Papers by Anette Frank

Exploring Large Language Models for Classical Philology (2023.acl-long)

Challenge:	Existing work on discourse-aware self-attention models for reading comprehension uses annotations .
Approach:	They propose to use linguistic annotations as a basis for a Discourse-Aware Semantic Self-Attention encoder for reading comprehension on narrative texts.
Outcome:	The proposed model improves reading comprehension performance on narrative texts up to +3.4 Rouge-L . it also improves inter- and cross-sentential discourse relations, sentence-internal semantic role relations, and long-distance coreference relations.

Copied to clipboard

Frederick Riemenschneider, Anette Frank

Language Models

Challenge:	Recent advances in NLP have led to the creation of powerful language models for many languages including Ancient Greek and Latin.
Approach:	They propose to use encoder-only and encoder decoder architectures to create four models for Ancient Greek that vary along two dimensions for tasks of interest for Classical languages.
Outcome:	The proposed models improve on existing models of Ancient Greek and Latin and provide a large pre-training corpus for Ancient Greek to support the creation of a larger, comparable model zoo for Classical Philology.

From Argumentation to Deliberation: Perspectivized Stance Vectors for Fine-grained (Dis)agreement Analysis (2025.findings-naacl)

Copied to clipboard

Moritz Plenz, Philipp Heinisch, Janosch Gehring, Philipp Cimiano, Anette Frank

Challenge:	Existing methods to identify conflict resolution points require a deeper analysis of arguments and the perspectives they are grounded in.
Approach:	They propose a framework for a deliberative analysis of arguments in a computational argumentation setup.
Outcome:	The proposed framework allows us to identify actionable options for conflict resolution, as a first step towards deliberation.

COCO-EX: A Tool for Linking Concepts from Texts to ConceptNet (2021.eacl-demos)

Copied to clipboard

Maria Becker, Katharina Korfhage, Anette Frank

Structured Data in NLP

Challenge:	ConceptNet is a semantic network which contains general commonsense facts about the world, e.g., Birds can fly or Computers are used for sending e-mails.
Approach:	They propose a tool for Extracting Concepts from texts and linking them to ConceptNet, using the maximum relational information stored in ConceptNet.
Outcome:	The proposed method extracts meaningful concepts from natural language texts and links them to conjunct concept nodes in ConceptNet, utilizing the maximum of relational information stored in the KnowledgeGraph.

Translate and Label! An Encoder-Decoder Approach for Cross-lingual Semantic Role Labeling (D19-1)

Copied to clipboard

Angel Daza, Anette Frank

Multilinguality

Challenge:	Unlike annotation projection techniques, our model does not need parallel data during inference time.
Approach:	They propose a cross-lingual Encoder-Decoder model that simultaneously translates and generates sentences with semantic role annotations in a resource-poor target language.
Outcome:	The proposed model can be applied in monolingual, multilingual and cross-lingual settings and produces dependency-based and span-based annotations.

DeModify: A Dataset for Analyzing Contextual Constraints on Modifier Deletion (L18-1)

Copied to clipboard

Vivi Nastase, Devon Fritz, Anette Frank

Challenge:	a text fragment is discarded when it has a smaller context, causing it to acquire a new meaning or even become false.
Approach:	They build a dataset to study the effect of modifiers on the larger context . they focus on single-word modifiers, the smallest unit that can be considered disposable .
Outcome:	The proposed dataset aims to determine whether modifiers can be removed without undesirable consequences.

Graph Language Models (2024.acl-long)

Copied to clipboard

Moritz Plenz, Anette Frank

SETI: Systematicity Evaluation of Textual Inference (2023.findings-acl)

Challenge:	Language Models (LMs) are the workhorses of NLP, but their interplay with structured knowledge graphs (KGs) is still actively researched.
Approach:	They propose a Graph Language Model (GLM) that integrates the strengths of both approaches and mitigates their weaknesses.
Outcome:	Empirical evaluations show that the proposed model surpasses both LM- and GNN-based baselines in supervised and zero-shot setting, demonstrating their versatility.

