Inductive knowledge graph completion models aim to perform link prediction on knowledge graphs with completely new entities and/or relations. The motivation and key goal of these methods is to transfer relational knowledge and patterns across disjoint knowledge graphs. Although such methods have shown promising empirical results across a variety of datasets, we argue here that these results do not provide sufficient evidence for the successful transfer of relational knowledge. We identify biases in the construction principles of available benchmark datasets, provide a simple baseline that does not rely on multi-step relational paths or more sophisticated relational patterns, and show empirically that this baseline performs on par with the state-of-the-art zero-shot model ULTRA. To provide evidence that relational knowledge transfer really takes place in inductive knowledge graph completion models, improved benchmarks and stronger empirical results are therefore required.

Abstract Multi-task learning (MTL) enables a single neural network to solve multiple tasks simultaneously, offering efficiency and improved generalization potential through shared representations. A central challenge in MTL is balancing task-specific losses during training to avoid performance degradation. While uncertainty-based loss weighting (UW) is a popular and competitive approach, we argue that it suffers from several limitations, including overfitting, rigid homoscedastic assumptions, and a lack of theoretical grounding for various loss functions. Therefore, we propose Soft Optimal Uncertainty Weighting (UW-SO), a novel loss weighting method that builds on UW by deriving analytically optimal weights and applying softmax normalization with adaptable temperature parameter, thereby alleviating several of the shortcomings of UW. Through extensive experiments across diverse datasets and architectures, we show that UW-SO achieves superior and robust performance compared to a variety of existing loss weighting methods. Additionally, we provide insights into the effects of temperature selection and propose measures to reduce computational demand.

Reinforcement learning (RL) has seen significant success across various domains, but its adoption is often limited by the black-box nature of neural network policies, making them difficult to interpret. In contrast, symbolic policies allow representing decision-making strategies in a compact and interpretable way. However, learning symbolic policies directly within on-policy methods remains challenging. In this paper, we introduce SYMPOL, a novel method for SYMbolic tree-based on-POLicy RL. SYMPOL employs a tree-based model integrated with a policy gradient method, enabling the agent to learn and adapt its actions while maintaining a high level of interpretability. We evaluate SYMPOL on a set of benchmark RL tasks, demonstrating its superiority over alternative tree-based RL approaches in terms of performance and interpretability. Unlike existing methods, it enables gradient-based, end-to-end learning of interpretable, axis-aligned decision trees within standard on-policy RL algorithms. Therefore, SYMPOL can become the foundation for a new class of interpretable RL based on decision trees. Our implementation is available under: https://github.com/s-marton/sympol

Urban data visualization poses distinct challenges, especially when it aims to engage the public effectively. This research investigates the impact of the placement of public displays on attracting visitors and facilitating comprehension. We designed two interactive dashboards tailored to visualize local and real-time information, to support questions collected through participatory workshops. These dashboards were then exhibited at a large semi-public event, with two distinct placements being analysed to assess visitor engagement. Our methodology comprised of an in-the-wild study, wherein the two placements were compared to gauge visitor understanding and interaction with the dashboards. The results of this research provide insights into optimizing the placement of public displays for effective data visualization and public engagement.

This research paper delves into the challenges of Supply Chain (SC) planning under demand uncertainty, focussing on Master Production Scheduling (MPS) with capacity constraints.Traditional SC planning methods, often based on point forecasts and basic safety stock calculations, overlook the distinct demand distribution of each product in each planning period.Considering the demand uncertainty is crucial for robust decision making, especially for time series with seasonal and non-stationary demand.To address this gap, the paper introduces a pioneering Separated Estimation and Optimisation (SEO) approach that estimates demand uncertainty using Machine Learning (ML)-based probabilistic forecasting, subsequently solved by stochastic optimisation.Through a comprehensive analysis involving 17 datasets with a total of 303 products, the study confirms the robustness of stochastic optimisation approaches.It also demonstrates the superiority of MLbased forecasting, which is particularly adept at capturing the intricacies of complex demand patterns.The research challenges the conventional reliance on the Gaussian distribution, instead advocating for the adoption of more flexible parametric distributions such as the Negative Binomial (Neg.-Bin.)distribution.Furthermore, it illustrates how to leverage advances in the research areas of ML-based probabilistic demand forecasting and stochastic MPS, as well as providing a basis for future research.Such avenues include the exploration of testing these approaches with a rolling planning horizon and incorporating both demand and supply uncertainties.

Background: With the discussion of fairness in Machine Learning (ML) gaining traction in recent years, the idea of viewing fairness through the causal lens has become prominent. The main idea behind this is that by looking at the causal structure underlying the data used for an ML model, we can see and evaluate more concisely which influences of the sensitive variables on the target variable are problematic and how they are problematic. Doing so allows not only a nuanced view of fairness and an informed choice of fairness measures but also more targeted approaches (such as path-specific bias mitigation) to handle fairness issues. Objectives: Mainly, two important points have hindered the practical use of the causal lens and causality-based bias mitigation. First, a classification of different graphical structures with different fairness implications involving a sensitive variable and a target variable is still missing, as is a discussion of how different contexts can shape our evaluation of fairness. Second, the construction of such graphical models is not trivial and error-prone. However, recent work showed that combining background knowledge and data-driven network structure learning may lead to more accurate graphs. In this work, we attempt to address and tackle these two practical shortcomings. Methods: Our first contribution is a classification and discussion of causal structures with different fairness implications and how contexts shape our assessment. Our second contribution is an advancement in learning more accurate graphs by adapting structure learning algorithms, and a detailed evaluation of graph correctness and subsequent fairness implications. Results: We show that when including background knowledge naturally available in fairness settings, graph learning becomes more accurate, which also has positive implications for accurate fairness assessments. Conclusions: Our work may pave the way for a broader adoption of causal ML fairness by providing concrete suggestions about the implications of causal structures and contexts, and learning more accurate graphs. We also address current limitations and highlight the need for stakeholder inclusion.

