In many situations a decision maker (DM) would like to grade a component, or rank several components of the same type. Often a component type has many features, which are deemed as valuable by the DM. Other vital features are not known by the DM but are needed for the component to function. However, it should be possible to guide the DM to find the desired business solution, without putting a requirement of detailed knowledge of the component type on the DM. We propose a framework for component selection with the help of fuzzy decision making. The work is based on algorithms from fuzzy decision making, which we have adapted or extended. The framework was validated by practitioners, which found the framework useful. © 2018 The Author(s).

In order to design intent-driven systems, the understanding of how the data is generated is essential. Without the understanding of the data generation process, it is not possible to use interventions, and counterfactuals. Interventions, and counterfactuals, are useful tools in order to achieve an artificial intelligence which can improve the system itself. We will create an understanding, and a model, of how data about decisions are generated, as well as used, by human decision makers. The research data were collected with the help of focus group interviews, and questionnaires. The models were built and evaluated with the help of, bayesian statistics, probability programming, and discussions with the practitioners. When we are combining, probabilistic programming models, extended machine learning algorithms, and data science processes, into a directed acyclic graph, we can mimic the process of human generated decision data. We believe the usage of a directed acyclic graph, to combine the functions and models, is a good base for mimic human generated decision data. Our next step is to evaluate if flow-based programming can be used as a framework for realization of components, useful in intent-driven systems. © Springer Nature Switzerland AG 2020.

When an actor is selecting an action in order to fulfill its intents, in a given context, the actor’s knowledge and beliefs about the specific context will impact the possibility to achieve a desired outcome. The context is often affected by unobserved, or unmeasured, factors, which can impact the result of the desired outcome.

The context specific knowledge and beliefs an actor has about a domain, together with the possibilities to evaluate and learn which actions shall be taken, are packaged into a context frame. Our intention with this study is to evaluate an implementation of such a context frame. The context frame concept is meant to support actors to fulfill their intents in a given knowledge domain, by enforcing the needed, and available, actions which cause effects on the outcomes. We have built our implementation of the context frame on the OODA-loop, Pask’s conversation theory, and structural causal models, by using a Bayesian approach, and probabilistic programming.

The research approach is based on evaluation research. We evaluated our implementation with the help of a proof of concept. During the proof of concept we used data sets containing decisions about treatment and survival analysis regarding cancer patients, information obtained during focus group interviews, and questionnaire data.

The proof of concept used to evaluate our implementation of a context frame was regarded as successful and the concept of context frames deemed as useful.

Our division of a context frame in three parts, supported by four different types of analysis functions, made it easier to create a solution which supports evaluation and learning.

Together with Ericsson AB, we are using the design science framework when investigating how to create an intent-driven system for their business support system and its business studio. The aim is to present our initial results on how an extended OODA-loop can be used to realize a robust, but still flexible, software architecture for an intent-driven system. We explain how an extended OODA-loop is constructed and provide suggestions of how different part of it can be implemented. The initial results are promising but further research is needed to use the extended OODA-loop as reusable components in intent-driven systems. Our next step is to extend the generic methods with knowledge representation and reasoning capabilities. © 2019 The Author(s). Published by Elsevier B.V.

Context: Decisions taken by medical practitioners may be based on explicit and implicit rules. By uncovering these rules, a medical practitioner may have the possibility to explain its decisions in a better way, both to itself and to the person which the decision is affecting. Objective: We investigate if it is possible for a machine learning pipe-line to uncover rules used by medical practitioners, when they decide if a patient could be operated or not. The uncovered rules should have a linguistic meaning. Method: We are evaluating two different algorithms, one of them is developed by us and named “The membership detection algorithm”. The evaluation is done with the help of real-world data provided by a hospital. Results: The membership detection algorithm has significantly better relevance measure, compared to the second algorithm. Conclusion: A machine learning pipe-line, based on our algorithm, makes it possibility to give the medical practitioners an understanding, or to question, how decisions have been taken. With the help of the uncovered fuzzy decision algorithm it is possible to test suggested changes to the feature limits. © Springer International Publishing AG, part of Springer Nature 2018.

Business-driven software architectures are emerging and gaining importance for many industries. As softwareintensive solutions continue to be more complex and operate in rapidly changing environments, there is a pressure for increased business flexibility realized by more efficient software architecture mechanisms to keep up with the necessary speed of change. We investigate how improved context descriptions could be implemented in software components, and support important software development practices like business modeling and requirement engineering. This paper proposes context descriptions as an architectural support for improving the connection between business flexibility and software components. We provide initial results regarding software architectural mechanisms which can support context descriptions as well as the context description’s support for business-driven software architecture, and the business flexibility demanded by the business ecosystems.

Context: Software supporting an enterprise’s business, also known as a business support system, needs to support the correlation of activities between actors as well as influence the activities based on knowledge about the value networks in which the enterprise acts. This requires the use of policies and rules to guide or enforce the execution of strategies or tactics within an enterprise as well as in collaborations between enterprises. With the help of policies and rules, an enterprise is able to capture an actor’s intent in its business support system, and act according to this intent on behalf of the actor. Since the value networks an enterprise is part of will change over time the business intents’ life cycle states might change. Achieving the changes in an effective and efficient way requires knowledge about the affected intents and the correlation between intents.

Objective: The aim of the study is to identify how a business support system can support continuous changes to business intents. The first step is to find a theoretical model which serves as a foundation for intent-driven systems.

Method: We conducted a case study using a focus group approach with employees from Ericsson. This case study was influenced by the spiral case study process.

Results: The study resulted in a model supporting continuous definition and execution of an enterprise. The model is divided into three layers; Define, Execute, and a com- mon governance view layer. This makes it possible to support continuous definition and execution of business intents and to identify the actors needed to support the business intents’ life cycles. This model is supported by a meta-model for capturing information into viewpoints.

Conclusion: The research question is addressed by suggesting a solution supporting con- tinuous definition and execution of an enterprise as a model of value architecture compo- nents and business functions. The results will affect how Ericsson will build the business studio for their next generation business support systems.

Context: The aim of intent-driven systems is to capture stakeholders’ intents and transform these into a form that enables computer processing of the intents. Only then are different computer- based agents able to negotiate with each other on behalf of their respective stakeholders and their intents, and suggest a mutually beneficial agreement. This requires a separation of concerns between the parts of the system used to execute the stakeholder business, and the parts which are used to design the business based on stakeholder intents.

Objective: The aim is to find out which methods/techniques as well as enabling aspects, useful for an intent-driven system, that are covered by research literature.

Method: As a part of a design science study, a Systematic Literature Review is conducted.

Results: The existence of methods/techniques which can be used as building blocks to construct intent-driven systems exist in the literature. How these methods/techniques can interact with the aspects needed to enabling flexible realizations of intent-driven systems is not evident in the existing literature.

Conclusion: The synthesis shows a need for further research regarding semantic interchange of information, actor interaction in intent-driven systems, and the governance of intent-driven systems.