Business process management employs process models and event logs to represent the behavior of the information systems under study. Traditional case-centric notions consider the order of activities and events in isolated process instances. The emerging field of object-centric processes challenges this assumption by putting objects in the center. Object-centric process mining and modeling approaches identify the structure of co-evolving data objects that influence the behavior of an information system to provide a comprehensive view of the system behavior. Object-centricity has been investigated independently in process modeling and in process mining, which resulted in the coexistence of seemingly contradictory assumptions and definitions. As a community effort, this research manifesto relates and aligns existing terminologies, definitions, and perspectives to provide a common ground for current and future research in object-centric business process management. Based on the current state of research, we propose a conceptualization that sets process models and event logs in relation to the information system’s behavior and the execution data it generates. The conceptualization aims at aligning different terminologies and, thus, providing a basis to model and analyze behavioral characteristics. Building on this common ground, we identify open research challenges along the most relevant research areas in object-centric process management. For each research area, its current status is investigated and an outline of the most relevant research challenges is presented.

Limited availability of multilingual text corpora for training language models often leads to poor performance on downstream tasks due to undertrained representation spaces for languages other than English.This 'under-representation' has motivated recent cross-lingual transfer methods to leverage the English representation space by e.g.mixing English and 'non-English' tokens at the input level or extending model parameters to accommodate new languages.However, these approaches often come at the cost of increased computational complexity.We propose Fusion for Language Representations (FLARE) in adapters, a novel method that enhances representation quality and downstream performance for languages other than English while maintaining parameter efficiency.FLARE integrates source and target language representations within low-rank (LoRA) adapters using lightweight linear transformations, maintaining parameter efficiency while improving transfer performance.A series of experiments across representative crosslingual natural language understanding tasks, including natural language inference, questionanswering and sentiment analysis, demonstrate FLARE's effectiveness.FLARE achieves performance improvements of 4.9% for Llama 3.1 and 2.2% for Gemma 2 compared to standard LoRA fine-tuning on question-answering tasks, as measured by the exact match metric. 1 method for cross-lingual language understanding.

Predictive Process Monitoring (PPM) in Process Mining (PM) uses predictive analytics to forecast business process progression. A key challenge is suffix prediction, forecasting future event sequences with activity labels, timestamps, and remaining runtime. Current techniques often focus on one-step-ahead predictions, rely on iterative feedback loops for suffix generation, and underutilize payload data. Additionally, many lag behind recent advancements in model architectures, sticking to LSTM-based models. Addressing these gaps, we propose SuTraN, a novel transformer architecture for PPM suffix prediction. SuTraN avoids iterative prediction loops and utilizes all available data, including event features, to forecast entire event suffixes in a single forward pass. Our approach integrates autoregressive suffix generation, data awareness, and seq2seq learning. Experimental results on real-life event logs demonstrate SuTraN’s superior performance in suffix prediction, highlighting its contributions often overlooked in current research.

Research interest in machine learning (ML) for graphs has skyrocketed in recent years. However, non-Euclidean graph structures inhibit the application of traditional ML algorithms. Consequently, scholars introduced graph learning algorithms tailored to network data, such as graph neural networks (GNNs). Most GNNs are designed for homogeneous and homophilous graphs and are evaluated on small, static, and balanced datasets, deviating from real-world conditions and industry applications. This paper introduces SHINE, a scalable heterogeneous inductive GNN for large imbalanced datasets. SHINE addresses four key challenges: scalability, network heterogeneity, inductive learning on dynamic graphs, and imbalanced node classification. SHINE comprises three core components: 1) a sampler based on nearest-neighbor (NN) search, 2) a heterogeneous GNN (HGNN) layer with a novel relationship aggregator, and 3) aggregator functions tailored to skewed class distributions. The components of SHINE are evaluated on benchmark datasets, while the integrated benefits of SHINE are demonstrated on two fraud detection datasets.

