kWzeNkdk..