Interpretability tools that offer explanations in the form of a dialogue have demonstrated their efficacy in enhancing users’ understanding (Slack et al., 2023; Shen et al., 2023), as one-off explanations may fall short in providing sufficient information to the user. Current solutions for dialogue-based explanations, however, often require external tools and modules and are not easily transferable to tasks they were not designed for. With LLMCheckup, we present an easily accessible tool that allows users to chat with any state-of-the-art large language model (LLM) about its behavior. We enable LLMs to generate explanations and perform user intent recognition without fine-tuning, by connecting them with a broad spectrum of Explainable AI (XAI) methods, including white-box explainability tools such as feature attributions, and self-explanations (e.g., for rationale generation). LLM-based (self-)explanations are presented as an interactive dialogue that supports follow-up questions and generates suggestions. LLMCheckup provides tutorials for operations available in the system, catering to individuals with varying levels of expertise in XAI and supporting multiple input modalities. We introduce a new parsing strategy that substantially enhances the user intent recognition accuracy of the LLM. Finally, we showcase LLMCheckup for the tasks of fact checking and commonsense question answering. Our code repository: https://github.com/DFKI-NLP/LLMCheckup

This survey analyses how external knowledge can be integrated into language models in the context of retrieval-augmentation.The main goal of this work is to give an overview of: (1) Which external knowledge can be augmented? (2) Given a knowledge source, how to retrieve from it and then integrate the retrieved knowledge? To achieve this, we define and give a mathematical formulation of retrieval-augmented knowledge integration (RAKI). We discuss retrieval and integration techniques separately in detail, for each of the following knowledge formats: knowledge graph, tabular and natural language.

Initially introduced as a machine translation model, the Transformer architecture has now become the foundation for modern deep learning architecture, with applications in a wide range of fields, from computer vision to natural language processing. Nowadays, to tackle increasingly more complex tasks, Transformer-based models are stretched to enormous sizes, requiring increasingly larger training datasets, and unsustainable amount of compute resources. The ubiquitous nature of the Transformer and its core component, the attention mechanism, are thus prime targets for efficiency research.In this work, we propose an alternative compatibility function for the self-attention mechanism introduced by the Transformer architecture. This compatibility function exploits an overlap in the learned representation of the traditional scaled dot-product attention, leading to a symmetric with pairwise coefficient dot-product attention. When applied to the pre-training of BERT-like models, this new symmetric attention mechanism reaches a score of 79.36 on the GLUE benchmark against 78.74 for the traditional implementation, leads to a reduction of 6% in the number of trainable parameters, and reduces the number of training steps required before convergence by half.

In this paper, we investigate the use of large language models (LLMs) to enhance the editorial process of rewriting customer help pages. We introduce a German-language dataset comprising Frequently Asked Question-Answer pairs, presenting both raw drafts and their revisions by professional editors. On this dataset, we evaluate the performance of four large language models (LLM) through diverse prompts tailored for the rewriting task. We conduct automatic evaluations of content and text quality using ROUGE, BERTScore, and ChatGPT. Furthermore, we let professional editors assess the helpfulness of automatically generated FAQ revisions for editorial enhancement. Our findings indicate that LLMs can produce FAQ reformulations beneficial to the editorial process. We observe minimal performance discrepancies among LLMs for this task, and our survey on helpfulness underscores the subjective nature of editors’ perspectives on editorial refinement.

This paper presents the dfki-mlst submission for the DialAM shared task (Ruiz-Dolz et al., 2024) on identification of argumentative and illocutionary relations in dialogue. Our model achieves best results in the global setting: 48.25 F1 at the focused level when looking only at the related arguments/locutions and 67.05 F1 at the general level when evaluating the complete argument maps. We describe our implementation of the data pre-processing, relation encoding and classification, evaluating 11 different base models and performing experiments with, e.g., node text combination and data augmentation. Our source code is publicly available.

Relation extraction (RE) is a fundamental task in information extraction, whose extension to multilingual settings has been hindered by the lack of supervised resources comparable in size to large English datasets such as TACRED (Zhang et al., 2017). To address this gap, we introduce the MultiTACRED dataset, covering 12 typologically diverse languages from 9 language families, which is created by machine-translating TACRED instances and automatically projecting their entity annotations. We analyze translation and annotation projection quality, identify error categories, and experimentally evaluate fine-tuned pretrained mono- and multilingual language models in common transfer learning scenarios. Our analyses show that machine translation is a viable strategy to transfer RE instances, with native speakers judging more than 83% of the translated instances to be linguistically and semantically acceptable. We find monolingual RE model performance to be comparable to the English original for many of the target languages, and that multilingual models trained on a combination of English and target language data can outperform their monolingual counterparts. However, we also observe a variety of translation and annotation projection errors, both due to the MT systems and linguistic features of the target languages, such as pronoun-dropping, compounding and inflection, that degrade dataset quality and RE model performance.

