Social biases such as gender or racial biases have been reported in language models (LMs), including Masked Language Models (MLMs). Given that MLMs are continuously trained with increasing amounts of additional data collected over time, an important yet unanswered question is how the social biases encoded with MLMs vary over time. In particular, the number of social media users continues to grow at an exponential rate, and it is a valid concern for the MLMs trained specifically on social media data whether their social biases (if any) would also amplify over time. To empirically analyse this problem, we use a series of MLMs pretrained on chronologically ordered temporal snapshots of corpora. Our analysis reveals that, although social biases are present in all MLMs, most types of social bias remain relatively stable over time (with a few exceptions). To further understand the mechanisms that influence social biases in MLMs, we analyse the temporal corpora used to train the MLMs. Our findings show that some demographic groups, such as male, obtain higher preference over the other, such as female on the training corpora constantly.

In the dynamic realm of social media, diverse topics are discussed daily, transcending linguistic boundaries. However, the complexities of understanding and categorising this content across various languages remain an important challenge with traditional techniques like topic modelling often struggling to accommodate this multilingual diversity. In this paper, we introduce X-Topic, a multilingual dataset featuring content in four distinct languages (English, Spanish, Japanese, and Greek), crafted for the purpose of tweet topic classification. Our dataset includes a wide range of topics, tailored for social media content, making it a valuable resource for scientists and professionals working on cross-linguistic analysis, the development of robust multilingual models, and computational scientists studying online dialogue. Finally, we leverage X-Topic to perform a comprehensive cross-linguistic and multilingual analysis, and compare the capabilities of current general- and domain-specific language models.

The annotation of ambiguous or subjective NLP tasks is usually addressed by various annotators. In most datasets, these annotations are aggregated into a single ground truth. However, this omits divergent opinions of annotators, hence missing individual perspectives. We propose FLEAD (Federated Learning for Exploiting Annotators’ Disagreements), a methodology built upon federated learning to independently learn from the opinions of all the annotators, thereby leveraging all their underlying information without relying on a single ground truth. We conduct an extensive experimental study and analysis in diverse text classification tasks to show the contribution of our approach with respect to mainstream approaches based on majority voting and other recent methodologies that also learn from annotator disagreements.

The ability to compare by analogy, metaphorically or not, lies at the core of how humans understand the world and communicate. In this paper, we study the likelihood of metaphoric outputs, and the capability of a wide range of pretrained transformer-based language models to identify metaphors from other types of analogies, including anomalous ones. In particular, we are interested in discovering whether language models recognise metaphorical analogies equally well as other types of analogies, and whether the model size has an impact on this ability. The results show that there are relevant differences using perplexity as a proxy, with the larger models reducing the gap when it comes to analogical processing, and for distinguishing metaphors from incorrect analogies. This behaviour does not result in increased difficulties for larger generative models in identifying metaphors in comparison to other types of analogies from anomalous sentences in a zero-shot generation setting, when perplexity values of metaphoric and non-metaphoric analogies are similar.

Most hate speech datasets neglect the cultural diversity within a single language, resulting in a critical shortcoming in hate speech detection. To address this, we introduce CREHate, a CRoss-cultural English Hate speech dataset. To construct CREHate, we follow a two-step procedure: 1) cultural post collection and 2) cross-cultural annotation. We sample posts from the SBIC dataset, which predominantly represents North America, and collect posts from four geographically diverse English-speaking countries (Australia, United Kingdom, Singapore, and South Africa) using culturally hateful keywords we retrieve from our survey. Annotations are collected from the four countries plus the United States to establish representative labels for each country. Our analysis highlights statistically significant disparities across countries in hate speech annotations. Only 56.2% of the posts in CREHate achieve consensus among all countries, with the highest pairwise label difference rate of 26%. Qualitative analysis shows that label disagreement occurs mostly due to different interpretations of sarcasm and the personal bias of annotators on divisive topics. Lastly, we evaluate large language models (LLMs) under a zero-shot setting and show that current LLMs tend to show higher accuracies on Anglosphere country labels in CREHate.Our dataset and codes are available at: https://github.com/nlee0212/CREHate

Relations such as “is influenced by”, “is known for” or “is a competitor of” are inherently graded: we can rank entity pairs based on how well they satisfy these relations, but it is hard to draw a line between those pairs that satisfy them and those that do not. Such graded relations play a central role in many applications, yet they are typically not covered by existing Knowledge Graphs. In this paper, we consider the possibility of using Large Language Models (LLMs) to fill this gap. To this end, we introduce a new benchmark, in which entity pairs have to be ranked according to how much they satisfy a given graded relation. The task is formulated as a few-shot ranking problem, where models only have access to a description of the relation and five prototypical instances. We use the proposed benchmark to evaluate state-of-the-art relation embedding strategies as well as several publicly available LLMs and closed conversational models such as GPT-4. We find that smaller language models struggle to outperform a naive baseline. Overall, the best results are obtained with the 11B parameter Flan-T5 model and the 13B parameter OPT model, where further increasing the model size does not seem to be beneficial. For all models, a clear gap with human performance remains.

In machine learning, temporal shifts occur when there are differences between training and test splits in terms of time. For streaming data such as news or social media, models are commonly trained on a fixed corpus from a certain period of time, and they can become obsolete due to the dynamism and evolving nature of online content. This paper focuses on temporal shifts in social media and, in particular, Twitter. We propose a unified evaluation scheme to assess the performance of language models (LMs) under temporal shift on standard social media tasks. LMs are tested on five diverse social media NLP tasks under different temporal settings, which revealed two important findings: (i) the decrease in performance under temporal shift is consistent across different models for entity-focused tasks such as named entity recognition or disambiguation, and hate speech detection, but not significant in the other tasks analysed (i.e., topic and sentiment classification); and (ii) continuous pre-training on the test period does not improve the temporal adaptability of LMs.

Social media is an integral part of the daily life of an increasingly large number of people worldwide. Used for entertainment, communication and news updates, it constitutes a source of information that has been extensively used to study human behaviour. Unfortunately, the open nature of social media platforms along with the difficult task of supervising their content has led to a proliferation of misinformation posts. In this paper, we aim to identify the textual differences between the profiles of user that share misinformation from questionable sources and those that do not. Our goal is to better understand user behaviour in order to be better equipped to combat this issue. To this end, we identify Twitter (X) accounts of potential misinformation spreaders and apply transformer models specialised in social media to extract characteristics such as sentiment, emotion, topic and presence of hate speech. Our results indicate that, while there may be some differences between the behaviour of users that share misinformation and those that do not, there are no large differences when it comes to the type of content shared.

