Large language models (LLMs) have shown great success in various Natural Language Processing (NLP) tasks, whist they still need updates after deployment to fix errors or keep pace with the changing knowledge in the world. Researchers formulate such problem as Model Editing and have developed various editors focusing on different axes of editing properties. However, current editors can hardly support all properties and rely on heavy computational resources. In this paper, we propose a plug-in Model Editing method based on neuron-indexed dynamic LoRA (MELO), which alters the behavior of language models by dynamically activating certain LoRA blocks according to the index built in an inner vector database. Our method satisfies various editing properties with high efficiency and can be easily integrated into multiple LLM backbones. Experimental results show that our proposed MELO achieves state-of-the-art editing performance on three sequential editing tasks (document classification, question answering and hallucination correction), while requires the least trainable parameters and computational cost.
This paper introduces a new large-scale dataset for Farsi document images, named SUT, which aims to tackle the challenges associated with obtaining diverse and substantial ground-truth data for supervised models in document image analysis (DIA) tasks, such as document image classification, text detection and recognition, and information retrieval. The dataset comprises 62,453 images that have been categorized into 21 distinct classes, including identity documents featuring synthetically generated personal information superimposed on various backgrounds. The dataset also includes corresponding files with labeling information for the images. The ground-truth data is organized in CSV files containing compiled image file paths and associated information about the embedded data. To demonstrate the efficacy of the SUT dataset in DIA tasks, it was utilized for document classification (achieving an accuracy of 86% using a convolutional neural network) and OCR (achieving a CER of 0.083 and 0.072 using Tesseract and EasyOCR engines, respectively). The SUT dataset represents a valuable resource for researchers who are interested in developing and evaluating supervised models in Farsi document image analysis.
Despite the prevalence of pretrained language models in natural language understanding tasks, understanding lengthy text such as document is still challenging due to the data sparseness problem. Inspired by that humans develop their ability of understanding lengthy text from reading shorter text, we propose a simple yet effective summarization-based data augmentation, SUMMaug, for document classification. We first obtain easy-to-learn examples for the target document classification task by summarizing the input of the original training examples, while optionally merging the original labels to conform to the summarized input. We then use the generated pseudo examples to perform curriculum learning. Experimental results on two datasets confirmed the advantage of our method compared to existing baseline methods in terms of robustness and accuracy. We release our code and data at https://github.com/etsurin/summaug.
Multi-label document classification is a traditional task in NLP. Compared to single-label classification, each document can be assigned multiple classes. This problem is crucially important in various domains, such as tagging scientific articles. Documents are often structured into several sections such as abstract and title. Current approaches treat different sections equally for multi-label classification. We argue that this is not a realistic assumption, leading to sub-optimal results. Instead, we propose a new method called Learning Section Weights (LSW), leveraging the contribution of each distinct section for multi-label classification. Via multiple feed-forward layers, LSW learns to assign weights to each section of, and incorporate the weights in the prediction. We demonstrate our approach on scientific articles. Experimental results on public (arXiv) and private (Elsevier) datasets confirm the superiority of LSW, compared to state-of-the-art multi-label document classification methods. In particular, LSW achieves a 1.3% improvement in terms of macro averaged F1-score while it achieves 1.3% in terms of macro averaged recall on the publicly available arXiv dataset.
In the fundamental statistics course, students are taught to remember the well-known saying: "Correlation is not Causation". Till now, statistics (i.e., correlation) have developed various successful frameworks, such as Transformer and Pre-training large-scale models, which have stacked multiple parallel self-attention blocks to imitate a wide range of tasks. However, in the causation community, how to build an integrated causal framework still remains an untouched domain despite its excellent intervention capabilities. In this paper, we propose the Causal Graph Routing (CGR) framework, an integrated causal scheme relying entirely on the intervention mechanisms to reveal the cause-effect forces hidden in data. Specifically, CGR is composed of a stack of causal layers. Each layer includes a set of parallel deconfounding blocks from different causal graphs. We combine these blocks via the concept of the proposed sufficient cause, which allows the model to dynamically select the suitable deconfounding methods in each layer. CGR is implemented as the stacked networks, integrating no confounder, back-door adjustment, front-door adjustment, and probability of sufficient cause. We evaluate this framework on two classical tasks of CV and NLP. Experiments show CGR can surpass the current state-of-the-art methods on both Visual Question Answer and Long Document Classification tasks. In particular, CGR has great potential in building the "causal" pre-training large-scale model that effectively generalizes to diverse tasks. It will improve the machines' comprehension of causal relationships within a broader semantic space.