Copied to clipboard

Xiyan Fu, Anette Frank

Language Models

Challenge:	Existing pre-trained language models (PLMs) have shown remarkable performance on this task, but little is known about their ability to address compositional generalization.
Approach:	They propose a benchmark to evaluate pre-trained language models' systematicity in the domain of textual inference.
Outcome:	The proposed benchmark evaluates pre-trained language models on six widely used PLMs.

MM-SHAP: A Performance-agnostic Metric for Measuring Multimodal Contributions in Vision and Language Models & Tasks (2023.acl-long)

Copied to clipboard

Letitia Parcalabescu, Anette Frank

Multimodality

Challenge:	Vision and language models exploit unrobust indicators in individual modalities instead of focusing on relevant information in each modality.
Approach:	They propose a performance-agnostic multimodality score based on Shapley values that quantifies in which proportions a multimodal model uses individual modalities.
Outcome:	The proposed model can quantify in which proportions a multimodal model uses individual modalities for different tasks and datasets.

Ranking and Selecting Multi-Hop Knowledge Paths to Better Predict Human Needs (N19-1)

Copied to clipboard

Explainability & Interpretability in NLP

Challenge:	Existing methods to categorize sentiments and emotions in text are limited.
Approach:	They propose to extract, rank, filter and select multi-hop relation paths from a commonsense knowledge resource to interpret the expression of sentiment in terms of their underlying human needs.
Outcome:	The proposed method improves on a recently published dataset for categorizing human needs.

X-SRL: A Parallel Cross-Lingual Semantic Role Labeling Dataset (2020.emnlp-main)

Copied to clipboard

Angel Daza, Anette Frank

Multilinguality

Challenge:	Existing multilingual SRL datasets contain disparate annotation styles or come from different domains, hampering generalization in multilingual learning.
Approach:	They propose to automatically construct an SRL corpus that is parallel in four languages with unified predicate and role annotations that are fully comparable across languages.
Outcome:	The proposed method improves performance for English SRL in weaker languages.

On Measuring Faithfulness or Self-consistency of Natural Language Explanations (2024.acl-long)

Copied to clipboard

Letitia Parcalabescu, Anette Frank

Explainability & Interpretability in NLP

Challenge:	Large language models (LLMs) can explain their predictions through post-hoc or Chain-of-Thought explanations, but they can also make unfaithful explanations that hide their sensitivity to biasing inputs.
Approach:	They propose to use a model-based consistency test to judge the faithfulness of post-hoc or Chain-of-Thought explanations rather than model-specific faithfulness tests.
Outcome:	The proposed tests do not measure faithfulness to model explainability but rather their self-consistency at output level.

Weisfeiler-Leman in the Bamboo: Novel AMR Graph Metrics and a Benchmark for AMR Graph Similarity (2021.tacl-1)

Copied to clipboard

Juri Opitz, Angel Daza, Anette Frank

SBERT studies Meaning Representations: Decomposing Sentence Embeddings into Explainable Semantic Features (2022.aacl-main)

Challenge:	Existing metrics for assessing the similarity of abstract meaning representations (AMRs) have complementary strengths and weaknesses, and are expensive to implement.
Approach:	They propose a Benchmark for AMR Metrics based on overt objectives that can be used to assess graph-based similarity metrics.
Outcome:	The proposed metrics match contextualized substructures and induce n:m alignments between their nodes.

Copied to clipboard

Juri Opitz, Anette Frank

VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena (2022.acl-long)

Challenge:	Abstract Meaning Representation (S3BERT) embeddings are composed of explainable sub-embeddings that emphasize various sentence meaning features.
Approach:	They propose to induce Semantically Structured Sentence BERT embeddings (S3BERT) that emphasize various sentence meaning features.
Outcome:	The proposed model shows high correlation to human similarity ratings, but lacks interpretability.