Gender bias has been widely studied by the NLP community. However, other more subtle variations of it, such as mansplaining, have yet received little attention. Mansplaining is a discriminatory behaviour that consists of a condescending treatment or discourse towards women. In this paper, we introduce and analyze Well, actually..., a corpus of 886 mansplaining stories experienced by women. We analyze the corpus in terms of features such as offensiveness, sentiment or misogyny, among others. We also explore to what extent Large Language Models (LLMs) can understand and identify mansplaining and other gender-related microaggressions. Specifically, we experiment with ChatGPT-3.5-Turbo and LLaMA-2 (13b and 70b), with both targeted and open questions. Our findings suggest that, although they can identify mansplaining to some extent, LLMs still struggle to point out this attitude and will even reproduce some of the social patterns behind mansplaining situations, for instance by praising men for giving unsolicited advice to women.

Recent foundational language models have shown state-of-the-art performance in many NLP tasks in zero- and few-shot settings. An advantage of these models over more standard approaches based on fine-tuning is the ability to understand instructions written in natural language (prompts), which helps them generalise better to different tasks and domains without the need for specific training data. This makes them suitable for addressing text classification problems for domains with limited amounts of annotated instances. However, existing research is limited in scale and lacks understanding of how text generation models combined with prompting techniques compare to more established methods for text classification such as fine-tuning masked language models. In this paper, we address this research gap by performing a large-scale evaluation study for 16 text classification datasets covering binary, multiclass, and multilabel problems. In particular, we compare zero- and few-shot approaches of large language models to fine-tuning smaller language models. We also analyse the results by prompt, classification type, domain, and number of labels. In general, the results show how fine-tuning smaller and more efficient language models can still outperform few-shot approaches of larger language models, which have room for improvement when it comes to text classification.

Large Language Models (LLMs) are showing emerging abilities, and one of the latest recognized ones deals with their ability to reason and answer questions from tabular data. Although there are some available datasets to assess question answering systems on tabular data, they are not large and diverse enough to properly assess the capabilities of LLMs. To this end, we propose DataBench, a benchmark composed of 65 real-world datasets over several domains, including 20 human-generated questions per dataset, totaling 1300 questions and answers overall. Using this benchmark, we perform a large-scale empirical comparison of several open and closed source models, including both code-generating and in-context learning models. The results highlight the current gap between open-source and closed-source models, with all types of model having room for improvement even in simple boolean questions or involving a single column.

Entity linking is a well-established task in NLP consisting of associating entity mentions with entries in a knowledge base. Current models have demonstrated competitive performance in standard text settings. However, when it comes to noisy domains such as social media, certain challenges still persist. Typically, to evaluate entity linking on existing benchmarks, a comprehensive knowledge base is necessary and models are expected to possess an understanding of all the entities contained within the knowledge base. However, in practical scenarios where the objective is to retrieve sentences specifically related to a particular entity, strict adherence to a complete understanding of all entities in the knowledge base may not be necessary. To address this gap, we introduce TweetTER (Tweet Target Entity Retrieval), a novel benchmark that aims to bridge the challenges in entity linking. The distinguishing feature of this benchmark is its approach of re-framing entity linking as a binary entity retrieval task. This enables the evaluation of language models’ performance without relying on a conventional knowledge base, providing a more practical and versatile evaluation framework for assessing the effectiveness of language models in entity retrieval tasks.

This paper presents the Visual Word Sense Disambiguation (Visual-WSD) task. The objective of Visual-WSD is to identify among a set of ten images the one that corresponds to the intended meaning of a given ambiguous word which is accompanied with minimal context. The task provides datasets for three different languages: English, Italian, and Farsi.We received a total of 96 different submissions. Out of these, 40 systems outperformed a strong zero-shot CLIP-based baseline. Participating systems proposed different zero- and few-shot approaches, often involving generative models and data augmentation. More information can be found on the task’s website: \url{https://raganato.github.io/vwsd/}.

Generating questions along with associated answers from a text has applications in several domains, such as creating reading comprehension tests for students, or improving document search by providing auxiliary questions and answers based on the query. Training models for question and answer generation (QAG) is not straightforward due to the expected structured output (i.e. a list of question and answer pairs), as it requires more than generating a single sentence. This results in a small number of publicly accessible QAG models. In this paper, we introduce AutoQG, an online service for multilingual QAG along with lmqg, an all-in-one python package for model fine-tuning, generation, and evaluation. We also release QAG models in eight languages fine-tuned on a few variants of pre-trained encoder-decoder language models, which can be used online via AutoQG or locally via lmqg. With these resources, practitioners of any level can benefit from a toolkit that includes a web interface for end users, and easy-to-use code for developers who require custom models or fine-grained controls for generation.

Question and answer generation (QAG) consists of generating a set of question-answer pairs given a context (e.g. a paragraph). This task has a variety of applications, such as data augmentation for question answering (QA) models, information retrieval and education. In this paper, we establish baselines with three different QAG methodologies that leverage sequence-to-sequence language model (LM) fine-tuning. Experiments show that an end-to-end QAG model, which is computationally light at both training and inference times, is generally robust and outperforms other more convoluted approaches. However, there are differences depending on the underlying generative LM. Finally, our analysis shows that QA models fine-tuned solely on generated question-answer pairs can be competitive when compared to supervised QA models trained on human-labeled data.

Despite its relevance, the maturity of NLP for social media pales in comparison with general-purpose models, metrics and benchmarks. This fragmented landscape makes it hard for the community to know, for instance, given a task, which is the best performing model and how it compares with others. To alleviate this issue, we introduce a unified benchmark for NLP evaluation in social media, SuperTweetEval, which includes a heterogeneous set of tasks and datasets combined, adapted and constructed from scratch. We benchmarked the performance of a wide range of models on SuperTweetEval and our results suggest that, despite the recent advances in language modelling, social media remains challenging.