Saliency maps can explain a neural model’s predictions by identifying important input features. They are difficult to interpret for laypeople, especially for instances with many features. In order to make them more accessible, we formalize the underexplored task of translating saliency maps into natural language and compare methods that address two key challenges of this approach – what and how to verbalize. In both automatic and human evaluation setups, using token-level attributions from text classification tasks, we compare two novel methods (search-based and instruction-based verbalizations) against conventional feature importance representations (heatmap visualizations and extractive rationales), measuring simulatability, faithfulness, helpfulness and ease of understanding. Instructing GPT-3.5 to generate saliency map verbalizations yields plausible explanations which include associations, abstractive summarization and commonsense reasoning, achieving by far the highest human ratings, but they are not faithfully capturing numeric information and are inconsistent in their interpretation of the task. In comparison, our search-based, model-free verbalization approach efficiently completes templated verbalizations, is faithful by design, but falls short in helpfulness and simulatability. Our results suggest that saliency map verbalization makes feature attribution explanations more comprehensible and less cognitively challenging to humans than conventional representations.

Pre-trained language models (PLM) are effective components of few-shot named entity recognition (NER) approaches when augmented with continued pre-training on task-specific out-of-domain data or fine-tuning on in-domain data. However, their performance in low-resource scenarios, where such data is not available, remains an open question. We introduce an encoder evaluation framework, and use it to systematically compare the performance of state-of-the-art pre-trained representations on the task of low-resource NER. We analyze a wide range of encoders pre-trained with different strategies, model architectures, intermediate-task fine-tuning, and contrastive learning. Our experimental results across ten benchmark NER datasets in English and German show that encoder performance varies significantly, suggesting that the choice of encoder for a specific low-resource scenario needs to be carefully evaluated.

Prompting pre-trained language models has achieved impressive performance on various NLP tasks, especially in low data regimes. Despite the success of prompting in monolingual settings, applying prompt-based methods in multilingual scenarios has been limited to a narrow set of tasks, due to the high cost of handcrafting multilingual prompts. In this paper, we present the first work on prompt-based multilingual relation classification (RC), by introducing an efficient and effective method that constructs prompts from relation triples and involves only minimal translation for the class labels. We evaluate its performance in fully supervised, few-shot and zero-shot scenarios, and analyze its effectiveness across 14 languages, prompt variants, and English-task training in cross-lingual settings. We find that in both fully supervised and few-shot scenarios, our prompt method beats competitive baselines: fine-tuning XLM-R_EM and null prompts. It also outperforms the random baseline by a large margin in zero-shot experiments. Our method requires little in-language knowledge and can be used as a strong baseline for similar multilingual classification tasks.

Relation classification models are conventionally evaluated using only a single measure, e.g., micro-F1, macro-F1 or AUC. In this work, we analyze weighting schemes, such as micro and macro, for imbalanced datasets. We introduce a framework for weighting schemes, where existing schemes are extremes, and two new intermediate schemes. We show that reporting results of different weighting schemes better highlights strengths and weaknesses of a model.

Scholarly Argumentation Mining (SAM) has recently gained attention due to its potential to help scholars with the rapid growth of published scientific literature. It comprises two subtasks: argumentative discourse unit recognition (ADUR) and argumentative relation extraction (ARE), both of which are challenging since they require e.g. the integration of domain knowledge, the detection of implicit statements, and the disambiguation of argument structure. While previous work focused on dataset construction and baseline methods for specific document sections, such as abstract or results, full-text scholarly argumentation mining has seen little progress. In this work, we introduce a sequential pipeline model combining ADUR and ARE for full-text SAM, and provide a first analysis of the performance of pretrained language models (PLMs) on both subtasks. We establish a new SotA for ADUR on the Sci-Arg corpus, outperforming the previous best reported result by a large margin (+7% F1). We also present the first results for ARE, and thus for the full AM pipeline, on this benchmark dataset. Our detailed error analysis reveals that non-contiguous ADUs as well as the interpretation of discourse connectors pose major challenges and that data annotation needs to be more consistent.

