arXiv subjects

Martin Theobald

Publications and source records attributed to Martin Theobald.

16 recordsLinked to original sources

arXiv2025

Automating Historical Insight Extraction from Large-Scale Newspaper Archives via Neural Topic Modeling

Extracting coherent and human-understandable themes from large collections of unstructured historical newspaper archives presents significant challenges due to topic evolution, Optical Character Recognition (OCR) noise, and the sheer volume of text. Traditional topic-modeling methods, such as Latent Dirichlet Allocation (LDA), often fall short in capturing the complexity and dynamic nature of discourse in historical texts. To address these limitations, we employ BERTopic. This neural topic-modeling approach leverages transformerbased embeddings to extract and classify topics, which, despite its growing popularity, still remains underused in historical research. Our study focuses on articles published between 1955 and 2018, specifically examining discourse on nuclear power and nuclear safety. We analyze various topic distributions across the corpus and trace their temporal evolution to uncover long-term trends and shifts in public discourse. This enables us to more accurately explore patterns in public discourse, including the co-occurrence of themes related to nuclear power and nuclear weapons and their shifts in topic importance over time. Our study demonstrates the scalability and contextual sensitivity of BERTopic as an alternative to traditional approaches, offering richer insights into historical discourses extracted from newspaper archives. These findings contribute to historical, nuclear, and social-science research while reflecting on current limitations and proposing potential directions for future work.

cs.CL↗

arXiv2025

RAGVUE: A Diagnostic View for Explainable and Automated Evaluation of Retrieval-Augmented Generation

Evaluating Retrieval-Augmented Generation (RAG) systems remains a challenging task: existing metrics often collapse heterogeneous behaviors into single scores and provide little insight into whether errors arise from retrieval,reasoning, or grounding. In this paper, we introduce RAGVUE, a diagnostic and explainable framework for automated, reference-free evaluation of RAG pipelines. RAGVUE decomposes RAG behavior into retrieval quality, answer relevance and completeness, strict claim-level faithfulness, and judge calibration. Each metric includes a structured explanation, making the evaluation process transparent. Our framework supports both manual metric selection and fully automated agentic evaluation. It also provides a Python API, CLI, and a local Streamlit interface for interactive usage. In comparative experiments, RAGVUE surfaces fine-grained failures that existing tools such as RAGAS often overlook. We showcase the full RAGVUE workflow and illustrate how it can be integrated into research pipelines and practical RAG development. The source code and detailed instructions on usage are publicly available on GitHub

cs.CL↗

arXiv2024

TensAIR: Real-Time Training of Neural Networks from Data-streams

Online learning (OL) from data streams is an emerging area of research that encompasses numerous challenges from stream processing, machine learning, and networking. Stream-processing platforms, such as Apache Kafka and Flink, have basic extensions for the training of Artificial Neural Networks (ANNs) in a stream-processing pipeline. However, these extensions were not designed to train ANNs in real-time, and they suffer from performance and scalability issues when doing so. This paper presents TensAIR, the first OL system for training ANNs in real time. TensAIR achieves remarkable performance and scalability by using a decentralized and asynchronous architecture to train ANN models (either freshly initialized or pre-trained) via DASGD (decentralized and asynchronous stochastic gradient descent). We empirically demonstrate that TensAIR achieves a nearly linear scale-out performance in terms of (1) the number of worker nodes deployed in the network, and (2) the throughput at which the data batches arrive at the dataflow operators. We depict the versatility of TensAIR by investigating both sparse (word embedding) and dense (image classification) use cases, for which TensAIR achieved from 6 to 116 times higher sustainable throughput rates than state-of-the-art systems for training ANN in a stream-processing pipeline.

cs.LG↗

arXiv2023

BigText-QA: Question Answering over a Large-Scale Hybrid Knowledge Graph

Answering complex questions over textual resources remains a challenge, particularly when dealing with nuanced relationships between multiple entities expressed within natural-language sentences. To this end, curated knowledge bases (KBs) like YAGO, DBpedia, Freebase, and Wikidata have been widely used and gained great acceptance for question-answering (QA) applications in the past decade. While these KBs offer a structured knowledge representation, they lack the contextual diversity found in natural-language sources. To address this limitation, BigText-QA introduces an integrated QA approach, which is able to answer questions based on a more redundant form of a knowledge graph (KG) that organizes both structured and unstructured (i.e., "hybrid") knowledge in a unified graphical representation. Thereby, BigText-QA is able to combine the best of both worlds$\unicode{x2013}$a canonical set of named entities, mapped to a structured background KB (such as YAGO or Wikidata), as well as an open set of textual clauses providing highly diversified relational paraphrases with rich context information. Our experimental results demonstrate that BigText-QA outperforms DrQA, a neural-network-based QA system, and achieves competitive results to QUEST, a graph-based unsupervised QA system.

