In this paper, we investigate how different semantic, or content-related, errors made by different types of data-to-text systems differ in terms of number and type. In total, we examine 15 systems: three rule-based and 12 neural systems including two large language models without training or fine-tuning. All systems were tested on the English WebNLG dataset version 3.0. We use a semantic error taxonomy and the brat annotation tool to obtain word-span error annotations on a sample of system outputs. The annotations enable us to establish how many semantic errors different (types of) systems make and what specific types of errors they make, and thus to get an overall understanding of semantic strengths and weaknesses among various types of NLG systems. Among our main findings, we observe that symbolic (rule and template-based) systems make fewer semantic errors overall, non-LLM neural systems have better fluency and data coverage, but make more semantic errors, while LLM-based systems require improvement particularly in addressing superfluous.

Four years on from two papers (Belz et al., 2020; Howcroft et al., 2020) that first called out the lack of standardisation and comparability in the quality criteria assessed in NLP system evaluations, researchers still use widely differing quality criteria names and definitions, meaning that it continues to be unclear when the same aspect of quality is being assessed in two evaluations. While normalised quality criteria were proposed at the time, the list was unwieldy and using it came with a steep learning curve. In this demo paper, our aim is to address these issues with an interactive taxonomy tool that enables quick perusal and selection of the quality criteria, and provides decision support and examples of use at each node.

Wikipedia is known to have systematic gaps in its coverage that correspond to under-resourced languages as well as underrepresented groups. This paper presents a new tool to support efforts to fill in these gaps by automatically generating draft articles and facilitating post-editing and uploading to Wikipedia. A rule-based generator and an input-constrained LLM are used to generate two alternative articles, enabling the often more fluent, but error-prone, LLM-generated article to be content-checked against the more reliable, but less fluent, rule-generated article.

Following numerous calls in the literature for improved practices and standardisation in human evaluation in Natural Language Processing over the past ten years, we held a tutorial on the topic at the 2024 INLG Conference. The tutorial addressed the structure, development, design, implementation, execution and analysis of human evaluations of NLP system quality. Hands-on practical sessions were run, designed to facilitate assimilation of the material presented. Slides, lecture recordings, code and data have been made available on GitHub (https://github.com/Human-Evaluation-Tutorial/INLG-2024-Tutorial). In this paper, we provide summaries of the content of the eight units of the tutorial, alongside its research context and aims.

In this paper, we present our approach to the GEM Shared Task at the INLG’24 Generation Challenges, which focuses on generating data-to-text in multiple languages, including low-resource languages, from WebNLG triples. We employ a combination of end-to-end and pipeline neural architectures for English text generation. To extend our methodology to Hindi, Korean, Arabic, and Swahili, we leverage a neural machine translation model. Our results demonstrate that our approach achieves competitive performance in the given task.

This paper presents the system description of “DCU_NMT’s” submission to the WMT-WAT24 English-to-Low-Resource Multimodal Translation Task. We participated in the English-to-Hindi track, developing both text-only and multimodal neural machine translation (NMT) systems. The text-only systems were trained from scratch on constrained data and augmented with back-translated data. For the multimodal approach, we implemented a context-aware transformer model that integrates visual features as additional contextual information. Specifically, image descriptions generated by an image captioning model were encoded using BERT and concatenated with the textual input.The results indicate that our multimodal system, trained solely on limited data, showed improvements over the text-only baseline in both the challenge and evaluation sets, suggesting the potential benefits of incorporating visual information.

Error analysis aims to provide insights into system errors at different levels of granularity. NLP as a field has a long-standing tradition of analysing and reporting errors which is generally considered good practice. There are existing error taxonomies tailored for different types of NLP task. In this paper, we report our work reviewing existing research on meaning/content error types in generated text, attempt to identify emerging consensus among existing meaning/content error taxonomies, and propose a standardised error taxonomy on this basis. We find that there is virtually complete agreement at the highest taxonomic level where errors of meaning/content divide into (1) Content Omission, (2) Content Addition, and (3) Content Substitution. Consensus in the lower levels is less pronounced, but a compact standardised consensus taxonomy can nevertheless be derived that works across generation tasks and application domains.

