Feature interactions occur when one feature interferes with the intended operation of another feature. To detect such interactions, each new feature must be tested against existing features. The detected interactions must then be resolved; many existing approaches to resolving interactions require the feature set be prioritized. Unfortunately, the cost to determine a priority ordering for a feature set increases dramatically as the number of features increases. This thesis explores strategies to decrease the cost of prioritizing features, and thus facilitates priority-based solutions to resolving feature interactions. Specifically, this thesis introduces a categorization approach that reduces the complexity of determining priorities for a large set of features by decomposing the prioritization problem. Our categorization approach reduces this cost by using abstraction to divide the system's features into categories based on their main goal or functionality (e.g., block unwanted calls, present call information). Next, in order to detect and resolve the interactions that occur between these seemingly unrelated categories, we identify a set of principles for proper system behaviour that define acceptable behaviour in the global system. For example, a call that should be blocked by a call-screening feature should never result in a voice connection. The categories are then ordered, such that adherence to the principles is optimized. We show that using category priorities, to order a large feature set, correctly resolves interactions between individual features and significantly reduces the cost to determine priority orderings. The four significant contributions that this thesis makes are: 1) the categorization of features, 2) the principles of proper system behaviour, 3) automatic generation of priority orderings for categories, and 4) devising several optimizations that reduce the search space when exploring call simulations during the automatic generation of the priority orderings. These contributions are examined with respect to the telephony domain and result in the identification of 12 feature categories and 9 principles of proper system behaviour. A Prolog model was also created to run call simulations on the categories, using the identified principles as correctness criteria. Our case studies showed the reduced cost of our categorization approach is approximately 1/10^(55) % of the cost of a traditional approach. Given this significant reduction in the cost and the ability of our model to accurately reproduce the manually identified priority orderings, we can confidently argue that our categorization approach was successful. The three main limitations of our categorization approach are: 1) not all features (e.g., 911 features in telephony) can be categorized or some categories will contain a small number of features, 2) the generated priority ordering may still need to be analyzed by a human expert, and 3) the run time for our automatic generation of priority orderings remains factorial with respect to the size of the number of categories. However, these limitations are small in comparison to the savings generated by the categorization approach.

Feature interactions occur when one feature interferes with the intended operation of another feature. To detect such interactions, each new feature must be tested against existing features. The detected interactions must then be resolved; many existing approaches to resolving interactions require the feature set be prioritized. Unfortunately, the cost to determine a priority ordering for a feature set increases dramatically as the number of features increases. This thesis explores strategies to decrease the cost of prioritizing features, and thus facilitates priority-based solutions to resolving feature interactions. Specifically, this thesis introduces a categorization approach that reduces the complexity of determining priorities for a large set of features by decomposing the prioritization problem. Our categorization approach reduces this cost by using abstraction to divide the system's features into categories based on their main goal or functionality (e.g., block unwanted calls, present call information). Next, in order to detect and resolve the interactions that occur between these seemingly unrelated categories, we identify a set of principles for proper system behaviour that define acceptable behaviour in the global system. For example, a call that should be blocked by a call-screening feature should never result in a voice connection. The categories are then ordered, such that adherence to the principles is optimized. We show that using category priorities, to order a large feature set, correctly resolves interactions between individual features and significantly reduces the cost to determine priority orderings. The four significant contributions that this thesis makes are: 1) the categorization of features, 2) the principles of proper system behaviour, 3) automatic generation of priority orderings for categories, and 4) devising several optimizations that reduce the search space when exploring call simulations during the automatic generation of the priority orderings. These contributions are examined with respect to the telephony domain and result in the identification of 12 feature categories and 9 principles of proper system behaviour. A Prolog model was also created to run call simulations on the categories, using the identified principles as correctness criteria. Our case studies showed the reduced cost of our categorization approach is approximately 1/10^(55) % of the cost of a traditional approach. Given this significant reduction in the cost and the ability of our model to accurately reproduce the manually identified priority orderings, we can confidently argue that our categorization approach was successful. The three main limitations of our categorization approach are: 1) not all features (e.g., 911 features in telephony) can be categorized or some categories will contain a small number of features, 2) the generated priority ordering may still need to be analyzed by a human expert, and 3) the run time for our automatic generation of priority orderings remains factorial with respect to the size of the number of categories. However, these limitations are small in comparison to the savings generated by the categorization approach.