We introduce CORE, a dataset for few-shot relation classification (RC) focused on company relations and business entities.CORE includes 4,708 instances of 12 relation types with corresponding textual evidence extracted from company Wikipedia pages.Company names and business entities pose a challenge for few-shot RC models due to the rich and diverse information associated with them.For example, a company name may represent the legal entity, products, people, or business divisions depending on the context.Therefore, deriving the relation type between entities is highly dependent on textual context.To evaluate the performance of state-of-the-art RC models on the CORE dataset, we conduct experiments in the few-shot domain adaptation setting.Our results reveal substantial performance gaps, confirming that models trained on different domains struggle to adapt to CORE.Interestingly, we find that models trained on CORE showcase improved out-of-domain performance, which highlights the importance of high-quality data for robust domain adaptation.Specifically, the information richness embedded in business entities allows models to focus on contextual nuances, reducing their reliance on superficial clues such as relation-specific verbs.In addition to the dataset, we provide relevant code snippets to facilitate reproducibility and encourage further research in the field. 1

This paper contains explicit statements that are potentially offensive.

Process mining encompasses a series of tasks aimed at automatically unveiling knowledge about business processes from event logs registered in underlying information systems deployed in organizations. As well as numerous machine learning approaches, process mining approaches often require a vector space as input. However, the choice of the representational scheme to map event log information to a vector space sig-nificantly influences the quality of the results. This mapping poses challenges due to the diverse information in event logs and the intricate relationships within a business process. Relying solely on automated approaches may overlook relevant information, necessitating the incorporation of domain knowledge from external sources. Unfortunately, this incorporation introduces complexity. To address these inherent issues in constructing adequate vector spaces for process mining, this paper proposes a novel approach leveraging graph embedding to organize process-related information. To this end, we present a novel and highly flexible graph structure to represent process-related information that is then mapped to a dense vector space by applying the metapath2vec algorithm. The resulting dense vector space was compared to traditional vector spaces in an exploratory study, in which we solved the trace clustering task. We employ the N3 measure to assess the quality of the clusters and to verify whether domain knowledge is adequately represented in the vector spaces. The results demonstrate a superior potential of the dense vector spaces obtained via graph embedding to adequately organize the information to be submitted to the trace clustering task.

In recent years, many predictive models have been successfully applied to predictive process monitoring, enabling tasks such as predicting the next activity, remaining time, or the future state of a process instance (case). However, recent developments have shown the vulnerability of these models to adversarial attacks, causing algorithms to make incorrect predictions. This paper addresses this issue by leveraging adversarial examples to evaluate the predictive performance of predictive process monitoring models in the face of adversarial threats. Although augmenting training data with adversarial examples has proven effective in defending against specific adversarial attacks, it often remains insufficient in mitigating vulnerabilities to other types of attacks. Our proposed approach explores the use of manifold learning techniques to restrict these examples within the range of data on which the model is trained. By learning from these specifically engineered (hidden) attacks, we seek to develop models that maintain accuracy on new, unseen data while effectively improving adversarial robustness against potential threats.

Process analytics is a collection of data-driven techniques for, among others, making predictions for individual process instances or overall process models. At the instance level, various novel techniques have been recently devised, tackling analytical tasks such as next activity, remaining time, or outcome prediction. However, there is a notable void regarding predictions at the process model level. It is the ambition of this article to fill this gap. More specifically, we develop a technique to forecast the entire process model from historical event data. A forecasted model is a will-be process model representing a probable description of the overall process for a given period in the future. Such a forecast helps, for instance, to anticipate and prepare for the consequences of upcoming process drifts and emerging bottlenecks. Our technique builds on a representation of event data as multiple time series, each capturing the evolution of a behavioural aspect of the process model, such that corresponding time series forecasting techniques can be applied. Our implementation demonstrates the feasibility of process model forecasting using real-world event data. A user study using our Process Change Exploration tool confirms the usefulness and ease of use of the produced process model forecasts.

