For text classification tasks, finetuned language models perform remarkably well. Yet, they tend to rely on spurious patterns in training data, thus limiting their performance on out-of-distribution (OOD) test data. Among recent models aiming to avoid this spurious pattern problem, adding extra counterfactual samples to the training data has proven to be very effective. Yet, counterfactual data generation is costly since it relies on human annotation. Thus, we propose a novel solution that only requires annotation of a small fraction (e.g., 1%) of the original training data, and uses automatic generation of extra counterfactuals in an encoding vector space. We demonstrate the effectiveness of our approach in sentiment classification, using IMDb data for training and other sets for OOD tests (i.e., Amazon, SemEval and Yelp). We achieve noticeable accuracy improvements by adding only 1% manual counterfactuals: +3% compared to adding +100% in-distribution training samples, +1.3% compared to alternate counterfactual approaches.

Most existing pre-trained language representation models (PLMs) are sub-optimal in sentiment analysis tasks, as they capture the sentiment information from word-level while under-considering sentence-level information. In this paper, we propose SentiWSP, a novel Sentiment-aware pre-trained language model with combined Word-level and Sentence-level Pre-training tasks. The word level pre-training task detects replaced sentiment words, via a generator-discriminator framework, to enhance the PLM's knowledge about sentiment words. The sentence level pre-training task further strengthens the discriminator via a contrastive learning framework, with similar sentences as negative samples, to encode sentiments in a sentence. Extensive experimental results show that SentiWSP achieves new state-of-the-art performance on various sentence-level and aspect-level sentiment classification benchmarks. We have made our code and model publicly available at https://github.com/XMUDM/SentiWSP.

Recently, there have merged a class of task-oriented dialogue (TOD) datasets collected through Wizard-of-Oz simulated games. However, the Wizard-of-Oz data are in fact simulated data and thus are fundamentally different from real-life conversations, which are more noisy and casual. Recently, the SereTOD challenge is organized and releases the MobileCS dataset, which consists of real-world dialog transcripts between real users and customer-service staffs from China Mobile. Based on the MobileCS dataset, the SereTOD challenge has two tasks, not only evaluating the construction of the dialogue system itself, but also examining information extraction from dialog transcripts, which is crucial for building the knowledge base for TOD. This paper mainly presents a baseline study of the two tasks with the MobileCS dataset. We introduce how the two baselines are constructed, the problems encountered, and the results. We anticipate that the baselines can facilitate exciting future research to build human-robot dialogue systems for real-life tasks.

This study investigated and compared public sentiment related to COVID-19 vaccines expressed on two popular social media platforms, Reddit and Twitter, harvested from January 1, 2020, to March 1, 2022. To accomplish this task, we created a fine-tuned DistilRoBERTa model to predict sentiments of approximately 9.5 million Tweets and 70 thousand Reddit comments. To fine-tune our model, our team manually labeled the sentiment of 3600 Tweets and then augmented our dataset by the method of back-translation. Text sentiment for each social media platform was then classified with our fine-tuned model using Python and the Huggingface sentiment analysis pipeline. Our results determined that the average sentiment expressed on Twitter was more negative (52% positive) than positive and the sentiment expressed on Reddit was more positive than negative (53% positive). Though average sentiment was found to vary between these social media platforms, both displayed similar behavior related to sentiment shared at key vaccine-related developments during the pandemic. Considering this similar trend in shared sentiment demonstrated across social media platforms, Twitter and Reddit continue to be valuable data sources that public health officials can utilize to strengthen vaccine confidence and combat misinformation. As the spread of misinformation poses a range of psychological and psychosocial risks (anxiety, fear, etc.), there is an urgency in understanding the public perspective and attitude toward shared falsities. Comprehensive educational delivery systems tailored to the population's expressed sentiments that facilitate digital literacy, health information-seeking behavior, and precision health promotion could aid in clarifying such misinformation.

Federated learning is particularly susceptible to model poisoning and backdoor attacks because individual users have direct control over the training data and model updates. At the same time, the attack power of an individual user is limited because their updates are quickly drowned out by those of many other users. Existing attacks do not account for future behaviors of other users, and thus require many sequential updates and their effects are quickly erased. We propose an attack that anticipates and accounts for the entire federated learning pipeline, including behaviors of other clients, and ensures that backdoors are effective quickly and persist even after multiple rounds of community updates. We show that this new attack is effective in realistic scenarios where the attacker only contributes to a small fraction of randomly sampled rounds and demonstrate this attack on image classification, next-word prediction, and sentiment analysis.