Decision Trees (DTs) are commonly used for many machine learning tasks due to their high degree of interpretability. However, learning a DT from data is a difficult optimization problem, as it is non-convex and non-differentiable. Therefore, common approaches learn DTs using a greedy growth algorithm that minimizes the impurity locally at each internal node. Unfortunately, this greedy procedure can lead to inaccurate trees. In this paper, we present a novel approach for learning hard, axis-aligned DTs with gradient descent. The proposed method uses backpropagation with a straight-through operator on a dense DT representation, to jointly optimize all tree parameters. Our approach outperforms existing methods on binary classification benchmarks and achieves competitive results for multi-class tasks. The implementation is available under: https://github.com/s-marton/GradTree

We present a new approach to goal recognition that involves comparing observed facts with their expected probabilities. These probabilities depend on a specified goal g and initial state s0. Our method maps these probabilities and observed facts into a real vector space to compute heuristic values for potential goals. These heuristic values estimate the likelihood of a given goal being the true objective of the observed agent. As obtaining exact expected probabilities for observed facts in an observation sequence is often practically infeasible, we propose and empirically validate a method for approximating these probabilities. Our empirical results show that the proposed approach offers improved goal recognition precision compared to state-of-the-art techniques while reducing computational complexity.

Despite the success of deep learning for text and image data, tree-based ensemble models are still state-of-the-art for machine learning with heterogeneous tabular data. However, there is a significant need for tabular-specific gradient-based methods due to their high flexibility. In this paper, we propose $\text{GRANDE}$, $\text{GRA}$die$\text{N}$t-Based $\text{D}$ecision Tree $\text{E}$nsembles, a novel approach for learning hard, axis-aligned decision tree ensembles using end-to-end gradient descent. GRANDE is based on a dense representation of tree ensembles, which affords to use backpropagation with a straight-through operator to jointly optimize all model parameters. Our method combines axis-aligned splits, which is a useful inductive bias for tabular data, with the flexibility of gradient-based optimization. Furthermore, we introduce an advanced instance-wise weighting that facilitates learning representations for both, simple and complex relations, within a single model. We conducted an extensive evaluation on a predefined benchmark with 19 classification datasets and demonstrate that our method outperforms existing gradient-boosting and deep learning frameworks on most datasets. The method is available under: https://github.com/s-marton/GRANDE

Business process simulation (BPS) is a versatile technique for estimating process performance across various scenarios. Traditionally, BPS approaches employ a control-flow-first perspective by enriching a process model with simulation parameters. Although such approaches can mimic the behavior of centrally orchestrated processes, such as those supported by workflow systems, current control-flow-first approaches cannot faithfully capture the dynamics of real-world processes that involve distinct resource behavior and decentralized decision-making. Recognizing this issue, this paper introduces AgentSimulator, a resource-first BPS approach that discovers a multi-agent system from an event log, modeling distinct resource behaviors and interaction patterns to simulate the underlying process. Our experiments show that AgentSimulator achieves state-of-the-art simulation accuracy with significantly lower computation times than existing approaches while providing high interpretability and adaptability to different types of process-execution scenarios.

Recently, discussions on fairness and algorithmic bias have gained prominence in the learning analytics and educational data mining communities. To quantify algorithmic bias, researchers and practitioners often use popular fairness metrics, e.g., demographic parity, without discussing their choices. This can be considered problematic, as the choices should strongly depend on the underlying data generation mechanism, the potential application, and normative beliefs. Likewise, whether and how one should deal with the indicated bias depends on these aspects. This paper presents and discusses several theoretical cases to highlight precisely this. By providing a set of examples, we hope to facilitate a practice where researchers discuss potential fairness concerns by default.

Multi-relational temporal graphs are powerful tools for modeling real-world data, capturing the evolving and interconnected nature of entities over time. Recently, many novel models are proposed for ML on such graphs intensifying the need for robust evaluation and standardized benchmark datasets. However, the availability of such resources remains scarce and evaluation faces added complexity due to reproducibility issues in experimental protocols. To address these challenges, we introduce Temporal Graph Benchmark 2.0 (TGB 2.0), a novel benchmarking framework tailored for evaluating methods for predicting future links on Temporal Knowledge Graphs and Temporal Heterogeneous Graphs with a focus on large-scale datasets, extending the Temporal Graph Benchmark. TGB 2.0 facilitates comprehensive evaluations by presenting eight novel datasets spanning five domains with up to 53 million edges. TGB 2.0 datasets are significantly largerthan existing datasets in terms of number of nodes, edges, or timestamps. In addition, TGB 2.0 provides a reproducible and realistic evaluation pipeline for multi-relational temporal graphs. Through extensive experimentation, we observe that 1) leveraging edge-type information is crucial to obtain high performance, 2) simple heuristic baselines are often competitive with more complex methods, 3) most methods fail to run on our largest datasets, highlighting the need for research on more scalable methods.

Tabular data is prevalent in real-world machine learning applications, and new models for supervised learning of tabular data are frequently proposed. Comparative studies assessing performance differences typically have model-centered evaluation setups with overly standardized data preprocessing. This limits the external validity of these studies, as in real-world modeling pipelines, models are typically applied after dataset-specific preprocessing and feature engineering. We address this gap by proposing a data-centric evaluation framework. We select 10 relevant datasets from Kaggle competitions and implement expert-level preprocessing pipelines for each dataset. We conduct experiments with different preprocessing pipelines and hyperparameter optimization (HPO) regimes to quantify the impact of model selection, HPO, feature engineering, and test-time adaptation. Our main findings reveal: 1) After dataset-specific feature engineering, model rankings change considerably, performance differences decrease, and the importance of model selection reduces. 2) Recent models, despite their measurable progress, still significantly benefit from manual feature engineering. This holds true for both tree-based models and neural networks. 3) While tabular data is typically considered static, samples are often collected over time, and adapting to distribution shifts can be important even in supposedly static data. These insights suggest that research efforts should be directed toward a data-centric perspective, acknowledging that tabular data requires feature engineering and often exhibits temporal characteristics.

