Prof. Dr. Alan Akbik
Profil
Forschungsthemen5
Eidetische Repräsentationen Natürlicher Sprache
Quelle ↗Förderer: DFG Nachwuchsgruppe Zeitraum: 07/2021 - 06/2028 Projektleitung: Prof. Dr. Alan Akbik
Inscribed
Quelle ↗Förderer: BMWE: EXIST Zeitraum: 04/2026 - 03/2027 Projektleitung: Prof. Dr. Alan Akbik
KI im Kundenservice (KIK)
Quelle ↗Förderer: Investitionsbank Berlin (IBB) Zeitraum: 02/2025 - 10/2027 Projektleitung: Prof. Dr. Alan Akbik, Prof. Dr. Stefan Lessmann
Modeling Neurogenesis for Continuous Learning
Quelle ↗Förderer: DFG Exzellenzstrategie Cluster Zeitraum: 01/2026 - 03/2027 Projektleitung: Prof. Dr. sc. nat. Verena Hafner
Mogli
Quelle ↗Förderer: BMWE: EXIST Zeitraum: 03/2026 - 02/2027 Projektleitung: Prof. Dr. Alan Akbik
Mögliche Industrie-Partner10
Stand: 26.4.2026, 19:48:44 (Top-K=20, Min-Cosine=0.4)
OMQ GmbH
KT2 Treffer85.0%- KI im Kundenservice (KIK)K85.0%
- KI im Kundenservice (KIK)
- 31 Treffer58.6%
- EU: Observatory for Political Texts in European Democracies: A European Research Infrastructure (OPTED)P58.6%
- EU: Observatory for Political Texts in European Democracies: A European Research Infrastructure (OPTED)
Higher Functions - Sistemas Informáticos Inteligentes
P77 Treffer58.4%- QTLeap - Qualitätsübersetzung durch tiefe SprachanalysemethodenP58.4%
- QTLeap - Qualitätsübersetzung durch tiefe Sprachanalysemethoden
- 28 Treffer58.0%
- Translation for Massive Open Online CoursesP58.0%
- Translation for Massive Open Online Courses
- 29 Treffer58.0%
- Translation for Massive Open Online CoursesP58.0%
- Translation for Massive Open Online Courses
- 28 Treffer58.0%
- Translation for Massive Open Online CoursesP58.0%
- Translation for Massive Open Online Courses
- 5 Treffer57.0%
- Zuwendung im Rahmen des Programms „exist – Existenzgründungen aus der Wissenschaft“ aus dem Bundeshaushalt, Einzelplan 09, Kapitel 02, Titel 68607, Haushaltsjahr 2026, sowie aus Mitteln des Europäischen Strukturfonds (hier Euro-päischer Sozialfonds Plus – ESF Plus) Förderperiode 2021-2027 – Kofinanzierung für das Vorhaben: „exist Women“T57.0%
- Zuwendung im Rahmen des Programms „exist – Existenzgründungen aus der Wissenschaft“ aus dem Bundeshaushalt, Einzelplan 09, Kapitel 02, Titel 68607, Haushaltsjahr 2026, sowie aus Mitteln des Europäischen Strukturfonds (hier Euro-päischer Sozialfonds Plus – ESF Plus) Förderperiode 2021-2027 – Kofinanzierung für das Vorhaben: „exist Women“
- 7 Treffer56.9%
- Lösung gekoppelter Probleme in der Nanoelektronik (nanoCOPS)P56.9%
- Lösung gekoppelter Probleme in der Nanoelektronik (nanoCOPS)
- 7 Treffer56.9%
- Lösung gekoppelter Probleme in der Nanoelektronik (nanoCOPS)P56.9%
- Lösung gekoppelter Probleme in der Nanoelektronik (nanoCOPS)
- 7 Treffer56.9%
- Lösung gekoppelter Probleme in der Nanoelektronik (nanoCOPS)P56.9%
- Lösung gekoppelter Probleme in der Nanoelektronik (nanoCOPS)
Publikationen25
Top 25 nach Zitationen — Quelle: OpenAlex (BAAI/bge-m3 embedded für Matching).