cs.CL↗

arXiv2023

Convergence Analysis of Decentralized ASGD

Over the last decades, Stochastic Gradient Descent (SGD) has been intensively studied by the Machine Learning community. Despite its versatility and excellent performance, the optimization of large models via SGD still is a time-consuming task. To reduce training time, it is common to distribute the training process across multiple devices. Recently, it has been shown that the convergence of asynchronous SGD (ASGD) will always be faster than mini-batch SGD. However, despite these improvements in the theoretical bounds, most ASGD convergence-rate proofs still rely on a centralized parameter server, which is prone to become a bottleneck when scaling out the gradient computations across many distributed processes. In this paper, we present a novel convergence-rate analysis for decentralized and asynchronous SGD (DASGD) which does not require partial synchronization among nodes nor restrictive network topologies. Specifically, we provide a bound of $\mathcal{O}(\sigma\epsilon^{-2}) + \mathcal{O}(QS_{avg}\epsilon^{-3/2}) + \mathcal{O}(S_{avg}\epsilon^{-1})$ for the convergence rate of DASGD, where $S_{avg}$ is the average staleness between models, $Q$ is a constant that bounds the norm of the gradients, and $\epsilon$ is a (small) error that is allowed within the bound. Furthermore, when gradients are not bounded, we prove the convergence rate of DASGD to be $\mathcal{O}(\sigma\epsilon^{-2}) + \mathcal{O}(\sqrt{\hat{S}_{avg}\hat{S}_{max}}\epsilon^{-1})$, with $\hat{S}_{max}$ and $\hat{S}_{avg}$ representing a loose version of the average and maximum staleness, respectively. Our convergence proof holds for a fixed stepsize and any non-convex, homogeneous, and L-smooth objective function. We anticipate that our results will be of high relevance for the adoption of DASGD by a broad community of researchers and developers.

cs.LG↗

arXiv2023

OPTWIN: Drift identification with optimal sub-windows

Online Learning (OL) is a field of research that is increasingly gaining attention both in academia and industry. One of the main challenges of OL is the inherent presence of concept drifts, which are commonly defined as unforeseeable changes in the statistical properties of an incoming data stream over time. The detection of concept drifts typically involves analyzing the error rates produced by an underlying OL algorithm in order to identify if a concept drift occurred or not, such that the OL algorithm can adapt accordingly. Current concept-drift detectors perform very well, i.e., with low false negative rates, but they still tend to exhibit high false positive rates in the concept-drift detection. This may impact the performance of the learner and result in an undue amount of computational resources spent on retraining a model that actually still performs within its expected range. In this paper, we propose OPTWIN, our "OPTimal WINdow" concept drift detector. OPTWIN uses a sliding window of events over an incoming data stream to track the errors of an OL algorithm. The novelty of OPTWIN is to consider both the means and the variances of the error rates produced by a learner in order to split the sliding window into two provably optimal sub-windows, such that the split occurs at the earliest event at which a statistically significant difference according to either the $t$- or the $f$-tests occurred. We assessed OPTWIN over the MOA framework, using ADWIN, DDM, EDDM, STEPD and ECDD as baselines over 7 synthetic and real-world datasets, and in the presence of both sudden and gradual concept drifts. In our experiments, we show that OPTWIN surpasses the F1-score of the baselines in a statistically significant manner while maintaining a lower detection delay and saving up to 21% of time spent on retraining the models.

cs.LG↗

arXiv2022

Enriching Relation Extraction with OpenIE

Relation extraction (RE) is a sub-discipline of information extraction (IE) which focuses on the prediction of a relational predicate from a natural-language input unit (such as a sentence, a clause, or even a short paragraph consisting of multiple sentences and/or clauses). Together with named-entity recognition (NER) and disambiguation (NED), RE forms the basis for many advanced IE tasks such as knowledge-base (KB) population and verification. In this work, we explore how recent approaches for open information extraction (OpenIE) may help to improve the task of RE by encoding structured information about the sentences' principal units, such as subjects, objects, verbal phrases, and adverbials, into various forms of vectorized (and hence unstructured) representations of the sentences. Our main conjecture is that the decomposition of long and possibly convoluted sentences into multiple smaller clauses via OpenIE even helps to fine-tune context-sensitive language models such as BERT (and its plethora of variants) for RE. Our experiments over two annotated corpora, KnowledgeNet and FewRel, demonstrate the improved accuracy of our enriched models compared to existing RE approaches. Our best results reach 92% and 71% of F1 score for KnowledgeNet and FewRel, respectively, proving the effectiveness of our approach on competitive benchmarks.