Abstract We consider generating explanations for neural networks in cases where the network’s training data is not accessible, for instance due to privacy or safety issues. Recently, Interpretation Nets ( $$\mathcal {I}$$ I -Nets) have been proposed as a sample-free approach to post-hoc, global model interpretability that does not require access to training data. They formulate interpretation as a machine learning task that maps network representations (parameters) to a representation of an interpretable function. In this paper, we extend the $$\mathcal {I}$$ I -Net framework to the cases of standard and soft decision trees as surrogate models. We propose a suitable decision tree representation and design of the corresponding $$\mathcal {I}$$ I -Net output layers. Furthermore, we make $$\mathcal {I}$$ I -Nets applicable to real-world tasks by considering more realistic distributions when generating the $$\mathcal {I}$$ I -Net’s training data. We empirically evaluate our approach against traditional global, post-hoc interpretability approaches and show that it achieves superior results when the training data is not accessible.

Abstract Knowledge graph completion is the task of predicting correct facts that can be expressed by the vocabulary of a given knowledge graph, which are not explicitly stated in that graph. Broadly, there are two main approaches for solving the knowledge graph completion problem. Sub-symbolic approaches embed the nodes and/or edges of a given graph into a low-dimensional vector space and use a scoring function to determine the plausibility of a given fact. Symbolic approaches learn a model that remains within the primary representation of the given knowledge graph. Rule-based approaches are well-known examples. One such approach is AnyBURL. It works by sampling random paths, which are generalized into Horn rules. Previously published results show that the prediction quality of AnyBURL is close to current state of the art with the additional benefit of offering an explanation for a predicted fact. In this paper, we propose several improvements and extensions of AnyBURL. In particular, we focus on AnyBURL’s capability to be successfully applied to large and very large datasets. Overall, we propose four separate extensions: (i) We add to each rule a set of pairwise inequality constraints which enforces that different variables cannot be grounded by the same entities, which results into more appropriate confidence estimations. (ii) We introduce reinforcement learning to guide path sampling in order to use available computational resources more efficiently. (iii) We propose an efficient sampling strategy to approximate the confidence of a rule instead of computing its exact value. (iv) We develop a new multithreaded AnyBURL, which incorporates all previously mentioned modifications. In an experimental study, we show that our approach outperforms both symbolic and sub-symbolic approaches in large-scale knowledge graph completion. It has a higher prediction quality and requires significantly less time and computational resources.

In this paper, we propose a new online learning algorithm tailored for data streams described by varying feature spaces (VFS), wherein new features constantly emerge and old features may stop to be observed over various time spans. Our proposed algorithm, named Online Random Feature Forests for Feature space Variabilities (ORF3V), provides a strategy to respect such feature dynamics by generating, updating, pruning, as well as online re-weighing an ensemble of what we call feature forests, which are generated and updated based on a compressed and storage efficient representation for each observed feature. We benchmark our algorithm on 12 datasets, including one novel real-world dataset of government COVID-19 responses collected through a crowd-sensing program in Spain. The empirical results substantiate the viability and effectiveness of our ORF3V algorithm and its superior accuracy performance over the state-of-the-art rival models.

Rule-based approaches have proven to be an efficient and explainable method for knowledge base completion. Their predictive quality is on par with classic knowledge graph embedding models such as TransE or ComplEx, however, they cannot achieve the results of neural models proposed recently. The performance of a rule-based approach depends crucially on the solution of the rule aggregation problem, which is concerned with the computation of a score for a prediction that is generated by several rules. Within this paper, we propose a supervised approach to learn a reweighted confidence value for each rule to get an optimal explanation for the training set given a specific aggregation function. In particular, we apply our approach to two aggregation functions: We learn weights for a noisy-or multiplication and apply logistic regression, which computes the score of a prediction as a sum of these weights. Due to the simplicity of both models the final score is fully explainable. Our experimental results show that we can significantly improve the predictive quality of a rule-based approach. We compare our method with current state-of-the-art latent models that lack explainability, and achieve promising results.

How much does a CEO’s personality impact the performanceof their company? Management theory posits a great influence, but it is difficult to show empirically—there is a lack of publicly available self-reported personality data of top managers. Instead, we propose a text-based personality regressor based on crowd-sourced Myers–Briggs Type Indicator (MBTI) assessments. The ratings have a high internal and external validity and can be predicted with moderate to strong correlations for three out of four dimensions. Providing evidence for the upper echelons theory, we demonstrate that the predicted CEO personalities have explanatory power of financial risk.

Human intelligence comes from the capability to describe and make sense of the world surrounding us, often in a lifelong manner. Online Learning (OL) allows a model to simulate this capability, which involves processing data in sequence, making predictions, and learning from predictive errors. However, traditional OL assumes a fixed set of features to describe data, which can be restrictive. In reality, new features may emerge and old features may vanish or become obsolete, leading to an open feature space. This dynamism can be caused by more advanced or outdated technology for sensing the world, or it can be a natural process of evolution. This paper reviews recent breakthroughs that strived to enable OL in open feature spaces, referred to as Utilitarian Online Learning (UOL). We taxonomize existing UOL models into three categories, analyze their pros and cons, and discuss their application scenarios. We also benchmark the performance of representative UOL models, highlighting open problems, challenges, and potential future directions of this emerging topic.

Process models play an important role in various software engineering contexts. Among others, they are used to capture business-related requirements and provide the basis for the development of process-oriented applications in low-code/no-code settings. To support modelers in creating, checking, and maintaining process models, dedicated tools are available. While these tools are generally considered as indispensable to capture process models for their later use, the initial version of a process model is often sketched on a whiteboard or a piece of paper. This has been found to have great advantages, especially with respect to communication and collaboration. It, however, also creates the need to subsequently transform the model sketch into a digital counterpart that can be further processed by modeling and analysis tools. Therefore, to automate this task, various so-called sketch recognition approaches have been defined in the past. Yet, these existing approaches are too limited for use in practice, since they, for instance, require sketches to be created on a digital device or do not address the recognition of edges or textual labels. Against this background, we use this paper to introduce Sketch2Process, the first end-to-end sketch recognition approach for process models captured using BPMN. Sketch2Process uses a neural network-based architecture to recognize the shapes, edges, and textual labels of highly expressive process models, covering 25 types of BPMN elements. To train and evaluate our approach, we created a dataset consisting of 704 hand-drawn and manually annotated BPMN models. Our experiments demonstrate that our approach is highly accurate and consistently outperforms the state of the art.