Multilingual language models (LMs) have become a powerful tool in NLP, especially for non-English languages. Nevertheless, model parameters of multilingual LMs remain large due to the larger embedding matrix of the vocabulary covering tokens in different languages. Instead, monolingual LMs can be trained in a target language with the language-specific vocabulary only. In this paper, we propose vocabulary-trimming (VT), a method to reduce a multilingual LM vocabulary to a target language by deleting potentially irrelevant tokens from its vocabulary. In theory, VT can compress any existing multilingual LM to any language covered by the original model. In our experiments, we show that VT can retain the original performance of the multilingual LM, while being considerably smaller in size than the original multilingual LM. The evaluation is performed over four NLP tasks (two generative and two classification tasks) among four widely used multilingual LMs in seven languages. The results show that this methodology can keep the best of both monolingual and multilingual worlds by keeping a small size as monolingual models without the need for specifically retraining them, and can even help limit potentially harmful social biases.

Metaphor identification aims at understanding whether a given expression is used figuratively in context. However, in this paper we show how existing metaphor identification datasets can be gamed by fully ignoring the potential metaphorical expression or the context in which it occurs. We test this hypothesis in a variety of datasets and settings, and show that metaphor identification systems based on language models without complete information can be competitive with those using the full context. This is due to the construction procedures to build such datasets, which introduce unwanted biases for positive and negative classes. Finally, we test the same hypothesis on datasets that are carefully sampled from natural corpora and where this bias is not present, making these datasets more challenging and reliable.

Various types of social biases have been reported with pretrained Masked Language Models (MLMs) in prior work. However, multiple underlying factors are associated with an MLM such as its model size, size of the training data, training objectives, the domain from which pretraining data is sampled, tokenization, and languages present in the pretrained corpora, to name a few. It remains unclear as to which of those factors influence social biases that are learned by MLMs. To study the relationship between model factors and the social biases learned by an MLM, as well as the downstream task performance of the model, we conduct a comprehensive study over 39 pretrained MLMs covering different model sizes, training objectives, tokenization methods, training data domains and languages. Our results shed light on important factors often neglected in prior literature, such as tokenization or model objectives.

The automatic detection of hate speech online is an active research area in NLP. Most of the studies to date are based on social media datasets that contribute to the creation of hate speech detection models trained on them. However, data creation processes contain their own biases, and models inherently learn from these dataset-specific biases. In this paper, we perform a large-scale cross-dataset comparison where we fine-tune language models on different hate speech detection datasets. This analysis shows how some datasets are more generalizable than others when used as training data. Crucially, our experiments show how combining hate speech detection datasets can contribute to the development of robust hate speech detection models. This robustness holds even when controlling by data size and compared with the best individual datasets.

Despite its importance, the time variable has been largely neglected in the NLP and language model literature. In this paper, we present TimeLMs, a set of language models specialized on diachronic Twitter data. We show that a continual learning strategy contributes to enhancing Twitter-based language models’ capacity to deal with future and out-of-distribution tweets, while making them competitive with standardized and more monolithic benchmarks. We also perform a number of qualitative analyses showing how they cope with trends and peaks in activity involving specific named entities or concept drift. TimeLMs is available at github.com/cardiffnlp/timelms.

Powerful generative models have led to recent progress in question generation (QG). However, it is difficult to measure advances in QG research since there are no standardized resources that allow a uniform comparison among approaches. In this paper, we introduce QG-Bench, a multilingual and multidomain benchmark for QG that unifies existing question answering datasets by converting them to a standard QG setting. It includes general-purpose datasets such as SQuAD for English, datasets from ten domains and two styles, as well as datasets in eight different languages. Using QG-Bench as a reference, we perform an extensive analysis of the capabilities of language models for the task. First, we propose robust QG baselines based on fine-tuning generative language models. Then, we complement automatic evaluation based on standard metrics with an extensive manual evaluation, which in turn sheds light on the difficulty of evaluating QG models. Finally, we analyse both the domain adaptability of these models as well as the effectiveness of multilingual models in languages other than English.QG-Bench is released along with the fine-tuned models presented in the paper (https://github.com/asahi417/lm-question-generation), which are also available as a demo (https://autoqg.net/).

In this paper we present TweetNLP, an integrated platform for Natural Language Processing (NLP) in social media. TweetNLP supports a diverse set of NLP tasks, including generic focus areas such as sentiment analysis and named entity recognition, as well as social media-specific tasks such as emoji prediction and offensive language identification. Task-specific systems are powered by reasonably-sized Transformer-based language models specialized on social media text (in particular, Twitter) which can be run without the need for dedicated hardware or cloud services. The main contributions of TweetNLP are: (1) an integrated Python library for a modern toolkit supporting social media analysis using our various task-specific models adapted to the social domain; (2) an interactive online demo for codeless experimentation using our models; and (3) a tutorial covering a wide variety of typical social media applications.

Data augmentation techniques are widely used for enhancing the performance of machine learning models by tackling class imbalance issues and data sparsity. State-of-the-art generative language models have been shown to provide significant gains across different NLP tasks. However, their applicability to data augmentation for text classification tasks in few-shot settings have not been fully explored, especially for specialised domains. In this paper, we leverage GPT-2 (Radford et al, 2019) for generating artificial training instances in order to improve classification performance. Our aim is to analyse the impact the selection process of seed training examples has over the quality of GPT-generated samples and consequently the classifier performance. We propose a human-in-the-loop approach for selecting seed samples. Further, we compare the approach to other seed selection strategies that exploit the characteristics of specialised domains such as human-created class hierarchical structure and the presence of noun phrases. Our results show that fine-tuning GPT-2 in a handful of label instances leads to consistent classification improvements and outperform competitive baselines. The seed selection strategies developed in this work lead to significant improvements over random seed selection for specialised domains. We show that guiding text generation through domain expert selection can lead to further improvements, which opens up interesting research avenues for combining generative models and active learning.

Understanding relational knowledge plays an integral part in natural language comprehension. When it comes to pre-trained language models (PLM), prior work has been focusing on probing relational knowledge this by filling the blanks in pre-defined prompts such as “The capital of France is —". However, these probes may be affected by the co-occurrence of target relation words and entities (e.g. “capital”, “France” and “Paris”) in the pre-training corpus. In this work, we extend these probing methodologies leveraging analogical proportions as a proxy to probe relational knowledge in transformer-based PLMs without directly presenting the desired relation. In particular, we analysed the ability of PLMs to understand (1) the directionality of a given relation (e.g. Paris-France is not the same as France-Paris); (2) the ability to distinguish types on a given relation (both France and Japan are countries); and (3) the relation itself (Paris is the capital of France, but not Rome). Our results show how PLMs are extremely accurate at (1) and (2), but have clear room for improvement for (3). To better understand the reasons behind this behaviour and mistakes made by PLMs, we provide an extended quantitative analysis based on relevant factors such as frequency.