cs.CL↗

arXiv2022

Robust and Provable Guarantees for Sparse Random Embeddings

In this work, we improve upon the guarantees for sparse random embeddings, as they were recently provided and analyzed by Freksen at al. (NIPS'18) and Jagadeesan (NIPS'19). Specifically, we show that (a) our bounds are explicit as opposed to the asymptotic guarantees provided previously, and (b) our bounds are guaranteed to be sharper by practically significant constants across a wide range of parameters, including the dimensionality, sparsity and dispersion of the data. Moreover, we empirically demonstrate that our bounds significantly outperform prior works on a wide range of real-world datasets, such as collections of images, text documents represented as bags-of-words, and text sequences vectorized by neural embeddings. Behind our numerical improvements are techniques of broader interest, which improve upon key steps of previous analyses in terms of (c) tighter estimates for certain types of quadratic chaos, (d) establishing extreme properties of sparse linear forms, and (e) improvements on bounds for the estimation of sums of independent random variables.

cs.LG↗

arXiv2020

Revisiting Initialization of Neural Networks

The proper initialization of weights is crucial for the effective training and fast convergence of deep neural networks (DNNs). Prior work in this area has mostly focused on balancing the variance among weights per layer to maintain stability of (i) the input data propagated forwards through the network and (ii) the loss gradients propagated backwards, respectively. This prevalent heuristic is however agnostic of dependencies among gradients across the various layers and captures only firstorder effects. In this paper, we propose and discuss an initialization principle that is based on a rigorous estimation of the global curvature of weights across layers by approximating and controlling the norm of their Hessian matrix. The proposed approach is more systematic and recovers previous results for DNN activations such as smooth functions, dropouts, and ReLU. Our experiments on Word2Vec and the MNIST/CIFAR image classification tasks confirm that tracking the Hessian norm is a useful diagnostic tool which helps to more rigorously initialize weights

cs.LG↗

arXiv2020

AIR: A Light-Weight Yet High-Performance Dataflow Engine based on Asynchronous Iterative Routing

Distributed Stream Processing Systems (DSPSs) are among the currently most emerging topics in data management, with applications ranging from real-time event monitoring to processing complex dataflow programs and big data analytics. The major market players in this domain are clearly represented by Apache Spark and Flink, which provide a variety of frontend APIs for SQL, statistical inference, machine learning, stream processing, and many others. Yet rather few details are reported on the integration of these engines into the underlying High-Performance Computing (HPC) infrastructure and the communication protocols they use. Spark and Flink, for example, are implemented in Java and still rely on a dedicated master node for managing their control flow among the worker nodes in a compute cluster. In this paper, we describe the architecture of our AIR engine, which is designed from scratch in C++ using the Message Passing Interface (MPI), pthreads for multithreading, and is directly deployed on top of a common HPC workload manager such as SLURM. AIR implements a light-weight, dynamic sharding protocol (referred to as "Asynchronous Iterative Routing"), which facilitates a direct and asynchronous communication among all client nodes and thereby completely avoids the overhead induced by the control flow with a master node that may otherwise form a performance bottleneck. Our experiments over a variety of benchmark settings confirm that AIR outperforms Spark and Flink in terms of latency and throughput by a factor of up to 15; moreover, we demonstrate that AIR scales out much better than existing DSPSs to clusters consisting of up to 8 nodes and 224 cores.

cs.DC↗

arXiv2019

Lineage-Aware Temporal Windows: Supporting Set Operations in Temporal-Probabilistic Databases

In temporal-probabilistic (TP) databases, the combination of the temporal and the probabilistic dimension adds significant overhead to the computation of set operations. Although set queries are guaranteed to yield linearly sized output relations, existing solutions exhibit quadratic runtime complexity. They suffer from redundant interval comparisons and additional joins for the formation of lineage expressions. In this paper, we formally define the semantics of set operations in TP databases and study their properties. For their efficient computation, we introduce the lineage-aware temporal window, a mechanism that directly binds intervals with lineage expressions. We suggest the lineage-aware window advancer (LAWA) for producing the windows of two TP relations in linearithmic time, and we implement all TP set operations based on LAWA. By exploiting the flexibility of lineage-aware temporal windows, we perform direct filtering of irrelevant intervals and finalization of output lineage expressions and thus guarantee that no additional computational cost or buffer space is needed. A series of experiments over both synthetic and real-world datasets show that (a) our approach has predictable performance, depending only on the input size and not on the number of time intervals per fact or their overlap, and that (b) it outperforms state-of-the-art approaches in both temporal and probabilistic databases.

cs.DB↗

arXiv2019

Generalized Lineage-Aware Temporal Windows: Supporting Outer and Anti Joins in Temporal-Probabilistic Databases