Developing LSTM neural networks that can accurately predict the future trajectory of ongoing cases and their remaining runtime is an active area of research in predictive process monitoring. In this work a novel complete remaining trace prediction (CRTP) LSTM is proposed. This model is trained to directly predict the complete remaining trace and runtime of cases in contrast to single event prediction as is considered in previously published research on this topic. This makes the CRTP-LSTM robust in terms of utilizing all available attributes of previously observed events for prediction, consequently it can be considered natively data aware. In an extensive experimental assessment the authors show that CRTP-LSTMs consistently outperform other considered approaches for both remaining trace and runtime prediction. Furthermore, the authors show that including all available information contained in previously observed events has a positive impact on the performance of the CRTP-LSTM model. This indicates that valuable information can be extracted from attributes of events in order to make more accurate trace and runtime predictions. This opens up interesting avenues for future research including the incorporation of inter-case features into a modeling setup when predicting the remaining trace and runtime of cases.

Novel joint techniques capture both the microscopic context and the mesoscopic structure of networks by leveraging two previously separated fields of research: node representation learning (NRL) and community detection (CD). However, several limitations exist in the literature. First, a comprehensive comparison between these joint NRL-CD techniques is non-existent. Second, baseline techniques, datasets, evaluation metrics, and classification algorithms differ significantly between each method. Thirdly, the literature lacks a synchronized experimental approach, thus rendering comparison between these methods strenuous. To overcome these limitations, we present a uni-fied experimental setup mutually comparing six joint NRL-CD techniques and comparing them with corresponding NRL/CD baselines in three different settings: non-overlapping and over-lapping CD and node classification. Our results show that joint methods underperform on the node classification task but achieve relatively solid results for overlapping community detection. Our research contribution is two-fold: first, we show specific weaknesses of selected joint techniques in different tasks and data sets; and second, we suggest a more thorough experimental setup to benchmark joint techniques with simpler NRL and CD techniques.

As machine and deep learning models are increasingly leveraged in predictive process monitoring, the focus has shifted towards making these models explainable. The successful adoption of a model is dependent on whether decision-makers can trust the predictions and explanations made. However, recent studies have shown that deep learning models are vulnerable to adversarial attacks -small perturbations to the inputs-which trick deep learning algorithms into making incorrect predictions. An additional crucial property is that the explanations are robust against these adversarial attacks when the model decision was not affected. Therefore, this paper introduces a robustness assessment framework by investigating the impact of adversarial attacks on the robustness of predictive accuracy and explanations used in the field of predictive process monitoring. First, adversarial examples of cases in the independent test set are generated to examine the robustness of the predictive model against intentionally manipulated data. Next, the predictive models are compared with similar models trained on data imputed with adversarial attacks. We monitor the impact on predictive performance in terms of AUC at different stages of the case execution. Finally, the robustness of the explanations is calculated as the distance between the original explanations and the explanations extracted from the model trained on attacked data. We test multiple machine and deep learning techniques, namely the transparent logistic regression, random forests with Shapley values, and LSTM neural networks with attention. Results show that especially neural networks suffer from adversarial attacks, and the former two are mostly robust in terms of both predictive accuracy and explanations.

Blended learning is gaining ground in contemporary education. However, studies on predictive learning analytics in the context of blended learning remain relatively scarce compared to Massive Open Online Courses (MOOCs), where such applications have gained a strong foothold. Data sets obtained from blended learning environments suffer from a high dimensionality and typically expose a limited number of instances, which makes predictive analysis a challenging task. In this work, we explore the log data of a master-level blended course to predict the students' grades based entirely on the data obtained from an online module (a small private online course), using and comparing logistic regression and random forest-based predictive models. The results of the analysis show that, despite the limited data, success vs. fail predictions can be made as early as in the middle of the course. This could be used in the future for timely interventions, both for failure prevention as well as for reinforcing positive learning behaviours of students.

Sequence classification deals with the task of finding discriminative and concise sequential patterns. To this purpose, many techniques have been proposed, which mainly resort to the use of partial orders to capture the underlying sequences in a database according to the labels. Partial orders, however, pose many limitations, especially on expressiveness, i.e., the aptitude towards capturing certain behavior, and on conciseness, i.e., doing so in a compact and informative way. These limitations can be addressed by using a better representation. In this paper, we present the interesting Behavioral Constraint Miner (iBCM), a sequence classification technique that discovers patterns using behavioral constraint templates. The templates comprise a variety of constraints and can express patterns ranging from simple occurrence, to looping and position-based behavior over a sequence. Furthermore, iBCM also captures negative constraints, i.e., absence of particular behavior. The constraints can be discovered by using simple string operations in an efficient way. Finally, deriving the constraints with a window-based approach allows to pinpoint where the constraints hold in a string, and to detect whether patterns are subject to concept drift. Through empirical evaluation, it is shown that iBCM is better capable of classifying sequences more accurately and concisely in a scalable manner.

