Document retrieval enables users to find their required documents accurately and quickly. To satisfy the requirement of retrieval efficiency, prevalent deep neural methods adopt a representation-based matching paradigm, which saves online matching time by pre-storing document representations offline. However, the above paradigm consumes vast local storage space, especially when storing the document as word-grained representations. To tackle this, we present TGTR, a Topic-Grained Text Representation-based Model for document retrieval. Following the representation-based matching paradigm, TGTR stores the document representations offline to ensure retrieval efficiency, whereas it significantly reduces the storage requirements by using novel topicgrained representations rather than traditional word-grained. Experimental results demonstrate that compared to word-grained baselines, TGTR is consistently competitive with them on TREC CAR and MS MARCO in terms of retrieval accuracy, but it requires less than 1/10 of the storage space required by them. Moreover, TGTR overwhelmingly surpasses global-grained baselines in terms of retrieval accuracy.
We present DiffusionBERT, a new generative masked language model based on discrete diffusion models. Diffusion models and many pre-trained language models have a shared training objective, i.e., denoising, making it possible to combine the two powerful models and enjoy the best of both worlds. On the one hand, diffusion models offer a promising training strategy that helps improve the generation quality. On the other hand, pre-trained denoising language models (e.g., BERT) can be used as a good initialization that accelerates convergence. We explore training BERT to learn the reverse process of a discrete diffusion process with an absorbing state and elucidate several designs to improve it. First, we propose a new noise schedule for the forward diffusion process that controls the degree of noise added at each step based on the information of each token. Second, we investigate several designs of incorporating the time step into BERT. Experiments on unconditional text generation demonstrate that DiffusionBERT achieves significant improvement over existing diffusion models for text (e.g., D3PM and Diffusion-LM) and previous generative masked language models in terms of perplexity and BLEU score.
In this paper we present our contribution to the TSAR-2022 Shared Task on Lexical Simplification of the EMNLP 2022 Workshop on Text Simplification, Accessibility, and Readability. Our approach builds on and extends the unsupervised lexical simplification system with pretrained encoders (LSBert) system in the following ways: For the subtask of simplification candidate selection, it utilizes a RoBERTa transformer language model and expands the size of the generated candidate list. For subsequent substitution ranking, it introduces a new feature weighting scheme and adopts a candidate filtering method based on textual entailment to maximize semantic similarity between the target word and its simplification. Our best-performing system improves LSBert by 5.9% accuracy and achieves second place out of 33 ranked solutions.
This preprint describes work in progress on LR-Sum, a new permissively-licensed dataset created with the goal of enabling further research in automatic summarization for less-resourced languages. LR-Sum contains human-written summaries for 40 languages, many of which are less-resourced. We describe our process for extracting and filtering the dataset from the Multilingual Open Text corpus (Palen-Michel et al., 2022). The source data is public domain newswire collected from from Voice of America websites, and LR-Sum is released under a Creative Commons license (CC BY 4.0), making it one of the most openly-licensed multilingual summarization datasets. We describe how we plan to use the data for modeling experiments and discuss limitations of the dataset.
In this paper, we have defined a novel task of affective feedback synthesis that deals with generating feedback for input text & corresponding image in a similar way as humans respond towards the multimodal data. A feedback synthesis system has been proposed and trained using ground-truth human comments along with image-text input. We have also constructed a large-scale dataset consisting of image, text, Twitter user comments, and the number of likes for the comments by crawling the news articles through Twitter feeds. The proposed system extracts textual features using a transformer-based textual encoder while the visual features have been extracted using a Faster region-based convolutional neural networks model. The textual and visual features have been concatenated to construct the multimodal features using which the decoder synthesizes the feedback. We have compared the results of the proposed system with the baseline models using quantitative and qualitative measures. The generated feedbacks have been analyzed using automatic and human evaluation. They have been found to be semantically similar to the ground-truth comments and relevant to the given text-image input.