The increase in performance in NLP due to the prevalence of distributional models and deep learning has brought with it a reciprocal decrease in interpretability. This has spurred a focus on what neural networks learn about natural language with less of a focus on how. Some work has focused on the data used to develop data-driven models, but typically this line of work aims to highlight issues with the data, e.g. highlighting and offsetting harmful biases. This work contributes to the relatively untrodden path of what is required in data for models to capture meaningful representations of natural language. This is entails evaluating how well English and Spanish semantic spaces capture a particular type of relational knowledge, namely the traits associated with concepts (e.g. bananas-yellow), and exploring the role of co-occurrences in this context.

Language models are ubiquitous in current NLP, and their multilingual capacity has recently attracted considerable attention. However, current analyses have almost exclusively focused on (multilingual variants of) standard benchmarks, and have relied on clean pre-training and task-specific corpora as multilingual signals. In this paper, we introduce XLM-T, a model to train and evaluate multilingual language models in Twitter. In this paper we provide: (1) a new strong multilingual baseline consisting of an XLM-R (Conneau et al. 2020) model pre-trained on millions of tweets in over thirty languages, alongside starter code to subsequently fine-tune on a target task; and (2) a set of unified sentiment analysis Twitter datasets in eight different languages and a XLM-T model trained on this dataset.

Recent progress in language model pre-training has led to important improvements in Named Entity Recognition (NER). Nonetheless, this progress has been mainly tested in well-formatted documents such as news, Wikipedia, or scientific articles. In social media the landscape is different, in which it adds another layer of complexity due to its noisy and dynamic nature. In this paper, we focus on NER in Twitter, one of the largest social media platforms, and construct a new NER dataset, TweetNER7, which contains seven entity types annotated over 11,382 tweets from September 2019 to August 2021. The dataset was constructed by carefully distributing the tweets over time and taking representative trends as a basis. Along with the dataset, we provide a set of language model baselines and perform an analysis on the language model performance on the task, especially analyzing the impact of different time periods. In particular, we focus on three important temporal aspects in our analysis: short-term degradation of NER models over time, strategies to fine-tune a language model over different periods, and self-labeling as an alternative to lack of recently-labeled data. TweetNER7 is released publicly (https://huggingface.co/datasets/tner/tweetner7) along with the models fine-tuned on it (NER models have been integrated into TweetNLP and can be found at https://github.com/asahi417/tner/tree/master/examples/tweetner7_paper).

This paper describes the experiments ran for SemEval-2022 Task 2, subtask A, zero-shot and one-shot settings for idiomaticity detection. Our main approach is based on fine-tuning transformer-based language models as a baseline to perform binary classification. Our system, CardiffNLP-Metaphor, ranked 8th and 7th (respectively on zero- and one-shot settings on this task. Our main contribution lies in the extensive evaluation of transformer-based language models and various configurations, showing, among others, the potential of large multilingual models over base monolingual models. Moreover, we analyse the impact of various input parameters, which offer interesting insights on how language models work in practice.

Language evolves over time, and word meaning changes accordingly. This is especially true in social media, since its dynamic nature leads to faster semantic shifts, making it challenging for NLP models to deal with new content and trends. However, the number of datasets and models that specifically address the dynamic nature of these social platforms is scarce. To bridge this gap, we present TempoWiC, a new benchmark especially aimed at accelerating research in social media-based meaning shift. Our results show that TempoWiC is a challenging benchmark, even for recently-released language models specialized in social media.

Social media platforms host discussions about a wide variety of topics that arise everyday. Making sense of all the content and organising it into categories is an arduous task. A common way to deal with this issue is relying on topic modeling, but topics discovered using this technique are difficult to interpret and can differ from corpus to corpus. In this paper, we present a new task based on tweet topic classification and release two associated datasets. Given a wide range of topics covering the most important discussion points in social media, we provide training and testing data from recent time periods that can be used to evaluate tweet classification models. Moreover, we perform a quantitative evaluation and analysis of current general- and domain-specific language models on the task, which provide more insights on the challenges and nature of the task.

In this paper we analyze the extent to which contextualized sense embeddings, i.e., sense embeddings that are computed based on contextualized word embeddings, are transferable across languages. To this end, we compiled a unified cross-lingual benchmark for Word Sense Disambiguation. We then propose two simple strategies to transfer sense-specific knowledge across languages and test them on the benchmark. Experimental results show that this contextualized knowledge can be effectively transferred to similar languages through pre-trained multilingual language models, to the extent that they can out-perform monolingual representations learnednfrom existing language-specific data.

Analogies play a central role in human commonsense reasoning. The ability to recognize analogies such as “eye is to seeing what ear is to hearing”, sometimes referred to as analogical proportions, shape how we structure knowledge and understand language. Surprisingly, however, the task of identifying such analogies has not yet received much attention in the language model era. In this paper, we analyze the capabilities of transformer-based language models on this unsupervised task, using benchmarks obtained from educational settings, as well as more commonly used datasets. We find that off-the-shelf language models can identify analogies to a certain extent, but struggle with abstract and complex relations, and results are highly sensitive to model architecture and hyperparameters. Overall the best results were obtained with GPT-2 and RoBERTa, while configurations using BERT were not able to outperform word embedding models. Our results raise important questions for future work about how, and to what extent, pre-trained language models capture knowledge about abstract semantic relations.

Social media is often used by individuals and organisations as a platform to spread misinformation. With the recent coronavirus pandemic we have seen a surge of misinformation on Twitter, posing a danger to public health. In this paper, we compile a large COVID-19 Twitter misinformation corpus and perform an analysis to discover patterns with respect to vocabulary usage. Among others, our analysis reveals that the variety of topics and vocabulary usage are considerably more limited and negative in tweets related to misinformation than in randomly extracted tweets. In addition to our qualitative analysis, our experimental results show that a simple linear model based only on lexical features is effective in identifying misinformation-related tweets (with accuracy over 80%), providing evidence to the fact that the vocabulary used in misinformation largely differs from generic tweets.