We propose a novel black-box approach for performing adversarial attacks against knowledge graph embedding models. An adversarial attack is a small perturbation of the data at training time to cause model failure at test time. We make use of an efficient rule learning approach and use abductive reasoning to identify triples which are logical explanations for a particular prediction. The proposed attack is then based on the simple idea to suppress or modify one of the triples in the most confident explanation. Although our attack scheme is model independent and only needs access to the training data, we report results on par with state-of-the-art white-box attack methods that additionally require full access to the model architecture, the learned embeddings, and the loss functions. This is a surprising result which indicates that knowledge graph embedding models can partly be explained post hoc with the help of symbolic methods.

Sum-Product Networks (SPNs) are expressive probabilistic models that provide exact, tractable inference. They achieve this efficiency by making use of local independence. On the other hand, mixtures of exchangeable variable models (MEVMs) are a class of tractable probabilistic models that make use of exchangeability of discrete random variables to render inference tractable. Exchangeability, which arises naturally in relational domains, has not been considered for efficient representation and inference in SPNs yet. The contribution of this paper is a novel probabilistic model which we call Exchangeability-Aware Sum-Product Networks (XSPNs). It contains both SPNs and MEVMs as special cases, and combines the ability of SPNs to efficiently learn deep probabilistic models with the ability of MEVMs to efficiently handle exchangeable random variables. We introduce a structure learning algorithm for XSPNs and empirically show that they can be more accurate than conventional SPNs when the data contains repeated, interchangeable parts.

In this paper, we introduce the task of political coalition signal prediction from text, that is, the task of recognizing from the news coverage leading up to an election the (un)willingness of political parties to form a government coalition.We decompose our problem into two related, but distinct tasks: (i) predicting whether a reported statement from a politician or a journalist refers to a potential coalition and (ii) predicting the polarity of the signal -namely, whether the speaker is in favour of or against the coalition.For this, we explore the benefits of multi-task learning and investigate which setup and task formulation is best suited for each sub-task.We evaluate our approach, based on hand-coded newspaper articles, covering elections in three countries (Ireland, Germany, Austria) and two languages (English, German).Our results show that the multi-task learning approach can further improve results over a strong monolingual transfer learning baseline.* election * AND ( * coalition * OR * pact * OR * collaboration * OR * alliance * ) 1The keyword-based sampling was necessary so that the coders did not spend too much time on articles that did not contain any signals.The keywords were optimised for recall by domain experts who,

Abstract We address the problem of offline handwritten diagram recognition. Recently, it has been shown that diagram symbols can be directly recognized with deep learning object detectors. However, object detectors are not able to recognize the diagram structure. We propose Arrow R-CNN, the first deep learning system for joint symbol and structure recognition in handwritten diagrams. Arrow R-CNN extends the Faster R-CNN object detector with an arrow head and tail keypoint predictor and a diagram-aware postprocessing method. We propose a network architecture and data augmentation methods targeted at small diagram datasets. Our diagram-aware postprocessing method addresses the insufficiencies of standard Faster R-CNN postprocessing. It reconstructs a diagram from a set of symbol detections and arrow keypoints. Arrow R-CNN improves state-of-the-art substantially: on a scanned flowchart dataset, we increase the rate of recognized diagrams from 37.7 to 78.6%.

Smart charging assigns charging capacities between vehicles in limited charging infrastructures. Smart charging solutions are becoming widespread with increasing numbers of commercial offerings. However, commercial solutions are intransparent regarding algorithm implementation. In contrast, open source solutions are transparent and open for collaborative development and scientific research. This article presents an open source package with a smart charging algorithm. The algorithm schedules heterogeneous fleets of vehicles for charging while ensuring a fair share for each vehicle. We present implementation aspects of the smart charging algorithm including data structures and representational state transfer application programming interfaces. Additionally, the smart charging algorithm was validated in a one-year field test. Experimental results of the field test show elcetronic vehicles (EVs) at six charging stations can be scheduled for charging when the grid connection only allows two EVs to charge concurrently. Runtime measurements demonstrate the smart charging algorithm is applicable in real-time and scales to large fleet sizes.

Electric vehicles (EVs) are increasingly used for commuting to the workplace where employees expect charging opportunities. Limited power supply in existing infrastructures prevents charging many EVs concurrently. Smart charging balances scarce charging resources and distributes power by prioritizing EVs. We maximize fair share among EVs by prioritizing for equal chances of reaching a sufficient state of charge by the time of departure. To address uncertain EV availability, we use regression models trained on historical data to predict departures. More sophisticated regression models show higher prediction accuracy. We improve a smart charging heuristic by incorporating these predictions. Simulations show accurate predictions improve EV prioritization and thus fair share.

Recognition of activities of daily living (ADLs) is an enabling technology for several ubiquitous computing applications. Most activity recognition systems rely on supervised learning to extract activity models from labeled datasets. A problem with that approach is the acquisition of comprehensive activity datasets, which is an expensive task. The problem is particularly challenging when focusing on complex ADLs characterized by large variability of execution. Moreover, several activity recognition systems are limited to offline recognition, while many applications claim for online activity recognition. In this paper, we propose POLARIS, a framework for unsupervised activity recognition. POLARIS can recognize complex ADLs exploiting the semantics of activities, context data, and sensors. Through ontological reasoning, our algorithm derives semantic correlations among activities and sensor events. By matching observed events with semantic correlations, a statistical reasoner formulates initial hypotheses about the occurred activities. Those hypotheses are refined through probabilistic reasoning, exploiting semantic constraints derived from the ontology. Our system supports online recognition, thanks to a novel segmentation algorithm. Extensive experiments with real-world datasets show that the accuracy of our unsupervised method is comparable to the one of supervised approaches. Moreover, the online version of our system achieves essentially the same accuracy of the offline version.