Large language models record impressive performance on many natural language processing tasks. However, their knowledge capacity is limited to the pretraining corpus. Retrieval augmentation offers an effective solution by retrieving context from external knowledge sources to complement the language model. However, existing retrieval augmentation techniques ignore the structural relationships between these documents. Furthermore, retrieval models are not explored much in scientific tasks, especially in regard to the faithfulness of retrieved documents. In this paper, we propose a novel structure-aware retrieval augmented language model that accommodates document structure during retrieval augmentation. We create a heterogeneous document graph capturing multiple types of relationships (e.g., citation, co-authorship, etc.) that connect documents from more than 15 scientific disciplines (e.g., Physics, Medicine, Chemistry, etc.). We train a graph neural network on the curated document graph to act as a structural encoder for the corresponding passages retrieved during the model pretraining. Particularly, along with text embeddings of the retrieved passages, we obtain structural embeddings of the documents (passages) and fuse them together before feeding them to the language model. We evaluate our model extensively on various scientific benchmarks that include science question-answering and scientific document classification tasks. Experimental results demonstrate that structure-aware retrieval improves retrieving more coherent, faithful and contextually relevant passages, while showing a comparable performance in the overall accuracy.
In recent times, large language models (LLMs) have shown impressive performance on various document-level tasks such as document classification, summarization, and question-answering. However, research on understanding their capabilities on the task of self-contradictions in long documents has been very limited. In this work, we introduce ContraDoc, the first human-annotated dataset to study self-contradictions in long documents across multiple domains, varying document lengths, self-contradictions types, and scope. We then analyze the current capabilities of four state-of-the-art open-source and commercially available LLMs: GPT3.5, GPT4, PaLM2, and LLaMAv2 on this dataset. While GPT4 performs the best and can outperform humans on this task, we find that it is still unreliable and struggles with self-contradictions that require more nuance and context. We release the dataset and all the code associated with the experiments (https://github.com/ddhruvkr/CONTRADOC).
US corporations regularly spend millions of dollars reviewing electronically-stored documents in legal matters. Recently, attorneys apply text classification to efficiently cull massive volumes of data to identify responsive documents for use in these matters. While text classification is regularly used to reduce the discovery costs of legal matters, it also faces a perception challenge: amongst lawyers, this technology is sometimes looked upon as a "black box". Put simply, no extra information is provided for attorneys to understand why documents are classified as responsive. In recent years, explainable machine learning has emerged as an active research area. In an explainable machine learning system, predictions or decisions made by a machine learning model are human understandable. In legal 'document review' scenarios, a document is responsive, because one or more of its small text snippets are deemed responsive. In these scenarios, if these responsive snippets can be located, then attorneys could easily evaluate the model's document classification decisions - this is especially important in the field of responsible AI. Our prior research identified that predictive models created using annotated training text snippets improved the precision of a model when compared to a model created using all of a set of documents' text as training. While interesting, manually annotating training text snippets is not generally practical during a legal document review. However, small increases in precision can drastically decrease the cost of large document reviews. Automating the identification of training text snippets without human review could then make the application of training text snippet-based models a practical approach.
Document image classification is different from plain-text document classification and consists of classifying a document by understanding the content and structure of documents such as forms, emails, and other such documents. We show that the only existing dataset for this task (Lewis et al., 2006) has several limitations and we introduce two newly curated multilingual datasets WIKI-DOC and MULTIEURLEX-DOC that overcome these limitations. We further undertake a comprehensive study of popular visually-rich document understanding or Document AI models in previously untested setting in document image classification such as 1) multi-label classification, and 2) zero-shot cross-lingual transfer setup. Experimental results show limitations of multilingual Document AI models on cross-lingual transfer across typologically distant languages. Our datasets and findings open the door for future research into improving Document AI models.
This paper introduces a deep learning model tailored for document information analysis, emphasizing document classification, entity relation extraction, and document visual question answering. The proposed model leverages transformer-based models to encode all the information present in a document image, including textual, visual, and layout information. The model is pre-trained and subsequently fine-tuned for various document image analysis tasks. The proposed model incorporates three additional tasks during the pre-training phase, including reading order identification of different layout segments in a document image, layout segments categorization as per PubLayNet, and generation of the text sequence within a given layout segment (text block). The model also incorporates a collective pre-training scheme where losses of all the tasks under consideration, including pre-training and fine-tuning tasks with all datasets, are considered. Additional encoder and decoder blocks are added to the RoBERTa network to generate results for all tasks. The proposed model achieved impressive results across all tasks, with an accuracy of 95.87% on the RVL-CDIP dataset for document classification, F1 scores of 0.9306, 0.9804, 0.9794, and 0.8742 on the FUNSD, CORD, SROIE, and Kleister-NDA datasets respectively for entity relation extraction, and an ANLS score of 0.8468 on the DocVQA dataset for visual question answering. The results highlight the effectiveness of the proposed model in understanding and interpreting complex document layouts and content, making it a promising tool for document analysis tasks.
