Large language models have driven major advances in Text-to-SQL generation. However, they suffer from high computational cost, long latency, and data privacy concerns, which make them impractical for many real-world applications. A natural alternative is to use small language models (SLMs), which enable efficient and private on-premise deployment. Yet, SLMs often struggle with weak reasoning and poor instruction following. Conventional reinforcement learning methods based on sparse binary rewards (0/1) provide little learning signal when the generated SQLs are incorrect, leading to unstable or collapsed training. To overcome these issues, we propose FINER-SQL, a scalable and reusable reinforcement learning framework that enhances SLMs through fine-grained execution feedback. Built on group relative policy optimization, FINER-SQL replaces sparse supervision with dense and interpretable rewards that offer continuous feedback even for incorrect SQLs. It introduces two key reward functions: a memory reward, which aligns reasoning with verified traces for semantic stability, and an atomic reward, which measures operation-level overlap to grant partial credit for structurally correct but incomplete SQLs. This approach transforms discrete correctness into continuous learning, enabling stable, critic-free optimization. Experiments on the BIRD and Spider benchmarks show that FINER-SQL achieves up to 67.73% and 85% execution accuracy with a 3B model – matching much larger LLMs while reducing inference latency to 5.57 s/sample. These results highlight a cost-efficient and privacy-preserving path toward high-performance Text-to-SQL generation. Our code is available at https://github.com/thanhdath/finer-sql.

An object-centric behavioral constraint (OCBC) model provides an integrated view on entities handled in a process-oriented information system. It defines behavioral constraints between activities operating on data objects and cardinality constraints for relations over activities and objects. Existing techniques for OCBC conformance checking evaluate conditions from the model on an event log. As such, conformance cannot be evaluated directly over the relational databases that typically store the event data. Instead, log construction is required, which induces computational overhead and decouples data management and process analysis. In this paper, we show how to incorporate OCBC conformance checking directly within relational database systems. To this end, we introduce conformance queries for two representations of process data that differ in how they capture the relationships between events and objects, i.e., explicitly by dedicated tables or implicitly through foreign keys. Moreover, we present strategies for continuous conformance monitoring under database updates. We demonstrate the feasibility of our approach using event data generated based on the TPC-H benchmark.

Scientific Workflows (SWF) encapsulate data processing tasks by organizing various tools, operators, and data in a logical flow. Due to their complex and domain-specific nature, developing SWFs remains laborious. Current code-generating large language models (Code LLMs) struggle to assist users in developing these workflows. This limitation arises primarily from the insufficient availability of relevant training data in public repositories, making it challenging for LLMs to learn specialized patterns and domain-specific logic. To address this, we propose FlowPilot , a suggestion framework for developing SWFs that assists developers by suggesting the next operator. Our system complements Code LLMs by enriching the source code to generate more accurate results. FlowPilot leverages a similarity knowledge base (SKB) that indexes historical workflows to find the ones matching the current context. To generate relevant suggestions, FlowPilot employs a statistical approach based on Markov chains to identify the most likely next step. As a proof of concept, we evaluated our system on NextFlow workflows and the results demonstrate the effectiveness of FlowPilot . It outperforms state-of-the-art code-generating models, e.g., Llama-4, and traditional methods, e.g., association rule mining techniques.

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.

The detection of sequential patterns in data is a basic functionality of modern data processing systems for complex event processing (CEP), OLAP, and retrieval-augmented generation (RAG). In practice, pattern matching is challenging, since common applications rely on a large set of patterns that shall be evaluated with tight latency bounds. At the same time, matching needs to maintain state, i.e., intermediate results, that grows exponentially in the input size. Hence, systems turn to best-effort processing, striving for maximal recall under a latency bound. Existing techniques, however, consider each pattern in isolation, neglecting the optimization potential induced by state sharing in pattern matching. In this paper, we present SHARP, a library that employs state reduction to achieve efficient best-effort pattern matching. To this end, SHARP incorporates state sharing between patterns through a new abstraction, coined pattern-sharing degree (PSD). At runtime, this abstraction facilitates the categorization and indexing of partial pattern matches. Based thereon, once a latency bound is exceeded, SHARP realizes best-effort processing by selecting a subset of partial matches for further processing in constant time. In experiments with real-world data, SHARP achieves a recall of 97

Process mining aims to obtain insights from event logs through the automated analyses of recorded process data in information systems, with the ultimate aim to improve business processes running in organisations. However, real-life event logs are often incomplete, noisy, or ambiguous, such as missing timestamps or having ambiguous event labels, which traditional deterministic models cannot capture. Recent process mining developments have considered uncertainty in process mining artifacts more explicitly: in logs of recorded process behaviour, uncertainty may implicitly or explicitly influence process mining outcomes, while in process models, explicit uncertainty allows analysts to interpret and value outcomes. In this paper, we provide a conceptual foundation for uncertainty in process mining by introducing a four-level specification that separately addresses uncertainty in log attributes (e.g., activity labels of events, frequencies) and model elements (e.g., service times, read guards). For each type of uncertainty, we illustrate the levels with concrete examples to help understanding and application. We then provide a structured overview of the state of the art in stochastic process mining, classified using our specification, and present key open research challenges.