We study the problem of explaining differences between two snapshots of the same database table including record insertions, deletions and in particular record updates. Unlike existing alternatives, our solution induces transformation functions and does not require knowledge of the correct alignment between the record sets. This allows profiling snapshots of tables with unspecified or modified primary keys. In such a problem setting, there are always multiple explanations for the differences. Our goal is to find the simplest explanation. We propose to measure the complexity of explanations on the basis of minimum description length in order to formulate the task as an optimization problem. We show that the problem is NP-hard and propose a heuristic search algorithm to solve practical problem instances. We implement a prototype called Affidavit to assess the explanatory qualities of our approach in experiments based on different real-world data sets. We show that it can scale to both a large number of records and attributes and is able to reliably provide correct explanations under practical levels of modifications.

Social media platforms have become an essential venue for online deliberation where users discuss arguments, debate, and form opinions.In this paper, we propose an unsupervised method to detect the stance of argumentative claims with respect to a topic.Most related work focuses on topic-specific supervised models that need to be trained for every emergent debate topic.To address this limitation, we propose a topic independent approach that focuses on a frequently encountered class of arguments, specifically, on arguments from consequences.We do this by extracting the effects that claims refer to, and proposing a means for inferring if the effect is a good or bad consequence.Our experiments provide promising results that are comparable to, and in particular regards even outperform BERT.Furthermore, we publish a novel dataset of arguments relating to consequences, annotated with Amazon Mechanical Turk.

We propose an unsupervised approach for assessing conceptual complexity of texts, based on spreading activation.Using DBpedia knowledge graph as a proxy to long-term memory, mentioned concepts become activated and trigger further activation as the text is sequentially traversed.Drawing inspiration from psycholinguistic theories of reading comprehension, we model memory processes such as semantic priming, sentence wrap-up, and forgetting.We show that our models capture various aspects of conceptual text complexity and significantly outperform current state of the art.

Several probabilistic extensions of description logic languages have been proposed and thoroughly studied. However, their practical use has been hampered by intractability of various reasoning tasks. While present-day knowledge bases (KBs) contain millions of instances and thousands of axioms, most state-of-the-art reasoners are capable of handling small scale KBs with thousands of instances. Thus, recent research has focused on leveraging the structure of KBs and queries in order to speed up inference runtime. However, these efforts have not been satisfactory in providing reasoners that are suitable for practical use in large scale KBs. In this study, we aim to tackle this challenging problem. In doing so, we use a probabilistic extension of OWL RL (called PRORL) as a modeling language and exploit graph neighborhoods (of undirected graphical models) for efficient approximate probabilistic inference. We show that subgraph extraction based inference is much faster and has comparable accuracy to full graph inference. We perform several experiments, in order to support our claim, over a NELL KB containing millions of instances and thousands of axioms. Furthermore, we propose a novel graph-based algorithm to automatically partition inferences rules based on their structure for efficient parallel inference.

We propose an anytime bottom-up technique for learning logical rules from large knowledge graphs. We apply the learned rules to predict candidates in the context of knowledge graph completion. Our approach outperforms other rule-based approaches and it is competitive with current state of the art, which is based on latent representations. Besides, our approach is significantly faster, requires less computational resources, and yields an explanation in terms of the rules that propose a candidate.

Financial risk, defined as the chance to deviate from return expectations, is most commonly measured with volatility. Due to its value for investment decision making, volatility prediction is probably among the most important tasks in finance and risk management. Although evidence exists that enriching purely financial models with natural language information can improve predictions of volatility, this task is still comparably underexplored. We introduce PRoFET, the first neural model for volatility prediction jointly exploiting both semantic language representations and a comprehensive set of financial features. As language data, we use transcripts from quarterly recurring events, so-called "earnings calls"; in these calls, the performance of publicly traded companies is summarized and prognosticated by their management. We show that our proposed architecture, which models verbal context with an attention mechanism, significantly outperforms the previous state-of-the-art and other strong baselines. Finally, we visualize this attention mechanism on the token-level, thus aiding interpretability and providing a use case of PRoFET as a tool for investment decision support.

The emergence of open information extraction as a tool for constructing and expanding knowledge graphs has aided the growth of temporal data, for instance, YAGO, NELL and Wikidata. While YAGO and Wikidata maintain the valid time of facts, NELL records the time point at which a fact is retrieved from some Web corpora. Collectively, these knowledge graphs (KG) store facts extracted from Wikipedia and other sources. Due to the imprecise nature of the extraction tools that are used to build and expand KG, such as NELL, the facts in the KG are weighted (a confidence value representing the correctness of a fact). Additionally, NELL can be considered as a transaction time KG because every fact is associated with extraction date. On the other hand, YAGO and Wikidata use the valid time model because they maintain facts together with their validity time (temporal scope). In this paper, we propose a bitemporal model (that combines transaction and valid time models) for maintaining and querying bitemporal probabilistic knowledge graphs. We study coalescing and scalability of marginal and MAP inference. Moreover, we show that complexity of reasoning tasks in atemporal probabilistic KG carry over to the bitemporal setting. Finally, we report our evaluation results of the proposed model.

There is an ever increasing number of rule learning algorithms and tools for automatic knowledge base (KB) construction. These tools often produce weighted rules and facts that make up a probabilistic KB (PKB). In such a PKB, probabilistic inference is used in order to perform marginal inference, consistency checking and other tasks. However, in general, inference is known to be intractable. Hence, recently, there are a number of studies aimed at lifting (making tractable or approximating) inference by exploiting symmetries in the structure of a PKB. These studies alleviate grounding entirely a given PKB which can generate a sizable factor graph for inference (e.g. to compute the probability of a query). In line with this, we propose a novel technique to automatically partition rules based on their structure for efficient parallel grounding. In addition, we perform query expansion so as to generate a factor graph small enough to be used for efficient probability computation. We present a novel approximate marginal inference algorithm that uses N-hop subgraph extraction and query expansion. Moreover, we show that our system is much faster than state-of-the-art systems.