Transformer-based language models have taken many fields in NLP by storm. BERT and its derivatives dominate most of the existing evaluation benchmarks, including those for Word Sense Disambiguation (WSD), thanks to their ability in capturing context-sensitive semantic nuances. However, there is still little knowledge about their capabilities and potential limitations in encoding and recovering word senses. In this article, we provide an in-depth quantitative and qualitative analysis of the celebrated BERT model with respect to lexical ambiguity. One of the main conclusions of our analysis is that BERT can accurately capture high-level sense distinctions, even when a limited number of examples is available for each word sense. Our analysis also reveals that in some cases language models come close to solving coarse-grained noun disambiguation under ideal conditions in terms of availability of training data and computing resources. However, this scenario rarely occurs in real-world settings and, hence, many practical challenges remain even in the coarse-grained setting. We also perform an in-depth comparison of the two main language model-based WSD strategies, namely, fine-tuning and feature extraction, finding that the latter approach is more robust with respect to sense bias and it can better exploit limited available training data. In fact, the simple feature extraction strategy of averaging contextualized embeddings proves robust even using only three training sentences per word sense, with minimal improvements obtained by increasing the size of this training data.

We present WiC-TSV, a new multi-domain evaluation benchmark for Word Sense Disambiguation. More specifically, we introduce a framework for Target Sense Verification of Words in Context which grounds its uniqueness in the formulation as binary classification task thus being independent of external sense inventories, and the coverage of various domains. This makes the dataset highly flexible for the evaluation of a diverse set of models and systems in and across domains. WiC-TSV provides three different evaluation settings, depending on the input signals provided to the model. We set baseline performance on the dataset using state-of-the-art language models. Experimental results show that even though these models can perform decently on the task, there remains a gap between machine and human performance, especially in out-of-domain settings. WiC-TSV data is available at https://competitions.codalab.org/competitions/23683.

Language model (LM) pretraining has led to consistent improvements in many NLP downstream tasks, including named entity recognition (NER). In this paper, we present T-NER (Transformer-based Named Entity Recognition), a Python library for NER LM finetuning. In addition to its practical utility, T-NER facilitates the study and investigation of the cross-domain and cross-lingual generalization ability of LMs finetuned on NER. Our library also provides a web app where users can get model predictions interactively for arbitrary text, which facilitates qualitative model evaluation for non-expert programmers. We show the potential of the library by compiling nine public NER datasets into a unified format and evaluating the cross-domain and cross- lingual performance across the datasets. The results from our initial experiments show that in-domain performance is generally competitive across datasets. However, cross-domain generalization is challenging even with a large pretrained LM, which has nevertheless capacity to learn domain-specific features if fine- tuned on a combined dataset. To facilitate future research, we also release all our LM checkpoints via the Hugging Face model hub.

Term weighting schemes are widely used in Natural Language Processing and Information Retrieval. In particular, term weighting is the basis for keyword extraction. However, there are relatively few evaluation studies that shed light about the strengths and shortcomings of each weighting scheme. In fact, in most cases researchers and practitioners resort to the well-known tf-idf as default, despite the existence of other suitable alternatives, including graph-based models. In this paper, we perform an exhaustive and large-scale empirical comparison of both statistical and graph-based term weighting methods in the context of keyword extraction. Our analysis reveals some interesting findings such as the advantages of the less-known lexical specificity with respect to tf-idf, or the qualitative differences between statistical and graph-based methods. Finally, based on our findings we discuss and devise some suggestions for practitioners. Source code to reproduce our experimental results, including a keyword extraction library, are available in the following repository: https://github.com/asahi417/kex

Pre-trained language models have been found to capture a surprisingly rich amount of lexical knowledge, ranging from commonsense properties of everyday concepts to detailed factual knowledge about named entities. Among others, this makes it possible to distill high-quality word vectors from pre-trained language models. However, it is currently unclear to what extent it is possible to distill relation embeddings, i.e. vectors that characterize the relationship between two words. Such relation embeddings are appealing because they can, in principle, encode relational knowledge in a more fine-grained way than is possible with knowledge graphs. To obtain relation embeddings from a pre-trained language model, we encode word pairs using a (manually or automatically generated) prompt, and we fine-tune the language model such that relationally similar word pairs yield similar output vectors. We find that the resulting relation embeddings are highly competitive on analogy (unsupervised) and relation classification (supervised) benchmarks, even without any task-specific fine-tuning. Source code to reproduce our experimental results and the model checkpoints are available in the following repository: https://github.com/asahi417/relbert

Pre-trained language models provide the foundations for state-of-the-art performance across a wide range of natural language processing tasks, including text classification. However, most classification datasets assume a large amount labeled data, which is commonly not the case in practical settings. In particular, in this paper we compare the performance of a light-weight linear classifier based on word embeddings, i.e., fastText (Joulin et al., 2017), versus a pre-trained language model, i.e., BERT (Devlin et al., 2019), across a wide range of datasets and classification tasks. In general, results show the importance of domain-specific unlabeled data, both in the form of word embeddings or language models. As for the comparison, BERT outperforms all baselines in standard datasets with large training sets. However, in settings with small training datasets a simple method like fastText coupled with domain-specific word embeddings performs equally well or better than BERT, even when pre-trained on domain-specific data.

Embeddings have been one of the most important topics of interest in NLP for the past decade. Representing knowledge through a low-dimensional vector which is easily integrable in modern machine learning models has played a central role in the development of the field. Embedding techniques initially focused on words but the attention soon started to shift to other forms. This tutorial will provide a high-level synthesis of the main embedding techniques in NLP, in the broad sense. We will start by conventional word embeddings (e.g., Word2Vec and GloVe) and then move to other types of embeddings, such as sense-specific and graph alternatives. We will finalize with an overview of the trending contextualized representations (e.g., ELMo and BERT) and explain their potential and impact in NLP.