Definition Extraction systems are a valuable knowledge source for both humans and algorithms. In this paper we describe our submissions to the DeftEval shared task (SemEval-2020 Task 6), which is evaluated on an English textbook corpus. We provide a detailed explanation of our system for the joint extraction of definition concepts and the relations among them. Furthermore we provide an ablation study of our model variations and describe the results of an error analysis.

We explore to what extent knowledge about the pre-trained language model that is used is beneficial for the task of abstractive summarization. To this end, we experiment with conditioning the encoder and decoder of a Transformer-based neural model on the BERT language model. In addition, we propose a new method of BERT-windowing, which allows chunk-wise processing of texts longer than the BERT window size. We also explore how locality modeling, i.e., the explicit restriction of calculations to the local context, can affect the summarization ability of the Transformer. This is done by introducing 2-dimensional convolutional self-attention into the first layers of the encoder. The results of our models are compared to a baseline and the state-of-the-art models on the CNN/Daily Mail dataset. We additionally train our model on the SwissText dataset to demonstrate usability on German. Both models outperform the baseline in ROUGE scores on two datasets and show its superiority in a manual qualitative analysis.

Despite the recent progress, little is known about the features captured by state-of-the-art neural relation extraction (RE) models. Common methods encode the source sentence, conditioned on the entity mentions, before classifying the relation. However, the complexity of the task makes it difficult to understand how encoder architecture and supporting linguistic knowledge affect the features learned by the encoder. We introduce 14 probing tasks targeting linguistic properties relevant to RE, and we use them to study representations learned by more than 40 different encoder architecture and linguistic feature combinations trained on two datasets, TACRED and SemEval 2010 Task 8. We find that the bias induced by the architecture and the inclusion of linguistic features are clearly expressed in the probing task performance. For example, adding contextualized word representations greatly increases performance on probing tasks with a focus on named entity and part-of-speech information, and yields better results in RE. In contrast, entity masking improves RE, but considerably lowers performance on entity type related probing tasks.

TACRED is one of the largest, most widely used crowdsourced datasets in Relation Extraction (RE). But, even with recent advances in unsupervised pre-training and knowledge enhanced neural RE, models still show a high error rate. In this paper, we investigate the questions: Have we reached a performance ceiling or is there still room for improvement? And how do crowd annotations, dataset, and models contribute to this error rate? To answer these questions, we first validate the most challenging 5K examples in the development and test sets using trained annotators. We find that label errors account for 8% absolute F1 test error, and that more than 50% of the examples need to be relabeled. On the relabeled test set the average F1 score of a large baseline model set improves from 62.1 to 70.1. After validation, we analyze misclassifications on the challenging instances, categorize them into linguistically motivated error groups, and verify the resulting error hypotheses on three state-of-the-art RE models. We show that two groups of ambiguous relations are responsible for most of the remaining errors and that models may adopt shallow heuristics on the dataset when entities are not masked.

In this work, we introduce a bootstrapped, iterative NER model that integrates a PU learning algorithm for recognizing named entities in a low-resource setting. Our approach combines dictionary-based labeling with syntactically-informed label expansion to efficiently enrich the seed dictionaries. Experimental results on a dataset of manually annotated e-commerce product descriptions demonstrate the effectiveness of the proposed framework.

Distantly supervised relation extraction is widely used to extract relational facts from text, but suffers from noisy labels. Current relation extraction methods try to alleviate the noise by multi-instance learning and by providing supporting linguistic and contextual information to more efficiently guide the relation classification. While achieving state-of-the-art results, we observed these models to be biased towards recognizing a limited set of relations with high precision, while ignoring those in the long tail. To address this gap, we utilize a pre-trained language model, the OpenAI Generative Pre-trained Transformer (GPT) (Radford et al., 2018). The GPT and similar models have been shown to capture semantic and syntactic features, and also a notable amount of “common-sense” knowledge, which we hypothesize are important features for recognizing a more diverse set of relations. By extending the GPT to the distantly supervised setting, and fine-tuning it on the NYT10 dataset, we show that it predicts a larger set of distinct relation types with high confidence. Manual and automated evaluation of our model shows that it achieves a state-of-the-art AUC score of 0.422 on the NYT10 dataset, and performs especially well at higher recall levels.