Due to the emerging popularity of pervasive healthcare applications, tools for monitoring activities in smart homes are gaining momentum. Existing methods mainly rely on supervised learning algorithms for recognizing activities based on sensor data. A key issue with those approaches is the acquisition of comprehensive training sets of activities. Indeed, that task incurs significant costs in terms of manual labeling effort; moreover, labeling by external observers violates the individual's privacy. For these reasons, there is an increasing interest in unsupervised activity recognition methods. A popular approach relies on knowledge-based models expressed by ontologies of activities, environment and sensors. Unfortunately, those models require significant knowledge engineering efforts, and are often limited to a specific application. In this paper, we address the issues of existing methods by proposing a novel hybrid approach. Our intuition is that a generic knowledge-based model of activities can be refined to target specific individuals and environments by collaboratively acquiring feedback from inhabitants. Specifically, we propose a collaborative active learning method to refine correlations among sensor events and activity types that are initially extracted from a high-level ontology. Generic correlations are personalized to each target smart-home considering the similarity between the feedback target and the feedback provider in terms of environment and inhabitant's profiles. Moreover, thanks to this method, new sensors installed in the home are seamlessly integrated in the recognition framework. In order to reduce the burden of providing feedback, we also propose a technique to carefully select the conditions that trigger a feedback request. We conducted experiments with a real-world dataset and a generic ontology of activities. Results show that our hybrid method outperforms state-of-the-art supervised and unsupervised activity recognition techniques while triggering an acceptable number of feedback queries.

Ambient Assisted Living using mobile device sensors is an active area of research in pervasive computing. Multiple approaches have shown that wearable sensors perform very well and distinguish falls reliably from Activities of Daily Living. However, these systems are tested in a controlled environment and are optimized for a given set of sensor types, sensor positions, and subjects. In this work, we propose a self-adaptive pervasive fall detection approach that is robust to the heterogeneity of real life situations. Therefore, we combine sensor data of four publicly available datasets, covering about 100 subjects, 5 devices, and 3 sensor placements. In a comprehensive evaluation, we show that our system is not only robust regarding the different dimensions of heterogeneity, but also adapts autonomously to spontaneous changes in the sensor's position at runtime.

The management of uncertainty is crucial when harvesting structured content from unstructured and noisy sources. Knowledge Graphs ( KGs ) are a prominent example. KGs maintain both numerical and non-numerical facts, with the support of an underlying schema. These facts are usually accompanied by a confidence score that witnesses how likely is for them to hold. Despite their popularity, most of existing KGs focus on static data thus impeding the availabilityof timewise knowledge. What is missing is a comprehensive solution for the management of uncertain and temporal data in KGs . The goal of this paper is to fill this gap. We rely on two main ingredients. The first is a numerical extension of Markov Logic Networks (MLNs) that provide the necessary underpinning to formalize the syntax and semantics of uncertain temporal KGs . The second is a set of Datalog constraints with inequalities that extend the underlying schema of the KGs and help to detect inconsistencies. From a theoretical point of view, we discuss the complexity of two important classes of queries for uncertain temporal KGs: maximuma-posteriori and conditional probability inference. Due to the hardness of these problems and the fact that MLN solvers do not scale well, we also explore the usage of Probabilistic Soft Logics (PSL) as a practical tool to support our reasoning tasks. We report on an experimental evaluation comparing the MLN and PSL approaches.

Lexically and syntactically simpler sentences result in shorter reading time and better understanding in many people. However, no reliable systems for automatic assessment of absolute sentence complexity have been proposed so far. Instead, the assessment is usually done manually, requiring expert human annotators. To address this problem, we first define the sentence complexity assessment as a five-level classification task, and build a ‘gold standard’ dataset. Next, we propose robust systems for sentence complexity assessment, using a novel set of features based on leveraging lexical properties of freely available corpora, and investigate the impact of the feature type and corpus size on the classification performance.

Human activity recognition using wearable devices is an active area of research in pervasive computing. In our work, we address the problem of reducing the effort for training and adapting activity recognition approaches to a specific person. We focus on the problem of cross-subjects based recognition models and introduce an approach that considers physical characteristics. Further, to adapt such a model to the behavior of a new user, we present a personalization approach that relies on online and active machine learning. In this context, we use online random forest as a classifier to continuously adapt the model without keeping the already seen data available and an active learning approach that uses user-feedback for adapting the model while minimizing the effort for the new user. We test our approaches on a real world data set that covers 15 participants, 8 common activities, and 7 different on-body device positions. We show that our cross-subjects based approach performs constantly +3% better than the standard approach. Further, the personalized cross-subjects models, gained through user-feedback, recognize dynamic activities with an F-measure of 87% where the user has significantly less effort than collecting and labeling data.

The management of uncertainty is crucial when harvesting structured content from unstructured and noisy sources. Knowledge Graphs ( kg s), maintaining both numerical and non-numerical facts supported by an underlying schema, are a prominent example. Knowledge Graph management is challenging because: (i) most of existing kg s focus on static data, thus impeding the availability of timewise knowledge; (ii) facts in kg s are usually accompanied by a confidence score, which witnesses how likely it is for them to hold. We demonstrate T e C o R e , a system for temporal inference and conflict resolution in uncertain temporal knowledge graphs ( utkg s). At the heart of T e C o R e are two state-of-the-art probabilistic reasoners that are able to deal with temporal constraints efficiently. While one is scalable, the other can cope with more expressive constraints. The demonstration will focus on enabling users and applications to find inconsistencies in utkg s. T e C o R e provides an interface allowing to select utkg s and editing constraints; shows the maximal consistent subset of the utkg , and displays statistics (e.g., number of noisy facts removed) about the debugging process.

Markov logic networks (MLNs) have proven to be useful tools for reasoning about uncertainty in complex knowledge bases. In this paper, we extend MLNs with numerical constraints and present an efficient implementation in terms of a cutting plane method. This extension is useful for reasoning over uncertain temporal data. To show the applicability of this extension, we enrich log-linear description logics (DLs) with concrete domains (datatypes). Thereby, allowing to reason over weighted DLs with datatypes. Moreover, we use the resulting formalism to reason about temporal assertions in DB-pedia, thus illustrating its practical use.