Cross-lingual word embeddings are vector representations of words in different languages where words with similar meaning are represented by similar vectors, regardless of the language. Recent developments which construct these embeddings by aligning monolingual spaces have shown that accurate alignments can be obtained with little or no supervision, which usually comes in the form of bilingual dictionaries. However, the focus has been on a particular controlled scenario for evaluation, and there is no strong evidence on how current state-of-the-art systems would fare with noisy text or for language pairs with major linguistic differences. In this paper we present an extensive evaluation over multiple cross-lingual embedding models, analyzing their strengths and limitations with respect to different variables such as target language, training corpora and amount of supervision. Our conclusions put in doubt the view that high-quality cross-lingual embeddings can always be learned without much supervision.

Large sense-annotated datasets are increasingly necessary for training deep supervised systems in Word Sense Disambiguation. However, gathering high-quality sense-annotated data for as many instances as possible is a laborious and expensive task. This has led to the proliferation of automatic and semi-automatic methods for overcoming the so-called knowledge-acquisition bottleneck. In this short survey we present an overview of sense-annotated corpora, annotated either manually- or (semi)automatically, that are currently available for different languages and featuring distinct lexical resources as inventory of senses, i.e. WordNet, Wikipedia, BabelNet. Furthermore, we provide the reader with general statistics of each dataset and an analysis of their specific features.

Textual definitions constitute a fundamental source of knowledge when seeking the meaning of words, and they are the cornerstone of lexical resources like glossaries, dictionaries, encyclopedia or thesauri. In this paper, we present an in-depth analytical study on the main features relevant to the task of definition extraction. Our main goal is to study whether linguistic structures from canonical (the Aristotelian or genus et differentia model) can be leveraged to retrieve definitions from corpora in different domains of knowledge and textual genres alike. To this end, we develop a simple linear classifier and analyze the contribution of several (sets of) linguistic features. Finally, as a result of our experiments, we also shed light on the particularities of existing benchmarks as well as the most challenging aspects of the task.

Semantic relations are core to how humans understand and express concepts in the real world using language. Recently, there has been a thread of research aimed at modeling these relations by learning vector representations from text corpora. Most of these approaches focus strictly on leveraging the co-occurrences of relationship word pairs within sentences. In this paper, we investigate the hypothesis that examples of a lexical relation in a corpus are fundamental to a neural word embedding’s ability to complete analogies involving the relation. Our experiments, in which we remove all known examples of a relation from training corpora, show only marginal degradation in analogy completion performance involving the removed relation. This finding enhances our understanding of neural word embeddings, showing that co-occurrence information of a particular semantic relation is not the main source of their structural regularity.

Depression and anxiety are psychiatric disorders that are observed in many areas of everyday life. For example, these disorders manifest themselves somewhat frequently in texts written by nondiagnosed users in social media. However, detecting users with these conditions is not a straightforward task as they may not explicitly talk about their mental state, and if they do, contextual cues such as immediacy must be taken into account. When available, linguistic flags pointing to probable anxiety or depression could be used by medical experts to write better guidelines and treatments. In this paper, we develop a dataset designed to foster research in depression and anxiety detection in Twitter, framing the detection task as a binary tweet classification problem. We then apply state-of-the-art classification models to this dataset, providing a competitive set of baselines alongside qualitative error analysis. Our results show that language models perform reasonably well, and better than more traditional baselines. Nonetheless, there is clear room for improvement, particularly with unbalanced training sets and in cases where seemingly obvious linguistic cues (keywords) are used counter-intuitively.

The experimental landscape in natural language processing for social media is too fragmented. Each year, new shared tasks and datasets are proposed, ranging from classics like sentiment analysis to irony detection or emoji prediction. Therefore, it is unclear what the current state of the art is, as there is no standardized evaluation protocol, neither a strong set of baselines trained on such domain-specific data. In this paper, we propose a new evaluation framework (TweetEval) consisting of seven heterogeneous Twitter-specific classification tasks. We also provide a strong set of baselines as starting point, and compare different language modeling pre-training strategies. Our initial experiments show the effectiveness of starting off with existing pre-trained generic language models, and continue training them on Twitter corpora.

State-of-the-art methods for Word Sense Disambiguation (WSD) combine two different features: the power of pre-trained language models and a propagation method to extend the coverage of such models. This propagation is needed as current sense-annotated corpora lack coverage of many instances in the underlying sense inventory (usually WordNet). At the same time, unambiguous words make for a large portion of all words in WordNet, while being poorly covered in existing sense-annotated corpora. In this paper, we propose a simple method to provide annotations for most unambiguous words in a large corpus. We introduce the UWA (Unambiguous Word Annotations) dataset and show how a state-of-the-art propagation-based model can use it to extend the coverage and quality of its word sense embeddings by a significant margin, improving on its original results on WSD.

The ability to correctly model distinct meanings of a word is crucial for the effectiveness of semantic representation techniques. However, most existing evaluation benchmarks for assessing this criterion are tied to sense inventories (usually WordNet), restricting their usage to a small subset of knowledge-based representation techniques. The Word-in-Context dataset (WiC) addresses the dependence on sense inventories by reformulating the standard disambiguation task as a binary classification problem; but, it is limited to the English language. We put forward a large multilingual benchmark, XL-WiC, featuring gold standards in 12 new languages from varied language families and with different degrees of resource availability, opening room for evaluation scenarios such as zero-shot cross-lingual transfer. We perform a series of experiments to determine the reliability of the datasets and to set performance baselines for several recent contextualized multilingual models. Experimental results show that even when no tagged instances are available for a target language, models trained solely on the English data can attain competitive performance in the task of distinguishing different meanings of a word, even for distant languages. XL-WiC is available at https://pilehvar.github.io/xlwic/.

This paper describes the system developed at the University of Alicante (UA) for the SemEval 2019 Task 5: Shared Task on Multilingual Detection of Hate. The purpose of this work is to build a strong baseline for hate speech detection, using a traditional machine learning approach with standard textual features, which could serve in a near future as a reference to compare with deep learning systems. We participated in both task A (Hate Speech Detection against Immigrants and Women) and task B (Aggressive behavior and Target Classification). Despite its simplicity, our system obtained a remarkable F1-score of 72.5 (sixth highest) and an accuracy of 73.6 (second highest) in Spanish (task A), outperforming more complex neural models from a total of 40 participant systems.

