Unsupervised Detection of Contextualized Embedding Bias with Application to Ideology

We propose a fully unsupervised method to detect bias in contextualized embeddings. The method leverages the assortative information latently encoded by social networks and combines orthogonality regularization, structured sparsity learning, and graph neural networks to find the embedding subspace capturing this information. As a concrete example, we focus on the phenomenon of ideological bias: we introduce the concept of an ideological subspace, show how it can be found by applying our method to online discussion forums, and present techniques to probe it. Our experiments suggest that the ideological subspace encodes abstract evaluative semantics and reflects changes in the political left-right spectrum during the presidency of Donald Trump.

Forecasting COVID-19 Caseloads Using Unsupervised Embedding Clusters of Social Media Posts

We present a novel approach incorporating transformer-based language models into infectious disease modelling. Text-derived features are quantified by tracking high-density clusters of sentence-level representations of Reddit posts within specific US states' COVID-19 subreddits. We benchmark these clustered embedding features against features extracted from other high-quality datasets. In a threshold-classification task, we show that they outperform all other feature types at predicting upward trend signals, a significant result for infectious disease modelling in areas where epidemiological data is unreliable. Subsequently, in a time-series forecasting task we fully utilise the predictive power of the caseload and compare the relative strengths of using different supplementary datasets as covariate feature sets in a transformer-based time-series model.

Geographic Adaptation of Pretrained Language Models

Geographic linguistic features are commonly used to improve the performance of pretrained language models (PLMs) on NLP tasks where geographic knowledge is intuitively beneficial (e.g., geolocation prediction and dialect feature prediction). Existing work, however, leverages such geographic information in task-specific fine-tuning, failing to incorporate it into PLMs' geo-linguistic knowledge, which would make it transferable across different tasks. In this work, we introduce an approach to task-agnostic geoadaptation of PLMs that forces the PLM to learn associations between linguistic phenomena and geographic locations. More specifically, geoadaptation is an intermediate training step that couples masked language modeling and geolocation prediction in a dynamic multitask learning setup. In our experiments, we geoadapt BERTi\'c -- a PLM for Bosnian, Croatian, Montenegrin, and Serbian (BCMS) -- using a corpus of geotagged BCMS tweets. Evaluation on three different tasks, namely unsupervised (zero-shot) and supervised geolocation prediction and (unsupervised) prediction of dialect features, shows that our geoadaptation approach is very effective: e.g., we obtain new state-of-the-art performance in supervised geolocation prediction and report massive gains over geographically uninformed PLMs on zero-shot geolocation prediction.

Two Contrasting Data Annotation Paradigms for Subjective NLP Tasks

Labelled data is the foundation of most natural language processing tasks. However, labelling data is difficult and there often are diverse valid beliefs about what the correct data labels should be. So far, dataset creators have acknowledged annotator subjectivity, but not actively managed it in the annotation process. This has led to partly-subjective datasets that fail to serve a clear downstream use. To address this issue, we propose two contrasting paradigms for data annotation. The descriptive paradigm encourages annotator subjectivity, whereas the prescriptive paradigm discourages it. Descriptive annotation allows for the surveying and modelling of different beliefs, whereas prescriptive annotation enables the training of models that consistently apply one belief. We discuss benefits and challenges in implementing both paradigms, and argue that dataset creators should explicitly aim for one or the other to facilitate the intended use of their dataset. Lastly, we design an annotation experiment to illustrate the contrast between the two paradigms.

Modeling Ideological Agenda Setting and Framing in Polarized Online Groups with Graph Neural Networks and Structured Sparsity

The increasing polarization of online political discourse calls for computational tools that are able to automatically detect and monitor ideological divides in social media. Here, we introduce a minimally supervised method that directly leverages the network structure of online discussion forums, specifically Reddit, to detect polarized concepts. We model polarization along the dimensions of agenda setting and framing, drawing upon insights from moral psychology. The architecture we propose combines graph neural networks with structured sparsity and results in representations for concepts and subreddits that capture phenomena such as ideological radicalization and subreddit hijacking. We also create a new dataset of political discourse covering 12 years and more than 600 online groups with different ideologies.

Temporal Adaptation of BERT and Performance on Downstream Document Classification: Insights from Social Media

Language use differs between domains and even within a domain, language use changes over time. Previous work shows that adapting pretrained language models like BERT to domain through continued pretraining improves performance on in-domain downstream tasks. In this article, we investigate whether adapting BERT to time in addition to domain can increase performance even further. For this purpose, we introduce a benchmark corpus of social media comments sampled over three years. The corpus consists of 36.36m unlabelled comments for adaptation and evaluation on an upstream masked language modelling task as well as 0.9m labelled comments for finetuning and evaluation on a downstream document classification task. We find that temporality matters for both tasks: temporal adaptation improves upstream task performance and temporal finetuning improves downstream task performance. However, we do not find clear evidence that adapting BERT to time and domain improves downstream task performance over just adapting to domain. Temporal adaptation captures changes in language use in the downstream task, but not those changes that are actually relevant to performance on it.

Superbizarre Is Not Superb: Improving BERT's Interpretations of Complex Words with Derivational Morphology

How does the input segmentation of pretrained language models (PLMs) affect their generalization capabilities? We present the first study investigating this question, taking BERT as the example PLM and focusing on the semantic representations of derivationally complex words. We show that PLMs can be interpreted as serial dual-route models, i.e., the meanings of complex words are either stored or else need to be computed from the subwords, which implies that maximally meaningful input tokens should allow for the best generalization on new words. This hypothesis is confirmed by a series of semantic probing tasks on which derivational segmentation consistently outperforms BERT's WordPiece segmentation by a large margin. Our results suggest that the generalization capabilities of PLMs could be further improved if a morphologically-informed vocabulary of input tokens were used.

Dynamic Contextualized Word Embeddings

Static word embeddings that represent words by a single vector cannot capture the variability of word meaning in different linguistic and extralinguistic contexts. Building on prior work on contextualized and dynamic word embeddings, we introduce dynamic contextualized word embeddings that represent words as a function of both linguistic and extralinguistic context. Based on a pretrained language model (PLM), dynamic contextualized word embeddings model time and social space jointly, which makes them attractive for various tasks in the computational social sciences. We highlight potential applications by means of qualitative and quantitative analyses.

Generating Derivational Morphology with BERT

Can BERT generate derivationally complex words? We present the first study investigating this question. We find that BERT with a derivational classification layer outperforms an LSTM-based model. Furthermore, our experiments show that the input segmentation crucially impacts BERT's derivational knowledge, both during training and inference.