In many real-world networks, it is important to explicitly differentiate between positive and negative links, thus considering the observed networks as signed. To derive useful features, just as in the case of unsigned networks, representation learning can be used to learn meaningful representations of a network that characterize its underlying topology. Several methods for learning representations on signed networks have already been proposed but have not been systematically benchmarked together before. Hence, in this paper, we bridge this literature gap providing a quantitative and qualitative benchmark of the four most prominent representation learning methods for signed networks. Results on three different datasets for link sign prediction showcase the superiority of the StEM method over its competitors both from a predictive performance and runtime perspective.

In process mining, one is often confronted with datasets that contain high degrees of control-flow variation. This causes the results of subsequent process mining steps to be less accurate or harder to understand. Trace clustering, or dividing the log into more homogeneous groups, can help limit this issue. Trace clustering techniques are commonly subdivided into two types: distance-based techniques and model-driven approaches. In this paper, a novel trace clustering technique is presented, that combines aspects of both paradigms. The core idea is to first learn so-called super-instances using a simple distance-driven technique and subsequently apply a model-driven technique to cluster the super-instances. Our technique not only shows qualitative improvements of the obtained clusterings, but also significantly improves the scalability, as shown in an experimental evaluation using real-life event logs.

In many real-life applications it is crucial to be able to, given a collection of link states of a network in a certain time period, accurately predict the link state of the network at a future time. This is known as dynamic link prediction, which compared to its static counterpart is more complex, as capturing the temporal characteristics is a non-trivial task. This explains while still majority of today's research in network representation learning focuses on static setting ignoring temporal information. In this work, we focus on one such case and aim at extending node2vec, representation learning method successfully applied for static link prediction, to a dynamic setup. This extended method is applied and validated on several real-life networks with different properties. Results show that taking into account dynamic aspect outperforms static approach. Additionally, based on the network properties, recommendations are given for the node2vec parameters.

Most of the recent studies on churn prediction in telco utilize social networks built on top of the call (and/or SMS) graphs to derive informative features. However, extracting features from large graphs, especially structural features, is an intricate process both from a methodological and computational perspective. Due to the former, feature extraction in the current literature has mainly been addressed in an ad-hoc and hand-crafted manner. Due to the latter, the full potential of the structural information is unexploited. In this work, we incorporate both interaction and structural information by devising two different ways of enriching original graphs with interaction information, delineated by the well-known RFM model. We circumvent the process of extensive manual feature engineering by enriching the networks and improving the scalability of the renowned node2vec approach to learn node representations. The obtained results demonstrate that our enriched network outperforms baseline RFM-based methods.

Providing individual and immediate feedback to students is a critical factor for improving knowledge and skills acquisition in higher education. However, a growing number of students with very different heterogeneous profiles and unequal problem solving skills, as well as the lack of teaching resources, make it very challenging and sometimes even impossible to give immediate feedback at individual level. This paper presents a synthesis and progress of a long-term project that addresses this challenge in the context of conceptual modelling by developing SAiLE@CoMo, a smart and adaptive learning environment. By crafting innovative process analytics techniques and expert knowledge on feedback automation, SAiLE@CoMo can automatically provide personalized and immediate feedback to leaners.

Organisations are constantly seeking new ways to improve operational efficiencies. This research study investigates a novel way to identify potential efficiency gains in business operations by observing how they are carried out in the past and then exploring better ways of executing them by taking into account trade-offs between time, cost and resource utilisation. This paper demonstrates how they can be incorporated in the assessment of alternative process execution scenarios by making use of a cost environment. A genetic algorithm-based approach is proposed to explore and assess alternative process execution scenarios, where the objective function is represented by a comprehensive cost structure that captures different process dimensions. Experiments conducted with different variants of the genetic algorithm evaluate the approach's feasibility. The findings demonstrate that a genetic algorithm-based approach is able to make use of cost reduction as a way to identify improved execution scenarios in terms of reduced case durations and increased resource utilisation. The ultimate aim is to utilise cost-related insights gained from such improved scenarios to put forward recommendations for reducing process-related cost within organisations.