While word embeddings have been shown to implicitly encode various forms of attributional knowledge, the extent to which they capture relational information is far more limited. In previous work, this limitation has been addressed by incorporating relational knowledge from external knowledge bases when learning the word embedding. Such strategies may not be optimal, however, as they are limited by the coverage of available resources and conflate similarity with other forms of relatedness. As an alternative, in this paper we propose to encode relational knowledge in a separate word embedding, which is aimed to be complementary to a given standard word embedding. This relational word embedding is still learned from co-occurrence statistics, and can thus be used even when no external knowledge base is available. Our analysis shows that relational word vectors do indeed capture information that is complementary to what is encoded in standard word embeddings.

By design, word embeddings are unable to model the dynamic nature of words’ semantics, i.e., the property of words to correspond to potentially different meanings. To address this limitation, dozens of specialized meaning representation techniques such as sense or contextualized embeddings have been proposed. However, despite the popularity of research on this topic, very few evaluation benchmarks exist that specifically focus on the dynamic semantics of words. In this paper we show that existing models have surpassed the performance ceiling of the standard evaluation dataset for the purpose, i.e., Stanford Contextual Word Similarity, and highlight its shortcomings. To address the lack of a suitable benchmark, we put forward a large-scale Word in Context dataset, called WiC, based on annotations curated by experts, for generic evaluation of context-sensitive representations. WiC is released in https://pilehvar.github.io/wic/.

Incorporating linguistic, world and common sense knowledge into AI/NLP systems is currently an important research area, with several open problems and challenges. At the same time, processing and storing this knowledge in lexical resources is not a straightforward task. We propose to address these complementary goals from two methodological perspectives: the use of NLP methods to help the process of constructing and enriching lexical resources and the use of lexical resources for improving NLP applications. This tutorial may be useful for two main types of audience: those working on language resources who are interested in becoming acquainted with automatic NLP techniques, with the end goal of speeding and/or easing up the process of resource curation; and on the other hand, researchers in NLP who would like to benefit from the knowledge of lexical resources to improve their systems and models.

Text preprocessing is often the first step in the pipeline of a Natural Language Processing (NLP) system, with potential impact in its final performance. Despite its importance, text preprocessing has not received much attention in the deep learning literature. In this paper we investigate the impact of simple text preprocessing decisions (particularly tokenizing, lemmatizing, lowercasing and multiword grouping) on the performance of a standard neural text classifier. We perform an extensive evaluation on standard benchmarks from text categorization and sentiment analysis. While our experiments show that a simple tokenization of input text is generally adequate, they also highlight significant degrees of variability across preprocessing techniques. This reveals the importance of paying attention to this usually-overlooked step in the pipeline, particularly when comparing different models. Finally, our evaluation provides insights into the best preprocessing practices for training word embeddings.

Cross-lingual word embeddings are becoming increasingly important in multilingual NLP. Recently, it has been shown that these embeddings can be effectively learned by aligning two disjoint monolingual vector spaces through linear transformations, using no more than a small bilingual dictionary as supervision. In this work, we propose to apply an additional transformation after the initial alignment step, which moves cross-lingual synonyms towards a middle point between them. By applying this transformation our aim is to obtain a better cross-lingual integration of the vector spaces. In addition, and perhaps surprisingly, the monolingual spaces also improve by this transformation. This is in contrast to the original alignment, which is typically learned such that the structure of the monolingual spaces is preserved. Our experiments confirm that the resulting cross-lingual embeddings outperform state-of-the-art models in both monolingual and cross-lingual evaluation tasks.

Human language has evolved towards newer forms of communication such as social media, where emojis (i.e., ideograms bearing a visual meaning) play a key role. While there is an increasing body of work aimed at the computational modeling of emoji semantics, there is currently little understanding about what makes a computational model represent or predict a given emoji in a certain way. In this paper we propose a label-wise attention mechanism with which we attempt to better understand the nuances underlying emoji prediction. In addition to advantages in terms of interpretability, we show that our proposed architecture improves over standard baselines in emoji prediction, and does particularly well when predicting infrequent emojis.

This paper describes the results of the first Shared Task on Multilingual Emoji Prediction, organized as part of SemEval 2018. Given the text of a tweet, the task consists of predicting the most likely emoji to be used along such tweet. Two subtasks were proposed, one for English and one for Spanish, and participants were allowed to submit a system run to one or both subtasks. In total, 49 teams participated to the English subtask and 22 teams submitted a system run to the Spanish subtask. Evaluation was carried out emoji-wise, and the final ranking was based on macro F-Score. Data and further information about this task can be found at https://competitions.codalab.org/competitions/17344.

This paper describes the SemEval 2018 Shared Task on Hypernym Discovery. We put forward this task as a complementary benchmark for modeling hypernymy, a problem which has traditionally been cast as a binary classification task, taking a pair of candidate words as input. Instead, our reformulated task is defined as follows: given an input term, retrieve (or discover) its suitable hypernyms from a target corpus. We proposed five different subtasks covering three languages (English, Spanish, and Italian), and two specific domains of knowledge in English (Medical and Music). Participants were allowed to compete in any or all of the subtasks. Overall, a total of 11 teams participated, with a total of 39 different systems submitted through all subtasks. Data, results and further information about the task can be found at https://competitions.codalab.org/competitions/17119.

In this paper we analyze the use of emojis in social media with respect to gender and skin tone. By gathering a dataset of over twenty two million tweets from United States some findings are clearly highlighted after performing a simple frequency-based analysis. Moreover, we carry out a semantic analysis on the usage of emojis and their modifiers (e.g. gender and skin tone) by embedding all words, emojis and modifiers into the same vector space. Our analyses reveal that some stereotypes related to the skin color and gender seem to be reflected on the use of these modifiers. For example, emojis representing hand gestures are more widely utilized with lighter skin tones, and the usage across skin tones differs significantly. At the same time, the vector corresponding to the male modifier tends to be semantically close to emojis related to business or technology, whereas their female counterparts appear closer to emojis about love or makeup.

Word embeddings are widely used in Natural Language Processing, mainly due to their success in capturing semantic information from massive corpora. However, their creation process does not allow the different meanings of a word to be automatically separated, as it conflates them into a single vector. We address this issue by proposing a new model which learns word and sense embeddings jointly. Our model exploits large corpora and knowledge from semantic networks in order to produce a unified vector space of word and sense embeddings. We evaluate the main features of our approach both qualitatively and quantitatively in a variety of tasks, highlighting the advantages of the proposed method in comparison to state-of-the-art word- and sense-based models.

