The robustness of a model for real-world deployment is decided by how well it performs on unseen data and distinguishes between in-domain and out-of-domain samples. Visual document classifiers have shown impressive performance on in-distribution test sets. However, they tend to have a hard time correctly classifying and differentiating out-of-distribution examples. Image-based classifiers lack the text component, whereas multi-modality transformer-based models face the token serialization problem in visual documents due to their diverse layouts. They also require a lot of computing power during inference, making them impractical for many real-world applications. We propose, GVdoc, a graph-based document classification model that addresses both of these challenges. Our approach generates a document graph based on its layout, and then trains a graph neural network to learn node and graph embeddings. Through experiments, we show that our model, even with fewer parameters, outperforms state-of-the-art models on out-of-distribution data while retaining comparable performance on the in-distribution test set.

Label noise—incorrectly or ambiguously labeled training examples—can negatively impact model performance. Although noise detection techniques have been around for decades, practitioners rarely apply them, as manual noise remediation is a tedious process. Examples incorrectly flagged as noise waste reviewers’ time, and correcting label noise without guidance can be difficult. We propose LNIC, a noise-detection method that uses an example’s neighborhood within the training set to (a) reduce false positives and (b) provide an explanation as to why the ex- ample was flagged as noise. We demonstrate on several short-text classification datasets that LNIC outperforms the state of the art on measures of precision and F0.5-score. We also show how LNIC’s training set context helps a reviewer to understand and correct label noise in a dataset. The LNIC tool lowers the barriers to label noise remediation, increasing its utility for NLP practitioners.

Clustering documents by type—grouping invoices with invoices and articles with articles—is a desirable first step for organizing large collections of document scans. Humans approaching this task use both the semantics of the text and the document layout to assist in grouping like documents. LayoutLM (Xu et al., 2019), a layout-aware transformer built on top of BERT with state-of-the-art performance on document-type classification, could reasonably be expected to outperform regular BERT (Devlin et al., 2018) for document-type clustering. However, we find experimentally that BERT significantly outperforms LayoutLM on this task (p <0.001). We analyze clusters to show where layout awareness is an asset and where it is a liability.

Most summarization research focuses on summarizing the entire given text, but in practice readers are often interested in only one aspect of the document or conversation. We propose targeted summarization as an umbrella category for summarization tasks that intentionally consider only parts of the input data. This covers query-based summarization, update summarization, and a new task we propose where the goal is to summarize a particular aspect of a document. However, collecting data for this new task is hard because directly asking annotators (e.g., crowd workers) to write summaries leads to data with low accuracy when there are a large number of facts to include. We introduce a novel crowdsourcing workflow, Pin-Refine, that allows us to collect high-quality summaries for our task, a necessary step for the development of automatic systems.

To be informative, an evaluation must measure how well systems generalize to realistic unseen data. We identify limitations of and propose improvements to current evaluations of text-to-SQL systems. First, we compare human-generated and automatically generated questions, characterizing properties of queries necessary for real-world applications. To facilitate evaluation on multiple datasets, we release standardized and improved versions of seven existing datasets and one new text-to-SQL dataset. Second, we show that the current division of data into training and test sets measures robustness to variations in the way questions are asked, but only partially tests how well systems generalize to new queries; therefore, we propose a complementary dataset split for evaluation of future work. Finally, we demonstrate how the common practice of anonymizing variables during evaluation removes an important challenge of the task. Our observations highlight key difficulties, and our methodology enables effective measurement of future development.