Image classification, which classifies images by pre-defined categories, has been the dominant approach to visual representation learning over the last decade. Visual learning through image-text alignment, however, has emerged to show promising performance, especially for zero-shot recognition. We believe that these two learning tasks are complementary, and suggest combining them for better visual learning. We propose a deep fusion method with three adaptations that effectively bridge two learning tasks, rather than shallow fusion through naive multi-task learning. First, we modify the previous common practice in image classification, a linear classifier, with a cosine classifier which shows comparable performance. Second, we convert the image classification problem from learning parametric category classifier weights to learning a text encoder as a meta network to generate category classifier weights. The learnt text encoder is shared between image classification and image-text alignment. Third, we enrich each class name with a description to avoid confusion between classes and make the classification method closer to the image-text alignment. We prove that this deep fusion approach performs better on a variety of visual recognition tasks and setups than the individual learning or shallow fusion approach, from zero-shot/few-shot image classification, such as the Kornblith 12-dataset benchmark, to downstream tasks of action recognition, semantic segmentation, and object detection in fine-tuning and open-vocabulary settings. The code will be available at https://github.com/weiyx16/iCAR.
Objective: The n2c2/UW SDOH Challenge explores the extraction of social determinant of health (SDOH) information from clinical notes. The objectives include the advancement of natural language processing (NLP) information extraction techniques for SDOH and clinical information more broadly. This paper presents the shared task, data, participating teams, performance results, and considerations for future work. Materials and Methods: The task used the Social History Annotated Corpus (SHAC), which consists of clinical text with detailed event-based annotations for SDOH events such as alcohol, drug, tobacco, employment, and living situation. Each SDOH event is characterized through attributes related to status, extent, and temporality. The task includes three subtasks related to information extraction (Subtask A), generalizability (Subtask B), and learning transfer (Subtask C). In addressing this task, participants utilized a range of techniques, including rules, knowledge bases, n-grams, word embeddings, and pretrained language models (LM). Results: A total of 15 teams participated, and the top teams utilized pretrained deep learning LM. The top team across all subtasks used a sequence-to-sequence approach achieving 0.901 F1 for Subtask A, 0.774 F1 Subtask B, and 0.889 F1 for Subtask C. Conclusions: Similar to many NLP tasks and domains, pretrained LM yielded the best performance, including generalizability and learning transfer. An error analysis indicates extraction performance varies by SDOH, with lower performance achieved for conditions, like substance use and homelessness, that increase health risks (risk factors) and higher performance achieved for conditions, like substance abstinence and living with family, that reduce health risks (protective factors).
Sentence summarization shortens given texts while maintaining core contents of the texts. Unsupervised approaches have been studied to summarize texts without human-written summaries. However, recent unsupervised models are extractive, which remove words from texts and thus they are less flexible than abstractive summarization. In this work, we devise an abstractive model based on reinforcement learning without ground-truth summaries. We formulate the unsupervised summarization based on the Markov decision process with rewards representing the summary quality. To further enhance the summary quality, we develop a multi-summary learning mechanism that generates multiple summaries with varying lengths for a given text, while making the summaries mutually enhance each other. Experimental results show that the proposed model substantially outperforms both abstractive and extractive models, yet frequently generating new words not contained in input texts.
This paper tackles the challenging problem of automating code updates to fix deprecated API usages of open source libraries by analyzing their release notes. Our system employs a three-tier architecture: first, a web crawler service retrieves deprecation documentation from the web; then a specially built parser processes those text documents into tree-structured representations; finally, a client IDE plugin locates and fixes identified deprecated usages of libraries in a given codebase. The focus of this paper in particular is the parsing component. We introduce a novel transition-based parser in two variants: based on a classical feature engineered classifier and a neural tree encoder. To confirm the effectiveness of our method, we gathered and labeled a set of 426 API deprecations from 7 well-known Python data science libraries, and demonstrated our approach decisively outperforms a non-trivial neural machine translation baseline.
Prior work on ideology prediction has largely focused on single modalities, i.e., text or images. In this work, we introduce the task of multimodal ideology prediction, where a model predicts binary or five-point scale ideological leanings, given a text-image pair with political content. We first collect five new large-scale datasets with English documents and images along with their ideological leanings, covering news articles from a wide range of US mainstream media and social media posts from Reddit and Twitter. We conduct in-depth analyses of news articles and reveal differences in image content and usage across the political spectrum. Furthermore, we perform extensive experiments and ablation studies, demonstrating the effectiveness of targeted pretraining objectives on different model components. Our best-performing model, a late-fusion architecture pretrained with a triplet objective over multimodal content, outperforms the state-of-the-art text-only model by almost 4% and a strong multimodal baseline with no pretraining by over 3%.