Most previous work in music emotion recognition assumes a single or a few song-level labels for the whole song. While it is known that different emotions can vary in intensity within a song, annotated data for this setup is scarce and difficult to obtain. In this work, we propose a method to predict emotion dynamics in song lyrics without song-level supervision. We frame each song as a time series and employ a State Space Model (SSM), combining a sentence-level emotion predictor with an Expectation-Maximization (EM) procedure to generate the full emotion dynamics. Our experiments show that applying our method consistently improves the performance of sentence-level baselines without requiring any annotated songs, making it ideal for limited training data scenarios. Further analysis through case studies shows the benefits of our method while also indicating the limitations and pointing to future directions.

Words often convey affect -- emotions, feelings, and attitudes. Further, different words can convey affect to various degrees (intensities). However, existing manually created lexicons for basic emotions (such as anger and fear) indicate only coarse categories of affect association (for example, associated with anger or not associated with anger). Automatic lexicons of affect provide fine degrees of association, but they tend not to be accurate as human-created lexicons. Here, for the first time, we present a manually created affect intensity lexicon with real-valued scores of intensity for four basic emotions: anger, fear, joy, and sadness. (We will subsequently add entries for more emotions such as disgust, anticipation, trust, and surprise.) We refer to this dataset as the NRC Affect Intensity Lexicon, or AIL for short. AIL has entries for close to 6,000 English words. We used a technique called best-worst scaling (BWS) to create the lexicon. BWS improves annotation consistency and obtains reliable fine-grained scores (split-half reliability > 0.91). We also compare the entries in AIL with the entries in the NRC VAD Lexicon, which has valence, arousal, and dominance (VAD) scores for 20K English words. We find that anger, fear, and sadness words, on average, have very similar VAD scores. However, sadness words tend to have slightly lower dominance scores than fear and anger words. The Affect Intensity Lexicon has applications in automatic emotion analysis in a number of domains such as commerce, education, intelligence, and public health. AIL is also useful in the building of natural language generation systems.

Recent years have witnessed the rise and success of pre-training techniques in visually-rich document understanding. However, most existing methods lack the systematic mining and utilization of layout-centered knowledge, leading to sub-optimal performances. In this paper, we propose ERNIE-Layout, a novel document pre-training solution with layout knowledge enhancement in the whole workflow, to learn better representations that combine the features from text, layout, and image. Specifically, we first rearrange input sequences in the serialization stage, and then present a correlative pre-training task, reading order prediction, to learn the proper reading order of documents. To improve the layout awareness of the model, we integrate a spatial-aware disentangled attention into the multi-modal transformer and a replaced regions prediction task into the pre-training phase. Experimental results show that ERNIE-Layout achieves superior performance on various downstream tasks, setting new state-of-the-art on key information extraction, document image classification, and document question answering datasets. The code and models are publicly available at http://github.com/PaddlePaddle/PaddleNLP/tree/develop/model_zoo/ernie-layout.

Automatically generated emotion arcs -- that capture how an individual or a population feels over time -- are widely used in industry and research. However, there is little work on evaluating the generated arcs. This is in part due to the difficulty of establishing the true (gold) emotion arc. Our work, for the first time, systematically and quantitatively evaluates automatically generated emotion arcs. We also compare two common ways of generating emotion arcs: Machine-Learning (ML) models and Lexicon-Only (LexO) methods. Using a number of diverse datasets, we systematically study the relationship between the quality of an emotion lexicon and the quality of the emotion arc that can be generated with it. We also study the relationship between the quality of an instance-level emotion detection system (say from an ML model) and the quality of emotion arcs that can be generated with it. We show that despite being markedly poor at instance level, LexO methods are highly accurate at generating emotion arcs by aggregating information from hundreds of instances. This has wide-spread implications for commercial development, as well as research in psychology, public health, digital humanities, etc. that values simple interpretable methods and disprefers the need for domain-specific training data, programming expertise, and high-carbon-footprint models.

We present an empirical study on methods for span finding, the selection of consecutive tokens in text for some downstream tasks. We focus on approaches that can be employed in training end-to-end information extraction systems, and find there is no definitive solution without considering task properties, and provide our observations to help with future design choices: 1) a tagging approach often yields higher precision while span enumeration and boundary prediction provide higher recall; 2) span type information can benefit a boundary prediction approach; 3) additional contextualization does not help span finding in most cases.