International Conference on Computational Linguistics · 1006 Zitationen
Recent advances in language modeling using recurrent neural networks have made it viable to model language as distributions over characters. By learning to predict the next character on the basis of previous characters, such models have been shown to automatically internalize linguistic concepts such as words, sentences, subclauses and even sentiment. In this paper, we propose to leverage the internal states of a trained character language model to produce a novel type of word embedding which we refer to as contextual string embeddings. Our proposed embeddings have the distinct properties that they (a) are trained without any explicit notion of words and thus fundamentally model words as sequences of characters, and (b) are contextualized by their surrounding text, meaning that the same word will have different embeddings depending on its contextual use. We conduct a comparative evaluation against previous embeddings and find that our embeddings are highly useful for downstream tasks: across four classic sequence labeling tasks we consistently outperform the previous state-of-the-art. In particular, we significantly outperform previous work on English and German named entity recognition (NER), allowing us to report new state-of-the-art F1-scores on the CoNLL03 shared task. We release all code and pre-trained language models in a simple-to-use framework to the research community, to enable reproduction of these experiments and application of our proposed embeddings to other tasks: https://github.com/zalandoresearch/flair
396 Zitationen · DOI
336 Zitationen · DOI
Alan Akbik, Tanja Bergmann, Duncan Blythe, Kashif Rasul, Stefan Schweter, Roland Vollgraf. Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics (Demonstrations). 2019.
308 Zitationen · DOI
Alan Akbik, Tanja Bergmann, Roland Vollgraf. Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). 2019.
109 Zitationen · DOI
81 Zitationen · DOI
Semantic role labeling (SRL) is crucial to natural language understanding as it identifies the predicate-argument structure in text with semantic labels. Unfortunately, resources required to construct SRL models are expensive to obtain and simply do not exist for most languages. In this paper, we present a two-stage method to enable the construction of SRL models for resourcepoor languages by exploiting monolingual SRL and multilingual parallel data. Experimental results show that our method outperforms existing methods. We use our method to generate Proposition Banks with high to reasonable quality for 7 languages in three language families and release these resources to the research community.
North American Chapter of the Association for Computational Linguistics · 64 Zitationen
Current techniques for Open Information Extraction (OIE) focus on the extraction of binary facts and suffer significant quality loss for the task of extracting higher order N-ary facts. This quality loss may not only affect the correctness, but also the completeness of an extracted fact. We present KrakeN, an OIE system specifically designed to capture N-ary facts, as well as the results of an experimental study on extracting facts from Web text in which we examine the issue of fact completeness. Our preliminary experiments indicate that KrakeN is a high precision OIE approach that captures more facts per sentence at greater completeness than existing OIE approaches, but is vulnerable to noisy and ungrammatical text.
50 Zitationen · DOI
State-of-the-art approaches for text classification leverage a transformer architecture with a linear layer on top that outputs a class distribution for a given prediction problem. While effective, this approach suffers from conceptual limitations that affect its utility in few-shot or zero-shot transfer learning scenarios. First, the number of classes to predict needs to be pre-defined. In a transfer learning setting, in which new classes are added to an already trained classifier, all information contained in a linear layer is therefore discarded, and a new layer is trained from scratch. Second, this approach only learns the semantics of classes implicitly from training examples, as opposed to leveraging the explicit semantic information provided by the natural language names of the classes. For instance, a classifier trained to predict the topics of news articles might have classes like "business" or "sports" that themselves carry semantic information. Extending a classifier to predict a new class named "politics" with only a handful of training examples would benefit from both leveraging the semantic information in the name of a new class and using the information contained in the already trained linear layer. This paper presents a novel formulation of text classification that addresses these limitations. It imbues the notion of the task at hand into the transformer model itself by factorizing arbitrary classification problems into a generic binary classification problem. We present experiments in few-shot and zero-shot transfer learning that show that our approach significantly outperforms previous approaches on small training data and can even learn to predict new classes with no training examples at all. The implementation of our model is publicly available at:
30 Zitationen
Unsupervised Relation Extraction (URE) is the task of extracting relations of a priori unknown semantic types using clustering methods on a vector space model of entity pairs and patterns. In this paper, we show that an informed feature generation technique based on dependency trees significantly improves clustering quality, as measured by the F-score, and therefore the ability of the URE method to discover relations in text. Furthermore, we extend URE to produce a set of weighted patterns for each identified relation that can be used by an information extraction system to find further instances of this relation. Each pattern is assigned to one or multiple relations with different confidence strengths, indicating how reliably a pattern evokes a relation, using the theory of Discriminative Category Matching. We evaluate our findings in two tasks against strong baselines and show significant improvements both in relation discovery and information extraction.
