Smart Journey Mining: Towards successful digitalisation of services

The digitalisation of our society's service systems has fundamentally changed the way services are delivered to, and experienced by, humans. Although digital services are supposed to simplify our lives and increase our efficiency, they often frustrate and burden customers, users, and employees. The overall goal is to increase the quality of services and support the successful digitalisation of services by uniting research on process mining and customer journeys using new developments in logic-based analysis and artificial intelligence. The partners are SINTEF Digital, University of Oslo, and Eindhoven University of Technology. The project funds two fellowships: a post.doc. fellow (employed by SINTEF) and a PhD fellow (employed by University of Oslo). Based on user journeys, we aim to analyze, model, and observe how humans experience digital services, shifting the focus from the perspective of service providers and systems. We will trace data generated by users across various systems during their repeated interactions with a service, focusing on individual-level experiences. So far, our primary sources of journey data have been Telenor and Greps. Currently, we are engaging with additional service providers to extract customer journey data. Acquiring structured and complete user journeys has proven challenging, and we also plan to explore methods to generate artificial journey data. In the first phase of the project we developed a machine-readable format for CJML (Customer Journey Modeling Language), enabling us to specify journeys in a standardized format and check the compliance of actual journeys against planned ones. This xCJML format represents the first step towards the automated capture of user journeys. We have also developed a graphical tool that can visualize customer journeys. It provides functionality for import, export, and basic analysis of customer journeys. The tool represents what we denote the “Level 1 journey analyzer”. Building on this, we will apply logic-based techniques and machine learning to identify deviations and predict potential behaviors in the journeys. In the second half of the project, we will extend and use executable modelling languages and their associated analysis tools to describe, predict, and prescribe user journeys as concurrent processes. We will base these languages and analysis tools on formal methods and concurrency theory, which build on the foundation of theoretical computer science. As part of the work on predicting how user journeys are likely to unfold, we have used several deep learning methods that were highly successful in other fields and adapted them to predict the continuation of a user journey. The main contribution is that we put them in a system that allows to better compare their quality. Currently, we are extending these predictions method to better take into account the context of a user journey. We have developed a game theoretic framework for user journeys that captures real user journeys, extracted form event logs. The interaction between users and a service provider is modelled as a game, hereby both actors have their own, independent goals. The analysis of such games reveals failure points and possible improvements for the service. The analysis allows us to have a better understanding of the real journeys since we can construct action recommendations for the companies by building and analysing strategies generated from the game. Such strategies are part of the recommendations of how service providers could improve their interaction with their users, guiding them to their goals, and avoiding/removing parts of the observed journeys where users struggle. The project team has had a productive meeting with the scientific advisory board to discuss ongoing activities and strategize for the upcoming midway evaluation scheduled for January 2023. The project has produced a number of conference papers and presentations showcased at scientific events like IEEE SCC 2021, BPM 2021, MODELS 2021, ICPM 2021, SAC 2022, SEFM 2022, ICPM 2022, BPM 2023. One journal article has been published in Software and Systems Modeling (SoSyM). Finally, the PhD work of Paul Kobialka has led to no less than TWO BEST-PAPER-AWARDS at international conferences (!) and subsequent invitations to submit papers to special issues of journals. The project web site www.smartjourneymining.no contains a complete publication list, including a blog.

Public and private service providers are under pressure to digitalise their service offerings, and the opportunities for increased flexibility and efficiency are huge. However, the digital transformation continues to frustrate and burden users, citizens and customers. The Smart Journey Mining (SJM) vision is to support successful digitalisation by extending the research front in process mining and service science with recent advances in formal analytics and artificial intelligence. The overall goal is to develop a theoretical fundament and a process analytics framework based on formalized user journeys. Process mining allows organisations to reverse-engineer their business processes by analysing digital traces left in information systems. However, process mining as a research field does not have a profound user focus, and new methods and tools are needed. We will develop a theoretical foundation and prototype tools to capture and transform user data of varying granularity levels into a semantic user journey database. This will form the basis for developing user journey analytics at three levels: (1) a descriptive level for discovery and monitoring; (2) a predictive level for anomaly detection and possible behaviours; and (3) a prescriptive level for pre-emptive problem solving and service recovery. This SJM Analyser will be a tool for user-centred process analytics and simulation. By involving two service providers, the analytics will be validated against empirical investigations of the user experience in realistic service environments. SJM will provide researchers and analysts the necessary methods and tools to work across backend systems and detect patterns in vast volumes of user data. This will inform novel guidelines for the successful digitalisation of services. The SJM results will be made available to academia, industry and the public sector, thus supporting improved service quality in our society.