Systems that are assembled from independently developed features suffer from feature interactions, in which features affect one another's behaviour in surprising ways. To ensure that a system behaves as intended, developers need to analyze all potential interactions -- and many of the identified interactions need to be fixed and their fixes verified. The feature-interaction problem states that the number of potential interactions to be considered is exponential in the number of features in a system. Resolution strategies combat the feature-interaction problem by offering general strategies that resolve entire classes of interactions, thereby reducing the work of the developer who is charged with the task of resolving interactions. In this thesis, we focus on resolving interactions due to conflict. We present an approach, language, and implementation based on resolver modules modelled in the situation calculus in which the developer can specify an appropriate resolution for each variable under conflict. We performed a case study involving 24 automotive features, and found that the number of resolutions to be specified was much smaller than the number of possible feature interactions (6 resolutions for 24 features), that what constitutes an appropriate resolution strategy is different for different variables, and that the subset of situation calculus we used was sufficient to construct nontrivial resolution strategies for six distinct output variables.

Feature based modelling represents the future of CAD systems. However, operations such as modelling and editing can corrupt the validity of a feature-based model representation. Feature interactions are a consequence of feature operations and the existence of a number of features in the same model. Feature interaction affects not only the solid representation of the part, but also the functional intentions embedded within features. A technique is thus required to assess the integrity of a feature-based model from various perspectives, including the functional intentional one, and this technique must take into account the problems brought about by feature interactions and operations. The understanding, reasoning and resolution of invalid feature-based models requires an understanding of the feature interaction phenomena, as well as the characterisation of these functional intentions. A system capable of such assessment is called a feature-based representation validation system. This research studies feature interaction phenomena and feature-based designer's intents as a medium to achieve a feature-based representation validation system.

abstract: Models using feature interactions have been applied successfully in many areas such as biomedical analysis, recommender systems. The popularity of using feature interactions mainly lies in (1) they are able to capture the nonlinearity of the data compared with linear effects and (2) they enjoy great interpretability. In this thesis, I propose a series of formulations using feature interactions for real world problems and develop efficient algorithms for solving them. Specifically, I first propose to directly solve the non-convex formulation of the weak hierarchical Lasso which imposes weak hierarchy on individual features and interactions but can only be approximately solved by a convex relaxation in existing studies. I further propose to use the non-convex weak hierarchical Lasso formulation for hypothesis testing on the interaction features with hierarchical assumptions. Secondly, I propose a type of bi-linear models that take advantage of interactions of features for drug discovery problems where specific drug-drug pairs or drug-disease pairs are of interest. These models are learned by maximizing the number of positive data pairs that rank above the average score of unlabeled data pairs. Then I generalize the method to the case of using the top-ranked unlabeled data pairs for representative construction and derive an efficient algorithm for the extended formulation. Last but not least, motivated by a special form of bi-linear models, I propose a framework that enables simultaneously subgrouping data points and building specific models on the subgroups for learning on massive and heterogeneous datasets. Experiments on synthetic and real datasets are conducted to demonstrate the effectiveness or efficiency of the proposed methods. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2018

With the introduction of software into cars, many functions are now realized with reduced cost, weight and energy. The development of these software systems is done in a distributed manner independently by suppliers, following the traditional approach of the automotive industry, while the car maker takes care of the integration. However, the integration can lead to unexpected and unintended interactions among software systems, a phenomena regarded as feature interaction. This dissertation addresses the problem of the automatic detection of feature interactions for automotive active safety features. Active safety features control the vehicle's motion control systems independently from the driver's request, with the intention of increasing passengers' safety (e.g., by applying hard braking in the case of an identified imminent collision), but their unintended interactions could instead endanger the passengers (e.g., simultaneous throttle increase and sharp narrow steering, causing the vehicle to roll over). My method decomposes the problem into three parts: (I) creation of a definition of feature interactions based on the set of actuators and domain expert knowledge; (II) translation of automotive active safety features designed using a subset of Matlab's Stateflow into the input language of the model checker SMV; (III) analysis using model checking at design time to detect a representation of all feature interactions based on partitioning the counterexamples into equivalence classes. The key novel characteristic of my work is exploiting domain-specific information about the feature interaction problem and the structure of the model to produce a method that finds a representation of all different feature interactions for automotive active safety features at design time. My method is validated by a case study with the set of non-proprietary automotive feature design models I created. The method generates a set of counterexamples that represent the whole set of feature interactions in the case study.By showing only a set of representative feature interaction cases, the information is concise and useful for feature designers. Moreover, by generating these results from feature models designed in Matlab's Stateflow translated into SMV models, the feature designers can trace the counterexamples generated by SMV and understand the results in terms of the Stateflow model. I believe that my results and techniques will have relevance to the solution of the feature interaction problem in other cyber-physical systems, and have a direct impact in assessing the safety of automotive systems.