Process mining encompasses the research area which is concerned with knowledge discovery from event logs. One common process mining task focuses on conformance checking, comparing discovered or designed process models with actual real-life behavior as captured in event logs in order to assess the “goodness” of the process model. This paper introduces a novel conformance checking method to measure how well a process model performs in terms of precision and generalization with respect to the actual executions of a process as recorded in an event log. Our approach differs from related work in the sense that we apply the concept of so-called weighted artificial negative events toward conformance checking, leading to more robust results, especially when dealing with less complete event logs that only contain a subset of all possible process execution behavior. In addition, our technique offers a novel way to estimate a process model's ability to generalize. Existing literature has focused mainly on the fitness (recall) and precision (appropriateness) of process models, whereas generalization has been much more difficult to estimate. The described algorithms are implemented in a number of ProM plugins, and a Petri net conformance checking tool was developed to inspect process model conformance in a visual manner.

Process mining encompasses the research area which is concerned with knowledge discovery from information system event logs. Within the process mining research area, two prominent tasks can be discerned. First of all, process discovery deals with the automatic construction of a process model out of an event log. Secondly, conformance checking focuses on the assessment of the quality of a discovered or designed process model in respect to the actual behavior as captured in event logs. Hereto, multiple techniques and metrics have been developed and described in the literature. However, the process mining domain still lacks a comprehensive framework for assessing the goodness of a process model from a quantitative perspective. In this study, we describe the architecture of an extensible framework within ProM, allowing for the consistent, comparative and repeatable calculation of conformance metrics. For the development and assessment of both process discovery as well as conformance techniques, such a framework is considered greatly valuable.

Process discovery is the learning task that entails the construction of process models from event logs of information systems. Typically, these event logs are large data sets that contain the process executions by registering what activity has taken place at a certain moment in time. By far the most arduous challenge for process discovery algorithms consists of tackling the problem of accurate and comprehensible knowledge discovery from highly flexible environments. Event logs from such flexible systems often contain a large variety of process executions which makes the application of process mining most interesting. However, simply applying existing process discovery techniques will often yield highly incomprehensible process models because of their inaccuracy and complexity. With respect to resolving this problem, trace clustering is one very interesting approach since it allows to split up an existing event log so as to facilitate the knowledge discovery process. In this paper, we propose a novel trace clustering technique that significantly differs from previous approaches. Above all, it starts from the observation that currently available techniques suffer from a large divergence between the clustering bias and the evaluation bias. By employing an active learning inspired approach, this bias divergence is solved. In an assessment using four complex, real-life event logs, it is shown that our technique significantly outperforms currently available trace clustering techniques.

Recent years have witnessed the ability to gather an enormous amount of data in a large number of domains. Also in the field of business process management, there exists an urgent need to beneficially use these data to retrieve actionable knowledge about the actual way of working in the context of a certain business process. The research field concerned is process mining, which can be defined as a whole family of analysis techniques for extracting knowledge from information system event logs. In this paper, we present a solution strategy to leverage traditional process discovery techniques in the flexible environment of incident management processes. In such environments, it is typically observed that single model discovery techniques are incapable of dealing with the large number of different types of execution traces. Accordingly, we propose a combination of trace clustering and text mining to enhance process discovery techniques with the purpose of retrieving more useful insights from process data.

Within process mining research, one of the most important fields of study is process discovery, which can be defined as the extraction of control-flow models from audit trails or information system event logs. The evaluation of discovered process models is an essential but difficult task for any process discovery analysis. With this paper, we propose a novel approach for evaluating discovered process models based on artificially generated negative events. This approach allows for the definition of a behavioral F-measure for discovered process models, which is the main contribution of this paper.