Information explosion leads to continuous growth of data distributed over different data sources. However, the increasing number of data sources increases the risk of inconsistency. In such a federative setting, description logics can be applied to define a central schema that serves as a conceptual view comprising and extending the semantics of each data source. Consequently, each data source is treated as a single knowledge base that is integrated in a federated knowledge base. Following this idea, we propose an approach for automated debugging of federated knowledge bases that targets the identification and repair of inconsistency. We report on experiments with a large distributed dataset from the domain of library science.

IT infrastructure is a crucial part in most of today''s business operations. High availability and reliability, and short response times to outages are essential. Thus a high amount of tool support and automation in risk management is desirable to decrease outages. We propose a new approach for calculating the root cause for an observed failure in an IT infrastructure. Our approach is based on Abduction in Markov Logic Networks. Abduction aims to find an explanation for a given observation in the light of some background knowledge. In failure diagnosis, the explanation corresponds to the root cause, the observation to the failure of a component, and the background knowledge to the dependency graph extended by potential risks. We apply a method to extend a Markov Logic Network in order to conduct abductive reasoning, which is not naturally supported in this formalism. Our approach exhibits a high amount of reusability and enables users without specific knowledge of a concrete infrastructure to gain viable insights in the case of an incident. We implemented the method in a tool and illustrate its suitability for root cause analysis by applying it to a sample scenario.

Recognition of activities of daily living (ADLs) is an enabling technology for several ubiquitous computing applications. In this field, most activity recognition systems rely on supervised learning methods to extract activity models from labeled datasets. An inherent problem of that approach consists in the acquisition of comprehensive activity datasets, which is expensive and may violate individuals' privacy. The problem is particularly challenging when focusing on complex ADLs, which are characterized by large intra- and inter-personal variability of execution. In this paper, we propose an unsupervised method to recognize complex ADLs exploiting the semantics of activities, context data, and sensing devices. Through ontological reasoning, we derive semantic correlations among activities and sensor events. By matching observed sensor events with semantic correlations, a statistical reasoner formulates initial hypotheses about the occurred activities. Those hypotheses are refined through probabilistic reasoning, exploiting semantic constraints derived from the ontology. Extensive experiments with real-world datasets show that the accuracy of our unsupervised method is comparable to the one of state of the art supervised approaches.

Human activity recognition using mobile device sensors is an active area of research in pervasive computing. In our work, we aim at implementing activity recognition approaches that are suitable for real life situations. This paper focuses on the problem of recognizing the on-body position of the mobile device which in a real world setting is not known a priori. We present a new real world data set that has been collected from 15 participants for 8 common activities were they carried 7 wearable devices in different positions. Further, we introduce a device localization method that uses random forest classifiers to predict the device position based on acceleration data. We perform the most complete experiment in on-body device location that includes all relevant device positions for the recognition of a variety of different activities. We show that the method outperforms other approaches achieving an F-Measure of 89% across different positions. We also show that the detection of the device position consistently improves the result of activity recognition for common activities.

This paper introduces the Protégé plugin uDecide. With the help of uDecide it is possible to solve multi-attribute decision making problems encoded in a straight forward extension of standard Description Logics. The formalism allows to specify background knowledge in terms of an ontology, while each attribute is represented as a weighted class expression. On top of such an approach one can compute the best choice (or the best k-choices) taking background knowledge into account in the appropriate way. We show how to implement the approach on top of existing semantic web technologies and demonstrate its benefits with the help of an interesting use case that illustrates how to convert an existing web resource into an expert system with the help of uDecide.

We propose an approach for semantifying web extracted facts. In particular, we map subject and object terms of these facts to instances; and relational phrases to object properties defined in a target knowledge base. By doing this we resolve the ambiguity inherent in the web extracted facts, while simultaneously enriching the target knowledge base with a significant number of new assertions. In this paper, we focus on the mapping of the relational phrases in the context of the overall work ow. Furthermore, in an open extraction setting identical semantic relationships can be represented by different surface forms, making it necessary to group these surface forms together. To solve this problem we propose the use of markov clustering. In this work we present a complete, ontology independent, generalized workflow which we evaluate on facts extracted by Nell and Reverb. Our target knowledge base is DBpedia. Our evaluation shows promising results in terms of producing highly precise facts. Moreover, the results indicate that the clustering of relational phrases pays of in terms of an improved instance and property mapping.

RockIt is a maximum a-posteriori (MAP) query engine for statistical relational models. MAP inference in graphical models is an optimization problem which can be compiled to integer linear programs (ILPs).We describe several advances in translating MAP queries to ILP instances and present the novel meta-algorithm cutting plane aggregation (CPA). CPA exploits local context-specific symmetries and bundles up sets of linear constraints. The resulting counting constraints lead to more compact ILPs and make the symmetry of the ground model more explicit to state-of-the-art ILP solvers. Moreover, RockIt parallelizes most parts of the MAP inference pipeline taking advantage of ubiquitous shared-memory multi-core architectures. We report on extensive experiments with Markov logic network (MLN) benchmarks showing that RockIt outperforms the state-of-the-art systems Alchemy, Markov TheBeast, and Tuffy both in terms of efficiency and quality of results.

A major challenge of ubiquitous computing resides in the acquisition and modelling of rich and heterogeneous context data, among which, ongoing human activities at different degrees of granularity. In a previous work, we advocated the use of probabilistic description logics (DLs) in a multilevel activity recognition framework. In this paper, we present an in-depth study of activity modeling and reasoning within that framework, as well as an experimental evaluation with a large real-world dataset. Our solution allows us to cope with the uncertain nature of ontological descriptions of activities, while exploiting the expressive power and inference tools of the OWL 2 language. Targeting a large dataset of real human activities, we developed a probabilistic ontology modeling nearly 150 activities and actions of daily living. Experiments with a prototype implementation of our framework confirm the viability of our solution.

A majority of the approaches to activity recognition in sensor environments are either based on manually constructed rules for recognizing activities or lack the ability to incorporate complex temporal dependencies. Furthermore, in many cases, the rather unrealistic assumption is made that the subject carries out only one activity at a time. In this paper, we describe the use of Markov logic as a declarative framework for recognizing interleaved and concurrent activities incorporating both input from pervasive light-weight sensor technology and common-sense background knowledge. In particular, we assess its ability to learn statistical-temporal models from training data and to combine these models with background knowledge to improve the overall recognition accuracy. To this end, we propose two Markov logic formulations for inferring the foreground activity as well as each activities' start and end times. We evaluate the approach on an established dataset. where it outperforms state-of-the-art algorithms for activity recognition.