The result of a temporal-probabilistic (TP) join with negation includes, at each time point, the probability with which a tuple of a positive relation ${\bf p}$ matches none of the tuples in a negative relation ${\bf n}$, for a given join condition $θ$. TP outer and anti joins thus resemble the characteristics of relational outer and anti joins also in the case when there exist time points at which input tuples from ${\bf p}$ have non-zero probabilities to be $true$ and input tuples from ${\bf n}$ have non-zero probabilities to be $false$, respectively. For the computation of TP joins with negation, we introduce generalized lineage-aware temporal windows, a mechanism that binds an output interval to the lineages of all the matching valid tuples of each input relation. We group the windows of two TP relations into three disjoint sets based on the way attributes, lineage expressions and intervals are produced. We compute all windows in an incremental manner, and we show that pipelined computations allow for the direct integration of our approach into PostgreSQL. We thereby alleviate the prevalent redundancies in the interval computations of existing approaches, which is proven by an extensive experimental evaluation with real-world datasets.

cs.DB↗

arXiv2018

A General Framework for Anytime Approximation in Probabilistic Databases

Anytime approximation algorithms that compute the probabilities of queries over probabilistic databases can be of great use to statistical learning tasks. Those approaches have been based so far on either (i) sampling or (ii) branch-and-bound with model-based bounds. We present here a more general branch-and-bound framework that extends the possible bounds by using 'dissociation', which yields tighter bounds.

cs.DB↗

arXiv2016

Distributed Processing of Generalized Graph-Pattern Queries in SPARQL 1.1

We propose an efficient and scalable architecture for processing generalized graph-pattern queries as they are specified by the current W3C recommendation of the SPARQL 1.1 "Query Language" component. Specifically, the class of queries we consider consists of sets of SPARQL triple patterns with labeled property paths. From a relational perspective, this class resolves to conjunctive queries of relational joins with additional graph-reachability predicates. For the scalable, i.e., distributed, processing of this kind of queries over very large RDF collections, we develop a suitable partitioning and indexing scheme, which allows us to shard the RDF triples over an entire cluster of compute nodes and to process an incoming SPARQL query over all of the relevant graph partitions (and thus compute nodes) in parallel. Unlike most prior works in this field, we specifically aim at the unified optimization and distributed processing of queries consisting of both relational joins and graph-reachability predicates. All communication among the compute nodes is established via a proprietary, asynchronous communication protocol based on the Message Passing Interface.

cs.DB↗

arXiv2016

Learning Tuple Probabilities

Learning the parameters of complex probabilistic-relational models from labeled training data is a standard technique in machine learning, which has been intensively studied in the subfield of Statistical Relational Learning (SRL), but---so far---this is still an under-investigated topic in the context of Probabilistic Databases (PDBs). In this paper, we focus on learning the probability values of base tuples in a PDB from labeled lineage formulas. The resulting learning problem can be viewed as the inverse problem to confidence computations in PDBs: given a set of labeled query answers, learn the probability values of the base tuples, such that the marginal probabilities of the query answers again yield in the assigned probability labels. We analyze the learning problem from a theoretical perspective, cast it into an optimization problem, and provide an algorithm based on stochastic gradient descent. Finally, we conclude by an experimental evaluation on three real-world and one synthetic dataset, thus comparing our approach to various techniques from SRL, reasoning in information extraction, and optimization.

cs.DB↗

arXiv2012

Proceedings of the first International Workshop On Open Data, WOD-2012

WOD-2012 aims at facilitating new trends and ideas from a broad range of topics concerned within the widely-spread Open Data movement, from the viewpoint of computer science research. While being most commonly known from the recent Linked Open Data movement, the concept of publishing data explicitly as Open Data has meanwhile developed many variants and facets that go beyond publishing large and highly structured RDF/S repositories. Open Data comprises text and semi-structured data, but also open multi-modal contents, including music, images, and videos. With the increasing amount of data that is published by governments (see, e.g., data.gov, data.gov.uk or data.gouv.fr), by international organizations (data.worldbank.org or data.undp.org) and by scientific communities (tdar.org, cds.u-strasbg.fr, GenBank, IRIS or KNB) explicitly under an Open Data policy, new challenges arise not only due to the scale at which this data becomes available. A number of community-based conferences accommodate tracks or workshops which are dedicated to Open Data. However, WOD aims to be a premier venue to gather researchers and practitioners who are contributing to and interested in the emerging field of managing Open Data from a computer science perspective. Hence, it is a unique opportunity to find in a single place up-to-date scientific works on Web-scale Open Data issues that have so far only partially been addressed by different research communities such as Databases, Data Mining and Knowledge Management, Distributed Systems, Data Privacy, and Data Visualization.

cs.DL↗