Representations in the hidden layers of Deep Neural Networks (DNN) are often hard to interpret since it is difficult to project them into an interpretable domain. Graph Convolutional Networks (GCN) allow this projection, but existing explainability methods do not exploit this fact, i.e. do not focus their explanations on intermediate states. In this work, we present a novel method that traces and visualizes features that contribute to a classification decision in the visible and hidden layers of a GCN. Our method exposes hidden cross-layer dynamics in the input graph structure. We experimentally demonstrate that it yields meaningful layerwise explanations for a GCN sentence classifier.

Comments on web news contain controversies that manifest as inter-group agreement-conflicts. Tracking such rapidly evolving controversy could ease conflict resolution or journalist-user interaction. However, this presupposes controversy online-prediction that scales to diverse domains using incidental supervision signals instead of manual labeling. To more deeply interpret comment-controversy model decisions we frame prediction as binary classification and evaluate baselines and multi-task CNNs that use an auxiliary news-genre-encoder. Finally, we use ablation and interpretability methods to determine the impacts of topic, discourse and sentiment indicators, contextual vs. global word influence, as well as genre-keywords vs. per-genre-controversy keywords – to find that the models learn plausible controversy features using only incidentally supervised signals.

A huge body of continuously growing written knowledge is available on the web in the form of social media posts, RSS feeds, and news articles. Real-time information extraction from such high velocity, high volume text streams requires scalable, distributed natural language processing pipelines. We introduce such a system for fine-grained event recognition within the big data framework Flink, and demonstrate its capabilities for extracting and geo-locating mobility- and industry-related events from heterogeneous text sources. Performance analyses conducted on several large datasets show that our system achieves high throughput and maintains low latency, which is crucial when events need to be detected and acted upon in real-time. We also present promising experimental results for the event extraction component of our system, which recognizes a novel set of event types. The demo system is available at http://dfki.de/sd4m-sta-demo/.

Web debates play an important role in enabling broad participation of constituencies in social, political and economic decision-taking. However, it is challenging to organize, structure, and navigate a vast number of diverse argumentations and comments collected from many participants over a long time period. In this paper we demonstrate Common Round, a next generation platform for large-scale web debates, which provides functions for eliciting the semantic content and structures from the contributions of participants. In particular, Common Round applies language technologies for the extraction of semantic essence from textual input, aggregation of the formulated opinions and arguments. The platform also provides a cross-lingual access to debates using machine translation.

Recent research shows the importance of linking linguistic knowledge resources for the creation of large-scale linguistic data. We describe our approach for combining two English resources, FrameNet and sar-graphs, and illustrate the benefits of the linked data in a relation extraction setting. While FrameNet consists of schematic representations of situations, linked to lexemes and their valency patterns, sar-graphs are knowledge resources that connect semantic relations from factual knowledge graphs to the linguistic phrases used to express instances of these relations. We analyze the conceptual similarities and differences of both resources and propose to link sar-graphs and FrameNet on the levels of relations/frames as well as phrases. The former alignment involves a manual ontology mapping step, which allows us to extend sar-graphs with new phrase patterns from FrameNet. The phrase-level linking, on the other hand, is fully automatic. We investigate the quality of the automatically constructed links and identify two main classes of errors.

The task of relation extraction is to recognize and extract relations between entities or concepts in texts. Dependency parse trees have become a popular source for discovering extraction patterns, which encode the grammatical relations among the phrases that jointly express relation instances. State-of-the-art weakly supervised approaches to relation extraction typically extract thousands of unique patterns only potentially expressing the target relation. Among these patterns, some are semantically equivalent, but differ in their morphological, lexical-semantic or syntactic form. Some express a relation that entails the target relation. We propose a new approach to structuring extraction patterns by utilizing entailment graphs, hierarchical structures representing entailment relations, and present a novel resource of gold-standard entailment graphs based on a set of patterns automatically acquired using distant supervision. We describe the methodology used for creating the dataset and present statistics of the resource as well as an analysis of inference types underlying the entailment decisions.

Determining the real-world referents for name mentions of persons, organizations and other named entities in texts has become an important task in many information retrieval scenarios and is referred to as Named Entity Disambiguation (NED). While comprehensive datasets support the development and evaluation of NED approaches for English, there are no public datasets to assess NED systems for other languages, such as German. This paper describes the construction of an NED dataset based on a large corpus of German news articles. The dataset is closely modeled on the datasets used for the Knowledge Base Population tasks of the Text Analysis Conference, and contains gold standard annotations for the NED tasks of Entity Linking, NIL Detection and NIL Clustering. We also present first experimental results on the new dataset for each of these tasks in order to establish a baseline for future research efforts.