Copied to clipboard

Letitia Parcalabescu, Michele Cafagna, Lilitta Muradjan, Anette Frank, Iacer Calixto, Albert Gatt

Implicit Knowledge in Argumentative Texts: An Annotated Corpus (2020.lrec-1)

Challenge:	Current models have difficulty addressing most phenomena, despite performance gains on many V&L tasks.
Approach:	They propose a benchmark for testing general-purpose pretrained vision and language models for their visio-linguistic grounding capabilities on specific linguistic phenomena.
Outcome:	The proposed model can be used to evaluate models on linguistic phenomena in the visual modality.

Copied to clipboard

Maria Becker, Katharina Korfhage, Anette Frank

Cross-Lingual Generalization and Compression: From Language-Specific to Shared Neurons (2025.acl-long)

Challenge:	Especially in argumentative texts, people omit information that seems clear and evident . a computational system typically does not possess commonsense or domain-specific knowledge to reconstruct implied information.
Approach:	They build a corpus of human annotations of missing and implied information in argumentative texts.
Outcome:	The proposed dataset can help to assist automated argument analysis and guide the process of revealing implicit information in argumentative texts automatically.

Copied to clipboard

Frederick Riemenschneider, Anette Frank

Cross-Lingual Transfer

Challenge:	Existing evidence suggests that multilingual language models can transfer knowledge across languages without explicit cross-lingual supervision.
Approach:	They analyze the parameter spaces of three multilingual language models to examine their representations . they find that models evolve from language-specific representations to more specialized layer functions .
Outcome:	The proposed model can generate coherent English text, rather than spanish text, and it can generate generalized representations, the authors show.

COINS: Dynamically Generating COntextualized Inference Rules for Narrative Story Completion (2021.acl-long)

Copied to clipboard

With a Little Push, NLI Models can Robustly and Efficiently Predict Faithfulness (2023.acl-short)

Challenge:	Existing inference models are opaque, but they can be made more interpretable by explicitly generating interim inference rules and using them to guide the generation of task-specific textual outputs.
Approach:	They propose a recursive inference framework that iteratively reads context sentences and dynamically generates contextualized inference rules, encodes them, and uses them to guide output generation.
Outcome:	The proposed framework generates better story sentences than baseline models, and is more interpretable than existing models.

Copied to clipboard

Julius Steen, Juri Opitz, Anette Frank, Katja Markert

Responsible & Trustworthy NLP

Challenge:	Conditional language models generate unfaithful output that is not supported by their input . this jeopardizes trust in real-world applications, raising a need for automatic faithfulness metrics.
Approach:	They propose to augment conditional language models with robust inference procedures to improve faithfulness.
Outcome:	The proposed approach outperforms existing models on the TRUE benchmark.

Generating Hypothetical Events for Abductive Inference (2021.starsem-1)

Copied to clipboard

Responsible & Trustworthy NLP

Challenge:	Abductive reasoning is inference to the best explanation given an incomplete set of observations about everyday situations.
Approach:	They propose a model that generates what could happen next from a hypothetical scenario and then proposes the most plausible explanation from varying hypothetical scenarios.
Outcome:	The proposed model improves over previous vanilla pre-trained models fine-tuned on Abductive NLI.

The Mystery of Compositional Generalization in Graph-based Generative Commonsense Reasoning (2024.findings-emnlp)

Copied to clipboard

Xiyan Fu, Anette Frank

Towards a Decomposable Metric for Explainable Evaluation of Text Generation from AMR (2021.eacl-main)

Challenge:	Existing studies have found that LLMs are limited in scenarios that require generalization abilities, such as out-of-domain tasks.
Approach:	They propose a Compositional Generalization Challenge for Graph-based Commonsense Reasoning that requires models to generate a natural sentence based on given concepts and a reasoning graph.
Outcome:	The proposed framework is based on seven well-known LLMs and shows that they struggle in compositional generalization.