Propminer: A Workflow for Interactive Information Extraction and Exploration using Dependency Trees
201321 Zitationen
The use of deep syntactic information such as typed dependencies has been shown to be very effective in Information Extraction. Despite this potential, the process of manually creating rule-based information extractors that operate on dependency trees is not intuitive for persons without an extensive NLP background. In this system demonstration, we present a tool and a workflow designed to enable initiate users to interactively explore the effect and expressivity of creating Information Extraction rules over dependency trees. We introduce the proposed five step workflow for creating information extractors, the graph query based rule language, as well as the core features of the PROP-MINER tool. 1
20 Zitationen · DOI
Semantic role labeling (SRL) identifies the predicate-argument structure in text with semantic labels. It plays a key role in understanding natural language. In this paper, we present POLYGLOT, a multilingual semantic role labeling system capable of semantically parsing sentences in 9 different languages from 4 different language groups. The core of POLYGLOT are SRL models for individual languages trained with automatically generated Proposition Banks (Akbik et al., 2015). The key feature of the system is that it treats the semantic labels of the English Proposition Bank as “universal semantic labels”: Given a sentence in any of the supported languages, POLYGLOT applies the corresponding SRL and predicts English PropBank frame and role annotation. The results are then visualized to facilitate the understanding of multilingual SRL with this unified semantic representation.
16 Zitationen · DOI
Annotation projection based on parallel corpora has shown great promise in inexpensively creating Proposition Banks for languages for which high-quality parallel corpora and syntactic parsers are available. In this paper, we present an experimental study where we apply this approach to three languages that lack such resources: Tamil, Bengali and Malayalam. We find an average quality difference of 6 to 20 absolute F-measure points vis-avis high-resource languages, which indicates that annotation projection alone is insufficient in low-resource scenarios. Based on these results, we explore the possibility of using annotation projection as a starting point for inexpensive data curation involving both experts and non-experts. We give an outline of what such a process may look like and present an initial study to discuss its potential and challenges.
arXiv (Cornell University) · 15 Zitationen · DOI
Current state-of-the-art approaches for named entity recognition (NER) typically consider text at the sentence-level and thus do not model information that crosses sentence boundaries. However, the use of transformer-based models for NER offers natural options for capturing document-level features. In this paper, we perform a comparative evaluation of document-level features in the two standard NER architectures commonly considered in the literature, namely "fine-tuning" and "feature-based LSTM-CRF". We evaluate different hyperparameters for document-level features such as context window size and enforcing document-locality. We present experiments from which we derive recommendations for how to model document context and present new state-of-the-art scores on several CoNLL-03 benchmark datasets. Our approach is integrated into the Flair framework to facilitate reproduction of our experiments.
Bioinformatics · 14 Zitationen · DOI
All our models are integrated into the Natural Language Processing (NLP) framework flair: https://github.com/flairNLP/flair. Code to reproduce our results is available at: https://github.com/hu-ner/hunflair2-experiments.
