The Event Causality Identification Shared Task of CASE 2022 involved two subtasks working on the Causal News Corpus. Subtask 1 required participants to predict if a sentence contains a causal relation or not. This is a supervised binary classification task. Subtask 2 required participants to identify the Cause, Effect and Signal spans per causal sentence. This could be seen as a supervised sequence labeling task. For both subtasks, participants uploaded their predictions for a held-out test set, and ranking was done based on binary F1 and macro F1 scores for Subtask 1 and 2, respectively. This paper summarizes the work of the 17 teams that submitted their results to our competition and 12 system description papers that were received. The best F1 scores achieved for Subtask 1 and 2 were 86.19% and 54.15%, respectively. All the top-performing approaches involved pre-trained language models fine-tuned to the targeted task. We further discuss these approaches and analyze errors across participants’ systems in this paper.
Despite the importance of understanding causality, corpora addressing causal relations are limited. There is a discrepancy between existing annotation guidelines of event causality and conventional causality corpora that focus more on linguistics. Many guidelines restrict themselves to include only explicit relations or clause-based arguments. Therefore, we propose an annotation schema for event causality that addresses these concerns. We annotated 3,559 event sentences from protest event news with labels on whether it contains causal relations or not. Our corpus is known as the Causal News Corpus (CNC). A neural network built upon a state-of-the-art pre-trained language model performed well with 81.20% F1 score on test set, and 83.46% in 5-folds cross-validation. CNC is transferable across two external corpora: CausalTimeBank (CTB) and Penn Discourse Treebank (PDTB). Leveraging each of these external datasets for training, we achieved up to approximately 64% F1 on the CNC test set without additional fine-tuning. CNC also served as an effective training and pre-training dataset for the two external corpora. Lastly, we demonstrate the difficulty of our task to the layman in a crowd-sourced annotation exercise. Our annotated corpus is publicly available, providing a valuable resource for causal text mining researchers.
This paper presents the system description of team BLUE for Task A of the CLPsych 2022 Shared Task on identifying changes in mood and behaviour in longitudinal textual data. These moments of change are signals that can be used to screen and prevent suicide attempts.To detect these changes, we experimented with several text representation methods, such as TF-IDF, sentence embeddings, emotion-informed embeddings and several classical machine learning classifiers. We chose to submit three runs of ensemble systems based on maximum voting on the predictions from the best performing models. Of the nine participating teams in Task A, our team ranked second in the Precision-oriented Coverage-based Evaluation, with a score of 0.499. Our best system was an ensemble of Support Vector Machine, Logistic Regression, and Adaptive Boosting classifiers using emotion-informed embeddings as input representation that can model both the linguistic and emotional information found in users? posts.
Benchmarking state-of-the-art text classification and information extraction systems in multilingual, cross-lingual, few-shot, and zero-shot settings for socio-political event information collection is achieved in the scope of the shared task Socio-political and Crisis Events Detection at the workshop CASE @ ACL-IJCNLP 2021. Socio-political event data is utilized for national and international policy- and decision-making. Therefore, the reliability and validity of these datasets are of the utmost importance. We split the shared task into three parts to address the three aspects of data collection (Task 1), fine-grained semantic classification (Task 2), and evaluation (Task 3). Task 1, which is the focus of this report, is on multilingual protest news detection and comprises four subtasks that are document classification (subtask 1), sentence classification (subtask 2), event sentence coreference identification (subtask 3), and event extraction (subtask 4). All subtasks had English, Portuguese, and Spanish for both training and evaluation data. Data in Hindi language was available only for the evaluation of subtask 1. The majority of the submissions, which are 238 in total, are created using multi- and cross-lingual approaches. Best scores are above 77.27 F1-macro for subtask 1, above 85.32 F1-macro for subtask 2, above 84.23 CoNLL 2012 average score for subtask 3, and above 66.20 F1-macro for subtask 4 in all evaluation settings. The performance of the best system for subtask 4 is above 66.20 F1 for all available languages. Although there is still a significant room for improvement in cross-lingual and zero-shot settings, the best submissions for each evaluation scenario yield remarkable results. Monolingual models outperformed the multilingual models in a few evaluation scenarios.
Identifying and extracting reports of medications, their abuse or adverse effects from social media is a challenging task. In social media, relevant reports are very infrequent, causes imbalanced class distribution for machine learning algorithms. Learning algorithms typically designed to optimize the overall accuracy without considering the relative distribution of each class. Thus, imbalanced class distribution is problematic as learning algorithms have low predictive accuracy for the infrequent class. Moreover, social media represents natural linguistic variation in creative language expressions. In this paper, we have used a combination of data balancing and neural language representation techniques to address the challenges. Specifically, we participated the shared tasks 1, 2 (all languages), 4, and 3 (only the span detection, no normalization was attempted) in Social Media Mining for Health applications (SMM4H) 2020 (Klein et al., 2020). The results show that with the proposed methodology recall scores are better than the precision scores for the shared tasks. The recall score is also better compared to the mean score of the total submissions. However, the F1-score is worse than the mean score except for task 2 (French).
We analyse Recurrent Neural Networks (RNNs) to understand the significance of multiple LSTM layers. We argue that the Weighted Finite-state Automata (WFA) trained using a spectral learning algorithm are helpful to analyse RNNs. Our results suggest that multiple LSTM layers in RNNs help learning distributed hidden states, but have a smaller impact on the ability to learn long-term dependencies. The analysis is based on the empirical results, however relevant theory (whenever possible) was discussed to justify and support our conclusions.
Word embeddings are increasingly used for the automatic detection of semantic change; yet, a robust evaluation and systematic comparison of the choices involved has been lacking. We propose a new evaluation framework for semantic change detection and find that (i) using the whole time series is preferable over only comparing between the first and last time points; (ii) independently trained and aligned embeddings perform better than continuously trained embeddings for long time periods; and (iii) that the reference point for comparison matters. We also present an analysis of the changes detected on a large Twitter dataset spanning 5.5 years.