Large language models have driven major advances in Text-to-SQL generation. However, they suffer from high computational cost, long latency, and data privacy concerns, which make them impractical for many real-world applications. A natural alternative is to use small language models (SLMs), which enable efficient and private on-premise deployment. Yet, SLMs often struggle with weak reasoning and poor instruction following. Conventional reinforcement learning methods based on sparse binary rewards (0/1) provide little learning signal when the generated SQLs are incorrect, leading to unstable or collapsed training. To overcome these issues, we propose FINER-SQL, a scalable and reusable reinforcement learning framework that enhances SLMs through fine-grained execution feedback. Built on group relative policy optimization, FINER-SQL replaces sparse supervision with dense and interpretable rewards that offer continuous feedback even for incorrect SQLs. It introduces two key reward functions: a memory reward, which aligns reasoning with verified traces for semantic stability, and an atomic reward, which measures operation-level overlap to grant partial credit for structurally correct but incomplete SQLs. This approach transforms discrete correctness into continuous learning, enabling stable, critic-free optimization. Experiments on the BIRD and Spider benchmarks show that FINER-SQL achieves up to 67.73% and 85% execution accuracy with a 3B model - matching much larger LLMs while reducing inference latency to 5.57 s/sample. These results highlight a cost-efficient and privacy-preserving path toward high-performance Text-to-SQL generation. Our code is available at https://github.com/thanhdath/finer-sql.

Process mining provides a broad portfolio of methods for extracting insights from event data. Many of these methods can benefit from incorporating semantic relations between activities. A common way to capture such relations is distributional similarity, which links activities based on the contexts in which they occur. Distributional information can be encoded in activity embeddings learned either by prediction-based approaches (adapting neural models from natural language processing) or by count-based models. In prior work, we showed that simple count-based embeddings can match or even outperform neural prediction-based methods for assessing activity similarity in event data. This finding, together with the limited interpretability and high computational cost of neural approaches, motivates the use of count-based alternatives. However, existing distributional similarity methods are limited to certain event data, where each event is assigned a single activity label. In this paper, we argue that many contemporary process mining applications also involve uncertain event data, where events are associated with multiple possible activities and corresponding likelihoods. We generalize count-based approaches and adapt state-of-the-art neural network approaches to the domain of uncertain event data. For both count-based and neural network approaches, probabilistic evidence is incorporated to provide a more robust and comprehensive similarity assessment. Finally, we contribute a comprehensive benchmarking framework for distributional similarity between activities on both certain and uncertain event data, evaluating (i) intrinsic similarity quality, (ii) performance in next-activity prediction as a downstream task, and (iii) computational efficiency. Across 28 certain and 9 uncertain event logs, count-based activity-activity embeddings achieve the highest intrinsic similarity quality overall and allow a simple activity-only predictor to match the state-of-the-art Tax model in next-activity prediction without temporal features, while running more than an order of magnitude faster than neural alternatives on uncertain data. Embedding quality is task-dependent: activity-activity representations lead in intrinsic retrieval while activity-context representations lead in downstream prediction.

The analysis of data series forms the basis of decision-making in various domains, so that it is essential to ensure data validity. Yet, current solutions for sanity checking of processing pipelines, such as GX, TFDV, Pandera or Deequ, fall short in accounting for data quality issues. In particular, irregular cadences, sparsity and value uncertainty limit the applicability of sanity checking and pose risks of false conclusions. In this paper, we present Sound to enable sanity checking of pipelines in the presence of typical quality issues in data series. In particular, Sound evaluates a set of sanity constraints that formalize validity expectations on the data, while incorporating data quality issues, i.e., uncertainty of individual data points and sparsity in a whole data series. To this end, it defines a statistical framework for constraint checking that is based on adaptive resampling and Bayesian hypothesis testing, minimizing computational costs while ensuring accurate results. If a constraint violation has been identified, Sound also includes drill-down strategies to guide users in the identification of the root cause of the violation. We demonstrate the feasibility and utility of Sound by applying it for pipelines developed in the domains of smart grid monitoring and astrophysics.

Modern recommender systems derive predictions from an interaction graph that links users and items. To this end, many of today's state-of-the-art systems use graph neural networks (GNNs) to learn effective representations of these graphs under the assumption of homophily, i.e., the idea that similar users will sit close to each other in the graph. However, recent studies have revealed that real-world recommendation graphs are often heterophilous, i.e., dissimilar users will also often sit close to each other. One of the reasons for this heterophilia is shilling attacks that obscure the inherent characteristics of the graph and make the derived recommendations less accurate as a consequence. Hence, to cope with low homophily in recommender systems, we propose a recommendation model called PGT4Rec that is based on a Partitioned Graph Transformer. The model integrates label information into the learning process, which allows discriminative neighbourhoods of users to be generated. As such, the framework can both detect shilling attacks and predict user ratings for items. Extensive experiments on real and synthetic datasets show PGT4Rec as not only providing superior performance in these two tasks but also significant robustness to a range of adversarial conditions.