Ontology matching is the problem of determining correspondences between concepts, properties, and individuals of different heterogeneous ontologies. With this paper we present a novel probabilistic-logical framework for ontology matching based on Markov logic. We define the syntax and semantics and provide a formalization of the ontology matching problem within the framework. The approach has several advantages over existing methods such as ease of experimentation, incoherence mitigation during the alignment process, and the incorporation of a-priori confidence values. We show empirically that the approach is efficient and more accurate than existing matchers on an established ontology alignment benchmark dataset.

The systematic evaluation of ontology alignments still faces a number of problems. One is the argued inadequacy of traditional quality measures adopted from the field of information retrieval. In previous work, Euzenat and others have proposed notions of semantic precision and recall that are supposed to better reflect the true quality of an alignment by considering its deductive closure rather than the explicitly stated correspondences. So far, these measures have only been investigated in theory. In this paper, we present the first implementation of a restricted version of semantic precision and recall as well as experiments in using it, we conducted on the results of the 2008 OAEI campaign.

Finding correct semantic correspondences between heterogeneous ontologies is one of the most challenging problems in the area of semantic web technologies. As manually constructing such mappings is not feasible in realistic scenarios, a number of automatic matching tools have been developed that propose mappings based on general heuristics. As these heuristics often produce incorrect results, a manual revision is inevitable in order to guarantee the quality of generated mappings. Experiences with benchmarking matching systems revealed that the manual revision of mappings is still a very difficult problem because it has to take the semantics of the ontologies as well as interactions between mappings into account. In this article, we propose methods for supporting human experts in the task of revising automatically created mappings. In particular, we present non-standard reasoning methods for detecting and propagating implications of expert decisions on the correctness of a mapping.

The benefits of modular ontologies in terms of easier creation and maintenance as well as better computational properties have been recognized by different researchers. As most real world ontologies, however, are still designed in a monolithic way, there is a need for methods that partition an existing ontology into a set of modules. Currently, existing work suffers from the fact that the notion of modularization is not as well understood in the context of ontologies as it is in software engineering. In this paper we present a flexible partitioning tool for large ontologies that can be adapted to the needs of different applications based on criteria that the resulting modular ontology should satisfy.

The use of description logics as one of the primary logical languages for knowledge representation on the Web has created new challenges with respect to reasoning in these logics. In order to support the vision of a semantic Web of interrelated ontologies, reasoning procedures have to be highly scalable and able to deal with physically distributed knowledge models. A natural way of addressing these problems is to rely on distributed inference procedures that can distribute the load between different solvers, thus reducing potential bottlenecks both in terms of memory and computation time. In this paper, we propose a distributed resolution approach that solves the problem by local resolution and propagation of derived axioms between different reasoners. The method is complete for first order logic, terminates for ALC ontologies and avoids duplication of axioms and inferences. The work can be seen as a building block for a large scale distributed reasoning infrastructure for the semantic Web as envisioned in recent activities such as the large knowledge collider (LarKC) project.

The use of thesaurus-based indexing is a common approach for increasing the performance of document retrieval. With the growing amount of documents available, manual indexing is not a feasible option. Statistical methods for automated document indexing are an attractive alternative. We argue that the quality of the thesaurus used as a basis for indexing in regard to its ability to adequately cover the contents to be indexed is of crucial importance inautomatic indexing because there is no human in the loop that can spot and avoid indexing errors. We propose a method for thesaurus evaluation that is based on a combination of statistical measures and appropriate visualization techniques that supports the detection of potential problems in a thesaurus. We describe this method and show its application in the context of two automatic indexing tasks. The examples show that the methods indeed eases the detection and correction of errors leading to a better indexing result. Please refer to http://www.kaiec.org for high resolution media of all figures used in this paper, as well as an animated presentation of the interactive tool.

No abstract available.

It is widely agreed that the semantic Web will be build on RDF. Therefore the success of the semantic Web also depends on the availability of a scalable and reliable infrastructure for storing and accessing RDF data. A number of storage and retrieval systems have been developed recently, but despite the inherently distributed nature of the semantic Web most of these systems do not support distributed storage and retrieval of data. In this paper we report our experiences with implementing a distributed storage and query infrastructure on top of an existing RDF infrastructure. We present the system architecture and discuss performance issues based on a set of experiments with the infrastructure. We conclude with an identification of remaining performance bottlenecks and point to further improvements.

A technical infrastructure for storing, querying and managing RDFdata is a key element in the current semantic web development. Systems like Jena, Sesame or the ICS-FORTH RDF Suite are widelyused for building semantic web applications. Currently, none ofthese systems supports the integrated querying of distributed RDF repositories. We consider this a major shortcoming since the semanticweb is distributed by nature. In this paper we present an architecture for querying distributed RDF repositories by extending the existing Sesame system. We discuss the implications of our architectureand propose an index structure as well as algorithms forquery processing and optimization in such a distributed context.

Web page categorization is an approach for improving precision and efficiency of information retrieval on the web by filtering out irrelevant pages. Current approaches to information filtering based on categorization assume the existence of a single classification hierarchy used for filtering. In this paper, we address the problem of filtering information categorized according to different classification hierarchies. We describe a method for approximating Boolean queries over class names across different class hierarchies.

Content-related metadata plays an important role in intelligent information systems. Especially on the world-wide web meaningful metadata describing the contents of a web-site is the key to intelligent retrieval and access of information. Metadata description standards like RDF and RDF schema have been developed and work in progress addresses the use of ontologies to provide a logical foundation for metadata. However, the acquisition of appropriate metadata is still a problem. The main part of the paper is concerned with the specification of ontologies and metadata models. We describe the Spectacle approach, a knowledge-based approach for metadata validation and generation as well as tools related to the ontology language OIL. We conclude that the specification of ontologies and the generation of metadata models are processes that supplement each other and propose a method for semi-automatic generation of metadata models on the basis of ontologies.