We study optimal transport (OT) problem for probability measures supported on a tree metric space. It is known that such OT problem (i.e., tree-Wasserstein (TW)) admits a closed-form expression, but depends fundamentally on the underlying tree structure over supports of input measures. In practice, the given tree structure may be, however, perturbed due to noisy or adversarial measurements. To mitigate this issue, we follow the max-min robust OT approach which considers the maximal possible distances between two input measures over an uncertainty set of tree metrics. In general, this approach is hard to compute, even for measures supported in one-dimensional space, due to its non-convexity and non-smoothness which hinders its practical applications, especially for large-scale settings. In this work, we propose novel uncertainty sets of tree metrics from the lens of edge deletion/addition which covers a diversity of tree structures in an elegant framework. Consequently, by building upon the proposed uncertainty sets, and leveraging the tree structure over supports, we show that the robust OT also admits a closed-form expression for a fast computation as its counterpart standard OT (i.e., TW). Furthermore, we demonstrate that the robust OT satisfies the metric property and is negative definite. We then exploit its negative definiteness to propose positive definite kernels and test them in several simulations on various real-world datasets on document classification and topological data analysis.
Automatic Keyphrase Extraction involves identifying essential phrases in a document. These keyphrases are crucial in various tasks such as document classification, clustering, recommendation, indexing, searching, summarization, and text simplification. This paper introduces a platform that integrates keyphrase datasets and facilitates the evaluation of keyphrase extraction algorithms. The platform includes BibRank, an automatic keyphrase extraction algorithm that leverages a rich dataset obtained by parsing bibliographic data in BibTeX format. BibRank combines innovative weighting techniques with positional, statistical, and word co-occurrence information to extract keyphrases from documents. The platform proves valuable for researchers and developers seeking to enhance their keyphrase extraction algorithms and advance the field of natural language processing.
This paper conducts a comprehensive investigation into applying large language models, particularly on BioBERT, in healthcare. It begins with thoroughly examining previous natural language processing (NLP) approaches in healthcare, shedding light on the limitations and challenges these methods face. Following that, this research explores the path that led to the incorporation of BioBERT into healthcare applications, highlighting its suitability for addressing the specific requirements of tasks related to biomedical text mining. The analysis outlines a systematic methodology for fine-tuning BioBERT to meet the unique needs of the healthcare domain. This approach includes various components, including the gathering of data from a wide range of healthcare sources, data annotation for tasks like identifying medical entities and categorizing them, and the application of specialized preprocessing techniques tailored to handle the complexities found in biomedical texts. Additionally, the paper covers aspects related to model evaluation, with a focus on healthcare benchmarks and functions like processing of natural language in biomedical, question-answering, clinical document classification, and medical entity recognition. It explores techniques to improve the model's interpretability and validates its performance compared to existing healthcare-focused language models. The paper thoroughly examines ethical considerations, particularly patient privacy and data security. It highlights the benefits of incorporating BioBERT into healthcare contexts, including enhanced clinical decision support and more efficient information retrieval. Nevertheless, it acknowledges the impediments and complexities of this integration, encompassing concerns regarding data privacy, transparency, resource-intensive requirements, and the necessity for model customization to align with diverse healthcare domains.
This work presents KoBigBird-large, a large size of Korean BigBird that achieves state-of-the-art performance and allows long sequence processing for Korean language understanding. Without further pretraining, we only transform the architecture and extend the positional encoding with our proposed Tapered Absolute Positional Encoding Representations (TAPER). In experiments, KoBigBird-large shows state-of-the-art overall performance on Korean language understanding benchmarks and the best performance on document classification and question answering tasks for longer sequences against the competitive baseline models. We publicly release our model here.