Process models may be automatically generated from event logs that contain as-is data of a business process. While such models generalize over the control-flow of specific, recorded process executions, they are often also annotated with behavioural statistics, such as execution frequencies.Based thereon, once a model is published, certain insights about the original process executions may be reconstructed, so that an external party may extract confidential information about the business process. This work is the first to empirically investigate such reconstruction attempts based on process models. To this end, we propose different play-out strategies that reconstruct the control-flow from process trees, potentially exploiting frequency annotations. To assess the potential success of such reconstruction attacks on process models, and hence the risks imposed by publishing them, we compare the reconstructed process executions with those of the original log for several real-world datasets.

Scheduling is adopted in various domains to assign jobs to resources, such that an objective is optimized. While schedules enable the analysis of the underlying system, publishing them also incurs a privacy risk. Recently, privacy attacks on schedules have been proposed, which may reveal sensitive information on the jobs by solving an inverse scheduling problem. In this work, we study the protection against such attacks. We formulate the problem of privacy-and-utility preservation of schedules, which bounds both, the privacy leakage and the loss in the utility of the schedule due to obfuscation. We address the problem based on a set of perturbation functions for schedules, study their instantiations for standard scheduling problems, and implement privacy-and-utility-aware publishing of a schedule using constraint programming. Experiments with synthetic and real-world schedules demonstrate the feasibility, robustness, and effectiveness of our mechanism.

Recommender systems have become an integral part of online services due to their ability to help users locate specific information in a sea of data. However, existing studies show that some recommender systems are vulnerable to poisoning attacks particularly those that involve learning schemes. A poisoning attack is where an adversary injects carefully crafted data into the process of training a model, with the goal of manipulating the system’s final recommendations. Based on recent advancements in artificial intelligence (AI), such attacks have gained importance recently. At present, we do not have a full and clear picture of why adversaries mount such attacks, nor do we have comprehensive knowledge of the full capacity to which such attacks can undermine a model or the impacts that might have. While numerous countermeasures to poisoning attacks have been developed, they have not yet been systematically linked to the properties of the attacks. Consequently, assessing the respective risks and potential success of mitigation strategies is difficult, if not impossible. This survey aims to fill this gap by primarily focusing on poisoning attacks and their countermeasures. This is in contrast to prior surveys that mainly focus on attacks and their detection methods. Through an exhaustive literature review, we provide a novel taxonomy for poisoning attacks, formalise its dimensions, and accordingly organise 31 attacks described in the literature. Further, we review 43 countermeasures to detect and/or prevent poisoning attacks, evaluating their effectiveness against specific types of attacks. This comprehensive survey should serve as a point of reference for protecting recommender systems against poisoning attacks. The article concludes with a discussion on open issues in the field and impactful directions for future research. A rich repository of resources associated with poisoning attacks is available at https://github.com/tamlhp/awesome-recsys-poisoning.

Complex Event Processing (CEP) continuously evaluates queries over event streams to detect patterns that yield actionable insights. In monitoring applications, however, not all patterns are relevant at any point in time. Especially for queries with permissive evaluation semantics yielding a large number of matches, it is often sufficient to compute only aggregated, summary results upon request, e.g., upon refreshing a dashboard. The optimization potential of on-demand aggregations over detected patterns has, so far, been mostly considered for centralized evaluation scenarios. For the setting of distributed event processors, it is not yet clear how to efficiently compute and distribute local aggregates. To fill this gap, we propose DIPSUM, a framework for the efficient on-demand evaluation of CEP aggregate queries in distributed environments. DIPSUM combines the efficient aggregation over computationally expensive operators (Kleene closure, negation), with fine-grained routing of partial aggregates. It relies on a summary data structure that compactly captures match information and can be decomposed for distributed query evaluation. Experiments with real-world and synthetic data show that DIPSUM improves transmission costs, detection latencies, and throughput by several orders of magnitude compared to baseline strategies.

Process mining traditionally assumes centralized event data collection and analysis. However, modern Industrial Internet of Things systems increasingly operate over distributed, resource-constrained edge-cloud infrastructures. This paper proposes a structured approach for decentralizing process mining by enabling event data to be mined directly within the IoT systems edge-cloud continuum. We introduce ContinuumConductor a layered decision framework that guides when to perform process mining tasks such as preprocessing, correlation, and discovery centrally or decentrally. Thus, enabling privacy, responsive and resource-efficient process mining. For each step in the process mining pipeline, we analyze the trade-offs of decentralization versus centralization across these layers and propose decision criteria. We demonstrate ContinuumConductor at a real-world use-case of process optimazition in inland ports. Our contributions lay the foundation for computing-aware process mining in cyber-physical and IIoT systems.