In this paper, we present the results of two reproduction studies for the human evaluation originally reported by Dušek and Kasner (2020) in which the authors comparatively evaluated outputs produced by a semantic error detection system for data-to-text generation against reference outputs. In the first reproduction, the original evaluators repeat the evaluation, in a test of the repeatability of the original evaluation. In the second study, two new evaluators carry out the evaluation task, in a test of the reproducibility of the original evaluation under otherwise identical conditions. We describe our approach to reproduction, and present and analyse results, finding different degrees of reproducibility depending on result type, data and labelling task. Our resources are available and open-sourced.

In this paper we describe our reproduction study of the human evaluation of text simplic- ity reported by Nisioi et al. (2017). The work was carried out as part of the ReproGen Shared Task 2022 on Reproducibility of Evaluations in NLG. Our aim was to repeat the evaluation of simplicity for nine automatic text simplification systems with a different set of evaluators. We describe our experimental design together with the known aspects of the original experimental design and present the results from both studies. Pearson correlation between the original and reproduction scores is moderate to high (0.776). Inter-annotator agreement in the reproduction study is lower (0.40) than in the original study (0.66). We discuss challenges arising from the unavailability of certain aspects of the origi- nal set-up, and make several suggestions as to how reproduction of similar evaluations can be made easier in future.

While automatically computing numerical scores remains the dominant paradigm in NLP system evaluation, error analysis is receiving increasing attention, with numerous error annotation schemes being proposed for automatically generated text. However, there is little agreement about what error annotation schemes should look like, how many different types of errors should be distinguished and at what level of granularity. In this paper, our aim is to map out recent work on annotating errors in automatically generated text, with a particular focus on error taxonomies. We describe our systematic paper selection process, and survey the error annotation schemes reported in the papers, drawing out similarities and differences between them. Finally, we characterise the issues that would make it difficult to move from the current situation to a standardised error taxonomy for annotating errors in automatically generated text.

This paper introduces a pretrained word embedding for Manipuri, a low-resourced Indian language. The pretrained word embedding based on FastText is capable of handling the highly agglutinating language Manipuri (mni). We then perform machine translation (MT) experiments using neural network (NN) models. In this paper, we confirm the following observations. Firstly, the reported BLEU score of the Transformer architecture with FastText word embedding model EM-FT performs better than without in all the NMT experiments. Secondly, we observe that adding more training data from a different domain of the test data negatively impacts translation accuracy. The resources reported in this paper are made available in the ELRA catalogue to help the low-resourced languages community with MT/NLP tasks.

In this paper, we introduce a sentence-level comparable text corpus crawled and created for the less-resourced language pair, Manipuri(mni) and English (eng). Our monolingual corpora comprise 1.88 million Manipuri sentences and 1.45 million English sentences, and our parallel corpus comprises 124,975 Manipuri-English sentence pairs. These data were crawled and collected over a year from August 2020 to March 2021 from a local newspaper website called ‘The Sangai Express.’ The resources reported in this paper are made available to help the low-resourced languages community for MT/NLP tasks.

Standard neural machine translation (NMT) allows a model to perform translation between a pair of languages. Multilingual neural machine translation (NMT), on the other hand, allows a model to perform translation between several language pairs, even between language pairs for which no sentences pair has been seen during training (zero-shot translation). This paper presents experiments with zero-shot translation on low resource Indian languages with a very small amount of data for each language pair. We first report results on balanced data over all considered language pairs. We then expand our experiments for additional three rounds by increasing the training data with 2,000 sentence pairs in each round for some of the language pairs. We obtain an increase in translation accuracy with its balanced data settings score multiplied by 7 for Manipuri to Hindi during Round-III of zero-shot translation.