This paper introduces a new task on Multilingual and Cross-lingual SemanticThis paper introduces a new task on Multilingual and Cross-lingual Semantic Word Similarity which measures the semantic similarity of word pairs within and across five languages: English, Farsi, German, Italian and Spanish. High quality datasets were manually curated for the five languages with high inter-annotator agreements (consistently in the 0.9 ballpark). These were used for semi-automatic construction of ten cross-lingual datasets. 17 teams participated in the task, submitting 24 systems in subtask 1 and 14 systems in subtask 2. Results show that systems that combine statistical knowledge from text corpora, in the form of word embeddings, and external knowledge from lexical resources are best performers in both subtasks. More information can be found on the task website: http://alt.qcri.org/semeval2017/task2/

Word Sense Disambiguation is a long-standing task in Natural Language Processing, lying at the core of human language understanding. However, the evaluation of automatic systems has been problematic, mainly due to the lack of a reliable evaluation framework. In this paper we develop a unified evaluation framework and analyze the performance of various Word Sense Disambiguation systems in a fair setup. The results show that supervised systems clearly outperform knowledge-based models. Among the supervised systems, a linear classifier trained on conventional local features still proves to be a hard baseline to beat. Nonetheless, recent approaches exploiting neural networks on unlabeled corpora achieve promising results, surpassing this hard baseline in most test sets.

In this paper we present BabelDomains, a unified resource which provides lexical items with information about domains of knowledge. We propose an automatic method that uses knowledge from various lexical resources, exploiting both distributional and graph-based clues, to accurately propagate domain information. We evaluate our methodology intrinsically on two lexical resources (WordNet and BabelNet), achieving a precision over 80% in both cases. Finally, we show the potential of BabelDomains in a supervised learning setting, clustering training data by domain for hypernym discovery.

Lexical ambiguity can impede NLP systems from accurate understanding of semantics. Despite its potential benefits, the integration of sense-level information into NLP systems has remained understudied. By incorporating a novel disambiguation algorithm into a state-of-the-art classification model, we create a pipeline to integrate sense-level information into downstream NLP applications. We show that a simple disambiguation of the input text can lead to consistent performance improvement on multiple topic categorization and polarity detection datasets, particularly when the fine granularity of the underlying sense inventory is reduced and the document is sufficiently large. Our results also point to the need for sense representation research to focus more on in vivo evaluations which target the performance in downstream NLP applications rather than artificial benchmarks.

Parallel corpora are widely used in a variety of Natural Language Processing tasks, from Machine Translation to cross-lingual Word Sense Disambiguation, where parallel sentences can be exploited to automatically generate high-quality sense annotations on a large scale. In this paper we present EuroSense, a multilingual sense-annotated resource based on the joint disambiguation of the Europarl parallel corpus, with almost 123 million sense annotations for over 155 thousand distinct concepts and entities from a language-independent unified sense inventory. We evaluate the quality of our sense annotations intrinsically and extrinsically, showing their effectiveness as training data for Word Sense Disambiguation.

The increasing amount of multilingual text collections available in different domains makes its automatic processing essential for the development of a given field. However, standard processing techniques based on statistical clues and keyword searches have clear limitations. Instead, we propose a knowledge-based processing pipeline which overcomes most of the limitations of these techniques. This, in turn, enables direct comparison across texts in different languages without the need of translation. In this paper we show the potential of this approach for semantically indexing multilingual text collections in the history domain. In our experiments we used a version of the Bible translated in four different languages, evaluating the precision of our semantic indexing pipeline and showing its reliability on the cross-lingual text retrieval task.

Linking concepts and named entities to knowledge bases has become a crucial Natural Language Understanding task. In this respect, recent works have shown the key advantage of exploiting textual definitions in various Natural Language Processing applications. However, to date there are no reliable large-scale corpora of sense-annotated textual definitions available to the research community. In this paper we present a large-scale high-quality corpus of disambiguated glosses in multiple languages, comprising sense annotations of both concepts and named entities from a unified sense inventory. Our approach for the construction and disambiguation of the corpus builds upon the structure of a large multilingual semantic network and a state-of-the-art disambiguation system; first, we gather complementary information of equivalent definitions across different languages to provide context for disambiguation, and then we combine it with a semantic similarity-based refinement. As a result we obtain a multilingual corpus of textual definitions featuring over 38 million definitions in 263 languages, and we make it freely available at http://lcl.uniroma1.it/disambiguated-glosses. Experiments on Open Information Extraction and Sense Clustering show how two state-of-the-art approaches improve their performance by integrating our disambiguated corpus into their pipeline.

WordNet is probably the best known lexical resource in Natural Language Processing. While it is widely regarded as a high quality repository of concepts and semantic relations, updating and extending it manually is costly. One important type of relation which could potentially add enormous value to WordNet is the inclusion of collocational information, which is paramount in tasks such as Machine Translation, Natural Language Generation and Second Language Learning. In this paper, we present ColWordNet (CWN), an extended WordNet version with fine-grained collocational information, automatically introduced thanks to a method exploiting linear relations between analogous sense-level embeddings spaces. We perform both intrinsic and extrinsic evaluations, and release CWN for the use and scrutiny of the community.

Representing the semantics of linguistic items in a machine ­interpretable form has been a major goal of Natural Language Processing since its earliest days. Among the range of different linguistic items, words have attracted the most research attention. However, word representations have an important limitation: they conflate different meanings of a word into a single vector. Representations of word senses have the potential to overcome this inherent limitation. Indeed, the representation of individual word senses and concepts has recently gained in popularity with several experimental results showing that a considerable performance improvement can be achieved across different NLP applications upon moving from word level to the deeper sense and concept levels. Another interesting point regarding the representation of concepts and word senses is that these models can be seamlessly applied to other linguistic items, such as words, phrases, sentences, etc.This tutorial will first provide a brief overview of the recent literature concerning word representation (both count based and neural network based). It will then describe the advantages of moving from the word level to the deeper level of word senses and concepts, providing an extensive review of state ­of ­the ­art systems. Approaches covered will not only include those which draw upon knowledge resources such as WordNet, Wikipedia, BabelNet or FreeBase as reference, but also the so ­called multi ­prototype approaches which learn sense distinctions by using different clustering techniques. Our tutorial will discuss the advantages and potential limitations of all approaches, showing their most successful applications to date. We will conclude by presenting current open problems and lines of future work.