A common bottleneck in deploying supervised learning systems is collecting human-annotated examples. In many domains, annotators form an opinion about the label of an example incrementally--e.g., each additional word read from a document or each additional minute spent inspecting a video helps inform the annotation. In this paper, we investigate whether we can train learning systems more efficiently by requesting an annotation before inspection is fully complete--e.g., after reading only 25 words of a document. While doing so may reduce the overall annotation time, it also introduces the risk that the annotator might not be able to provide a label if interrupted too early. We propose an anytime active learning approach that optimizes the annotation time and response rate simultaneously. We conduct user studies on two document classification datasets and develop simulated annotators that mimic the users. Our simulated experiments show that anytime active learning outperforms several baselines on these two datasets. For example, with an annotation budget of one hour, training a classifier by annotating the first 25 words of each document reduces classification error by 17% over annotating the first 100 words of each document.
The selection of features for text classification is a fundamental task in text mining and information retrieval. Despite being the sixth most widely spoken language in the world, Bangla has received little attention due to the scarcity of text datasets. In this research, we collected, annotated, and prepared a comprehensive dataset of 212,184 Bangla documents in seven different categories and made it publicly accessible. We implemented three deep learning generative models: LSTM variational autoencoder (LSTM VAE), auxiliary classifier generative adversarial network (AC-GAN), and adversarial autoencoder (AAE) to extract text features, although their applications are initially found in the field of computer vision. We utilized our dataset to train these three models and used the feature space obtained in the document classification task. We evaluated the performance of the classifiers and found that the adversarial autoencoder model produced the best feature space.
This paper highlights the need to bring document classification benchmarking closer to real-world applications, both in the nature of data tested ($X$: multi-channel, multi-paged, multi-industry; $Y$: class distributions and label set variety) and in classification tasks considered ($f$: multi-page document, page stream, and document bundle classification, ...). We identify the lack of public multi-page document classification datasets, formalize different classification tasks arising in application scenarios, and motivate the value of targeting efficient multi-page document representations. An experimental study on proposed multi-page document classification datasets demonstrates that current benchmarks have become irrelevant and need to be updated to evaluate complete documents, as they naturally occur in practice. This reality check also calls for more mature evaluation methodologies, covering calibration evaluation, inference complexity (time-memory), and a range of realistic distribution shifts (e.g., born-digital vs. scanning noise, shifting page order). Our study ends on a hopeful note by recommending concrete avenues for future improvements.}
Accurately evaluating the similarity of object vector embeddings is of critical importance for natural language processing, information retrieval and classification tasks. Popular similarity scores (e.g cosine similarity) are based on pairs of embedding vectors and disregard the distribution of the ensemble from which objects are drawn. Human perception of object similarity significantly depends on the context in which the objects appear. In this work we propose the $\textit{surprise score}$, an ensemble-normalized similarity metric that encapsulates the contrast effect of human perception and significantly improves the classification performance on zero- and few-shot document classification tasks. This score quantifies the surprise to find a given similarity between two elements relative to the pairwise ensemble similarities. We evaluate this metric on zero/few shot classification and clustering tasks and typically find 10-15 % better performance compared to raw cosine similarity. Our code is available at https://github.com/MeetElise/surprise-similarity.
Materials language processing (MLP) is one of the key facilitators of materials science research, as it enables the extraction of structured information from massive materials science literature. Prior works suggested high-performance MLP models for text classification, named entity recognition (NER), and extractive question answering (QA), which require complex model architecture, exhaustive fine-tuning and a large number of human-labelled datasets. In this study, we develop generative pretrained transformer (GPT)-enabled pipelines where the complex architectures of prior MLP models are replaced with strategic designs of prompt engineering. First, we develop a GPT-enabled document classification method for screening relevant documents, achieving comparable accuracy and reliability compared to prior models, with only small dataset. Secondly, for NER task, we design an entity-centric prompts, and learning few-shot of them improved the performance on most of entities in three open datasets. Finally, we develop an GPT-enabled extractive QA model, which provides improved performance and shows the possibility of automatically correcting annotations. While our findings confirm the potential of GPT-enabled MLP models as well as their value in terms of reliability and practicability, our scientific methods and systematic approach are applicable to any materials science domain to accelerate the information extraction of scientific literature.
Prompting is used to guide or steer a language model in generating an appropriate response that is consistent with the desired outcome. Chaining is a strategy used to decompose complex tasks into smaller, manageable components. In this study, we utilize prompt chaining for extensive legal document classification tasks, which present difficulties due to their intricate domain-specific language and considerable length. Our approach begins with the creation of a concise summary of the original document, followed by a semantic search for related exemplar texts and their corresponding annotations from a training corpus. Finally, we prompt for a label - based on the task - to assign, by leveraging the in-context learning from the few-shot prompt. We demonstrate that through prompt chaining, we can not only enhance the performance over zero-shot, but also surpass the micro-F1 score achieved by larger models, such as ChatGPT zero-shot, using smaller models.