Copied to clipboard

Juri Opitz, Anette Frank

Juri Opitz, Letitia Parcalabescu, Anette Frank

Challenge:	Abstract meaning representations are typically evaluated using surface matching metrics . however, there are problems with these metrics, since they allow multiple surface realizations .
Approach:	They propose a decomposable metric that measures the distance between the original and reconstructed AMRs.
Outcome:	The proposed metric reduces the complexity of the evaluation process by allowing for multiple surface realizations.

Copied to clipboard

Moritz Plenz, Juri Opitz, Philipp Heinisch, Philipp Cimiano, Anette Frank

Challenge:	Abstract Meaning Representation (AMR) graphs are rooted, acyclic, directed, and edge-labeled.
Approach:	They propose a canonical Smatch metric that aligns variables of two graphs and assesses triple matches.
Outcome:	The proposed metric is more benevolent to only very slight meaning deviations and targets the fulfilment of all established criteria.

Copied to clipboard

Multimodality

Challenge:	Arguments often do not make explicit how a conclusion follows from its premises . we present a method for constructing Contextualized Commonsense Knowledge Graphs (CCKGs) that is efficient and high-quality .
Approach:	They propose an unsupervised method for constructing Contextualized Commonsense Knowledge Graphs (CCKGs) they use triplet similarities to extract contextually relevant knowledge paths .
Outcome:	The proposed method outperforms baselines and a GPT-3 based system in a knowledge-intense argumentation task.

A Dynamic, Interpreted CheckList for Meaning-oriented NLG Metric Evaluation – through the Lens of Semantic Similarity Rating (2022.starsem-1)

Copied to clipboard

Laura Zeidler, Juri Opitz, Anette Frank

Social Commonsense Reasoning with Multi-Head Knowledge Attention (2020.findings-emnlp)

Challenge:	Abstract Meaning Representation (AMR) is a popular NLG evaluation metric . however, it is difficult to judge where exactly such a metric fails .
Approach:	They propose a checklist for NLG evaluation metrics that focus on meaning by organizing them around meaning-relevant linguistic phenomena.
Outcome:	The proposed metric GraCo computes lexical cohesion graphs over AMR concepts.

Copied to clipboard

MAGMA – Multimodal Augmentation of Generative Models through Adapter-based Finetuning (2022.findings-emnlp)

Challenge:	Social Commonsense Reasoning requires understanding of text, knowledge about social events and their pragmatic implications, as well as commonsense reasoning skills.
Approach:	They propose a multi-head knowledge attention model that encodes semi-structured commonsense inference rules and learns to incorporate them in a transformer-based reasoning cell.
Outcome:	The proposed model improves performance on two reasoning tasks that require different reasoning skills.

Copied to clipboard

Constantin Eichenberg, Sidney Black, Samuel Weinbach, Letitia Parcalabescu, Anette Frank

Knowledgeable Reader: Enhancing Cloze-Style Reading Comprehension with External Commonsense Knowledge (P18-1)

Challenge:	Large-scale pretraining is becoming the norm in Vision-Language (VL) modeling.
Approach:	They propose a method for augmenting generative language models with additional modalities using adapter-based finetuning.
Outcome:	The proposed method outperforms Frozen on open-ended generative tasks while maintaining the language model weights.

Copied to clipboard

Todor Mihaylov, Anette Frank

SRL4ORL: Improving Opinion Role Labeling Using Multi-Task Learning with Semantic Role Labeling (N18-1)

Challenge:	a new model for reading comprehension integrates external commonsense knowledge . cloze-style reading comprehension is a language understanding task similar to question answering .
Approach:	They propose a reading comprehension model that integrates external commonsense knowledge in a cloze-style setting.
Outcome:	The proposed model improves results over a very strong baseline on a hard Common Nouns dataset, making it a strong competitor of more complex models.

Copied to clipboard

Ana Marasović, Anette Frank