We present the results and the main findings of SemEval-2024 Task 8: Multigenerator, Multidomain, and Multilingual Machine-Generated Text Detection. The task featured three subtasks. Subtask A is a binary classification task determining whether a text is written by a human or generated by a machine. This subtask has two tracks: a monolingual track focused solely on English texts and a multilingual track. Subtask B is to detect the exact source of a text, discerning whether it is written by a human or generated by a specific LLM. Subtask C aims to identify the changing point within a text, at which the authorship transitions from human to machine. The task attracted a large number of participants: subtask A monolingual (126), subtask A multilingual (59), subtask B (70), and subtask C (30). In this paper, we present the task, analyze the results, and discuss the system submissions and the methods they used. For all subtasks, the best systems used LLMs.

Large language models (LLMs) have demonstrated remarkable capability to generate fluent responses to a wide variety of user queries. However, this has also raised concerns about the potential misuse of such texts in journalism, education, and academia. In this study, we strive to create automated systems that can detect machine-generated texts and pinpoint potential misuse. We first introduce a large-scale benchmark M4, which is a multi-generator, multi-domain, and multi-lingual corpus for machine-generated text detection. Through an extensive empirical study of this dataset, we show that it is challenging for detectors to generalize well on instances from unseen domains or LLMs. In such cases, detectors tend to misclassify machine-generated text as human-written. These results show that the problem is far from solved and that there is a lot of room for improvement. We believe that our dataset will enable future research towards more robust approaches to this pressing societal problem. The dataset is available at https://github.com/mbzuai-nlp/M4

The advent of Large Language Models (LLMs) has brought an unprecedented surge in machine-generated text (MGT) across diverse channels. This raises legitimate concerns about its potential misuse and societal implications. The need to identify and differentiate such content from genuine human-generated text is critical in combating disinformation, preserving the integrity of education and scientific fields, and maintaining trust in communication. In this work, we address this problem by introducing a new benchmark based on a multilingual, multi-domain and multi-generator corpus of MGTs — M4GT-Bench. The benchmark is compiled of three tasks: (1) mono-lingual and multi-lingual binary MGT detection; (2) multi-way detection where one need to identify, which particular model generated the text; and (3) mixed human-machine text detection, where a word boundary delimiting MGT from human-written content should be determined. On the developed benchmark, we have tested several MGT detection baselines and also conducted an evaluation of human performance. We see that obtaining good performance in MGT detection usually requires an access to the training data from the same domain and generators. The benchmark is available at https://github.com/mbzuai-nlp/M4GT-Bench.

We present an approach for medical text coding with SNOMED CT. Our approach uses publicly available linked open data from terminologies and ontologies as training data for the algorithms. We claim that even small training corpora made of short text snippets can be used to train models for the given task. We propose a method based on transformers enhanced with clustering and filtering of the candidates. Further, we adopt a classical machine learning approach - support vector classification (SVC) using transformer embeddings. The resulting approach proves to be more accurate than the predictions given by Large Language Models. We evaluate on a dataset generated from linked open data for SNOMED codes related to morphology and topography for four use cases. Our transformers-based approach achieves an F1-score of 0.82 for morphology and 0.99 for topography codes. Further, we validate the applicability of our approach in a clinical context using labelled real clinical data that are not used for model training.

Event extraction from textual data is a NLP research task relevant to a plethora of domains. Most approaches aim to recognize events from a predefined event schema, consisting of event types and their corresponding arguments. For domains, such as disinformation, where new event types emerge frequently, there is a need to adapt such fixed event schemas to accommodate for new event types. We present NEXT (New Event eXTraction) - a resource-sparse approach to extending a close-domain model to novel event types, that requires a very small number of annotated samples for fine-tuning performed on a single GPU. Furthermore, our results suggest that this approach is suitable not only for extraction of new event types, but also for recognition of existing event types, as the use of this approach on a new dataset leads to improved recall for all existing events while retaining precision.