14 Zitationen · DOI
Matthias Vogt, Ulf Leser, Alan Akbik. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
International Conference on Computational Linguistics · 13 Zitationen
Semantic role labeling (SRL) is the task of identifying and labeling predicate-argument structures in sentences with semantic frame and role labels. A known challenge in SRL is the large number of low-frequency exceptions in training data, which are highly context-specific and difficult to generalize. To overcome this challenge, we propose the use of instance-based learning that performs no explicit generalization, but rather extrapolates predictions from the most similar instances in the training data. We present a variant of k-nearest neighbors (kNN) classification with composite features to identify nearest neighbors for SRL. We show that high-quality predictions can be derived from a very small number of similar instances. In a comparative evaluation we experimentally demonstrate that our instance-based learning approach significantly outperforms current state-of-the-art systems on both in-domain and out-of-domain data, reaching F1-scores of 89,28% and 79.91% respectively.
International Conference on Computational Linguistics · 13 Zitationen
The WIKIA project maintains wikis across a diverse range of subjects from areas of popular culture. Each wiki consists of collaboratively authored content and focuses on a particular topic, including franchises such as “Star Trek”, “Star Wars” and “The Simpsons”. In this paper, we investigate the use of such wikis to create Question-Answering (QA) systems for a given topic. Our key idea is to use a wiki as seed to gather large amounts of relevant text and to use semantic role labeling (SRL) methods to extract N-ary facts from this data. By applying our method to very large amounts of topically focused text, we propose to address the coverage issues that have been noted for QA systems built using such techniques. To illustrate the strengths and weaknesses of the proposed approach, we make a Web demonstrator of our system publicly available; it provides a QA view that enables users to pose natural language questions to the system and that visualizes how questions are interpreted and matched to answers. In addition, the demonstrator provides a graph exploration view in which users can directly browse the fact base in order to inspect the scope of the extracted information.
11 Zitationen · DOI
Crowdsourcing has proven to be an effective method for generating labeled data for a range of NLP tasks. However, multiple recent attempts of using crowdsourcing to generate gold-labeled training data for semantic role labeling (SRL) reported only modest results, indicating that SRL is perhaps too difficult a task to be effectively crowdsourced. In this paper, we postulate that while producing SRL annotation does require expert involvement in general, a large subset of SRL labeling tasks is in fact appropriate for the crowd. We present a novel workflow in which we employ a classifier to identify difficult annotation tasks and route each task either to experts or crowd workers according to their difficulties. Our experimental evaluation shows that the proposed approach reduces the workload for experts by over two-thirds, and thus significantly reduces the cost of producing SRL annotation at little loss in quality.
International Conference on Computational Linguistics · 11 Zitationen
In this paper, we propose and demonstrate Exploratory Relation Extraction (ERE), a novel approach to identifying and extracting relations from large text corpora based on user-driven and data-guided incremental exploration. We draw upon ideas from the information seeking paradigm of Exploratory Search (ES) to enable an exploration process in which users begin with a vaguely defined information need and progressively sharpen their definition of extraction tasks as they identify relations of interest in the underlying data. This process extends the application of Relation Extraction to use cases characterized by imprecise information needs and uncertainty regarding the information content of available data. We present an interactive workflow that allows users to build extractors based on entity types and human-readable extraction patterns derived from subtrees in dependency trees. In order to evaluate the viability of our approach on large text corpora, we conduct experiments on a dataset of over 160 million sentences with mentions of over 6 million FREEBASE entities extracted from the CLUEWEB09 corpus. Our experiments indicate that even non-expert users can intuitively use our approach to identify relations and create high precision extractors with minimal effort.
International Joint Conference on Natural Language Processing · 11 Zitationen
Unsupervised Relation Extraction (URE) methods automatically discover semantic relations in text corpora of unknown content and extract for each discovered relation a set of relation instances. Due to the sparsity of the feature space, URE is vulnerable to ambiguities and underspecification in patterns. In this paper, we propose to increase the discriminative power of patterns in URE using selectional restrictions (SR). We propose a method that utilizes a Web-derived soft clustering of n-grams to model selectional restrictions in the open domain. We comparatively evaluate our method against a baseline without SR, a setup in which standard 7class Named Entity types are used as SR and a setup that models SR using a finegrained entity type system. Our results indicate that modeling SR into patterns significantly improves the ability of URE to discover relations and enables the discovery of more fine-granular relations.