With the introduction of software into cars, many functions are now realized with reduced cost, weight and energy. The development of these software systems is done in a distributed manner independently by suppliers, following the traditional approach of the automotive industry, while the car maker takes care of the integration. However, the integration can lead to unexpected and unintended interactions among software systems, a phenomena regarded as feature interaction. This dissertation addresses the problem of the automatic detection of feature interactions for automotive active safety features. Active safety features control the vehicle's motion control systems independently from the driver's request, with the intention of increasing passengers' safety (e.g., by applying hard braking in the case of an identified imminent collision), but their unintended interactions could instead endanger the passengers (e.g., simultaneous throttle increase and sharp narrow steering, causing the vehicle to roll over). My method decomposes the problem into three parts: (I) creation of a definition of feature interactions based on the set of actuators and domain expert knowledge; (II) translation of automotive active safety features designed using a subset of Matlab's Stateflow into the input language of the model checker SMV; (III) analysis using model checking at design time to detect a representation of all feature interactions based on partitioning the counterexamples into equivalence classes. The key novel characteristic of my work is exploiting domain-specific information about the feature interaction problem and the structure of the model to produce a method that finds a representation of all different feature interactions for automotive active safety features at design time. My method is validated by a case study with the set of non-proprietary automotive feature design models I created. The method generates a set of counterexamples that represent the whole set of feature interactions in the case study.By showing only a set of representative feature interaction cases, the information is concise and useful for feature designers. Moreover, by generating these results from feature models designed in Matlab's Stateflow translated into SMV models, the feature designers can trace the counterexamples generated by SMV and understand the results in terms of the Stateflow model. I believe that my results and techniques will have relevance to the solution of the feature interaction problem in other cyber-physical systems, and have a direct impact in assessing the safety of automotive systems.

People trying to understand, develop and maintain software have faced greater challenges as the complexity of software systems has increased. These challenges include the difficulty of cleanly separating different intertwined parts of a system, or relating parts of the system spread across many modules. This makes it difficult to neatly identify an area of interest, which in turn makes it difficult to understand or edit that area. The ability to separate these areas of interest, called concerns, into their own modules has been shown to improve the situation. Several approaches have been developed to enable this separation: aspect-oriented programming allows program code to be divided into smaller modules that better match areas of interest; reverse engineering tools help programmers extract information from an existing system; requirements traceability tools track individual requirements through the development process. This thesis describes a technique that works in a wide variety of circumstances. This technique allows users to create simple diagrams that describe the concern and then annotate this diagram with query expressions which link the diagram to related development artefacts like source code or documents. This research has used the tool in a set of common scenarios and compared the results to those achieved using other approaches.

No description available.

Home automation is maturing with the increased deployment of networks and intelligent devices in the home. Along with new protocols and devices, new software services will emerge and work together releasing the full potential of networked consumer devices. Services may include home security, climate control or entertainment. With such extensive interworking the phenomenon known as service interaction, or feature interaction, appears. The problem occurs when services interfere with one another causing unexpected or undesirable outcomes. The main goal of this work is to detect undesired interactions between devices and services while allowing positive interactions between services and devices. If the interaction is negative, the approach should be able to handle it in an appropriate way. Being able to carry out interaction detection in the home poses certain challenges. Firstly, the devices and services are provided by a number of vendors and will be using a variety of protocols. Secondly, the configuration will not be fixed, the network will change as devices join and leave. Services may also change and adapt to user needs and to devices available at runtime. The developed approach is able to work with such challenges. Since the goal of the automated home is to make life simpler for the occupant, the approach should require minimal user intervention. With the above goals, an approach was developed which tackles the problem. Whereas previous approaches solving service interaction have focused on the service, the technique presented here concentrates on the devices and their surrounds, as some interactions occur through conflicting effects on the environment. The approach introduces the concept of environmental variables. A variable may be room temperature, movement or perhaps light. Drawing inspiration from the Operating Systems domain, locks are used to control access to the devices and environmental variables. Using this technique, undesirable interactions are avoided. The inclusion of the environment is a key element of this approach as many interactions can happen indirectly, through the environment. Since the configuration of a home’s devices and services is continually changing, developing an off-line solution is not practical. Therefore, an on-line approach in the form of an interaction manager has been developed. It is the manager’s role to detect interactions. The approach was shown to work successfuly. The manager was able to successfully detect interactions and prevent negative interactions from occurring. Interactions were detected at both device and service level. The approach is flexible: it is protocol independent, services are unaware of the manager, and the manager can cope with new devices and services joining the network. Further, there is little user intervention required for the approach to operate.