9 Zitationen · DOI
Angelo Ziletti, Alan Akbik, Christoph Berns, Thomas Herold, Marion Legler, Martina Viell. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Industry Track. 2022.
International Conference on Computational Linguistics · 9 Zitationen
We present PolyglotIE, a web-based tool for developing extractors that perform Information Extraction (IE) over multilingual data. Our tool has two core features: First, it allows users to develop extractors against a unified abstraction that is shared across a large set of natural languages. This means that an extractor needs only be created once for one language, but will then run on multilingual data without any additional effort or language-specific knowledge on part of the user. Second, it embeds this abstraction as a set of views within a declarative IE system, allowing users to quickly create extractors using a mature IE query language. We present PolyglotIE as a hands-on demo in which users can experiment with creating extractors, execute them on multilingual text and inspect extraction results. Using the UI, we discuss the challenges and potential of using unified, crosslingual semantic abstractions as basis for downstream applications. We demonstrate multilingual IE for 9 languages from 4 different language groups: English, German, French, Spanish, Japanese, Chinese, Arabic, Russian and Hindi.
arXiv (Cornell University) · 8 Zitationen · DOI
Instruction-tuned Large Language Models (LLMs) have recently showcased remarkable advancements in their ability to generate fitting responses to natural language instructions. However, many current works rely on manual evaluation to judge the quality of generated responses. Since such manual evaluation is time-consuming, it does not easily scale to the evaluation of multiple models and model variants. In this short paper, we propose a straightforward but remarkably effective evaluation metric called SemScore, in which we directly compare model outputs to gold target responses using semantic textual similarity (STS). We conduct a comparative evaluation of the model outputs of 12 prominent instruction-tuned LLMs using 8 widely-used evaluation metrics for text generation. We find that our proposed SemScore metric outperforms all other, in many cases more complex, evaluation metrics in terms of correlation to human evaluation. These findings indicate the utility of our proposed metric for the evaluation of instruction-tuned LLMs.
International Conference on Computational Linguistics · 8 Zitationen
Semantic Role Labeling (SRL) is the task of identifying the predicate-argument structure in sentences with semantic frame and role labels. For the English language, the Proposition Bank provides both a lexicon of all possible semantic frames and large amounts of labeled training data. In order to expand SRL beyond English, previous work investigated automatic approaches based on parallel corpora to automatically generate Proposition Banks for new target languages (TLs). However, this approach heuristically produces the frame lexicon from word alignments, leading to a range of lexicon-level errors and inconsistencies. To address these issues, we propose to manually alias TL verbs to existing English frames. For instance, the German verb drehen may evoke several meanings, including “turn something” and “film something”. Accordingly, we alias the former to the frame TURN.01 and the latter to a group of frames that includes FILM.01 and SHOOT.03. We execute a large-scale manual aliasing effort for three target languages and apply the new lexicons to automatically generate large Proposition Banks for Chinese, French and German with manually curated frames. We present a detailed evaluation in which we find that our proposed approach significantly increases the quality and consistency of the generated Proposition Banks. We release these resources to the research community.
Language Resources and Evaluation · 8 Zitationen
Kooperationen1
Bestätigte Forscher↔Partner-Paare aus HU-FIS — Gold-Standard-Positive für das Matching.
KI im Kundenservice (KIK)
company
Stammdaten
Identität, Organisation und Kontakt aus HU-FIS.
- Name
- Prof. Dr. Alan Akbik
- Titel
- Prof. Dr.
- Fakultät
- Mathematisch-Naturwissenschaftliche Fakultät
- Institut
- Institut für Informatik
- Arbeitsgruppe
- Maschinelles Lernen (J)
- Telefon
- +49 30 2093-41185
- HU-FIS-Profil
- Quelle ↗
- Zuletzt gescrapt
- 26.4.2026, 01:01:51