In this work, we build a dense retrieval based semantic search engine on scientific articles from Elsevier. The major challenge is that there is no labeled data for training and testing. We apply a state-of-the-art unsupervised dense retrieval model called Generative Pseudo Labeling that generates high-quality pseudo training labels. Furthermore, since the articles are unbalanced across different domains, we select passages from multiple domains to form balanced training data. For the evaluation, we create two test sets: one manually annotated and one automatically created from the meta-information of our data. We compare the semantic search engine with the currently deployed lexical search engine on the two test sets. The results of the experiment show that the semantic search engine trained with pseudo training labels can significantly improve search performance.

Automatic extraction of funding information from academic articles adds significant value to industry and research communities, including tracking research outcomes by funding organizations, profiling researchers and universities based on the received funding, and supporting open access policies. Two major challenges of identifying and linking funding entities are: (i) sparse graph structure of the Knowledge Base (KB), which makes the commonly used graph-based entity linking approaches suboptimal for the funding domain, (ii) missing entities in KB, which (unlike recent zero-shot approaches) requires marking entity mentions without KB entries as NIL. We propose an entity linking model that can perform NIL prediction and overcome data scarcity issues in a time and data-efficient manner. Our model builds on a transformer-based mention detection and a bi-encoder model to perform entity linking. We show that our model outperforms strong existing baselines.

Despite the important evolutions in few-shot and zero-shot learning techniques, domain specific applications still require expert knowledge and significant effort in annotating and labeling a large volume of unstructured textual data. To mitigate this problem, active learning, and meta-learning attempt to reach a high performance with the least amount of labeled data. In this paper, we introduce a novel approach to combine both lines of work by initializing an active learner with meta-learned parameters obtained through meta-training on tasks similar to the target task during active learning. In this approach we use the pre-trained BERT as our text-encoder and meta-learn its parameters with LEOPARD, which extends the model-agnostic meta-learning method by generating task dependent softmax weights to enable learning across tasks with different number of classes. We demonstrate the effectiveness of our method by performing active learning on five natural language understanding tasks and six datasets with five different acquisition functions. We train two different meta-initializations, and we use the pre-trained BERT base initialization as baseline. We observe that our approach performs better than the baseline at low budget, especially when closely related tasks were present during meta-learning. Moreover, our results show that better performance in the initial phase, i.e., with fewer labeled samples, leads to better performance when larger acquisition batches are used. We also perform an ablation study of the proposed method, showing that active learning with only the meta-learned weights is beneficial and adding the meta-learned learning rates and generating the softmax have negative consequences for the performance.

This paper provides an overview of the DAGPap22 shared task on the detection of automatically generated scientific papers at the Scholarly Document Process workshop colocated with COLING. We frame the detection problem as a binary classification task: given an excerpt of text, label it as either human-written or machine-generated. We shared a dataset containing excerpts from human-written papers as well as artificially generated content and suspicious documents collected by Elsevier publishing and editorial teams. As a test set, the participants are provided with a 5x larger corpus of openly accessible human-written as well as generated papers from the same scientific domains of documents. The shared task saw 180 submissions across 14 participating teams and resulted in two published technical reports. We discuss our findings from the shared task in this overview paper.

Ever since the COVID-19 pandemic broke out, the academic and scientific research community, as well as industry and governments around the world have joined forces in an unprecedented manner to fight the threat. Clinicians, biologists, chemists, bioinformaticians, nurses, data scientists, and all of the affiliated relevant disciplines have been mobilized to help discover efficient treatments for the infected population, as well as a vaccine solution to prevent further the virus spread. In this combat against the virus responsible for the pandemic, key for any advancements is the timely, accurate, peer-reviewed, and efficient communication of any novel research findings. In this paper we present a novel framework to address the information need of filtering efficiently the scientific bibliography for relevant literature around COVID-19. The contributions of the paper are summarized in the following: we define and describe the information need that encompasses the major requirements for COVID-19 articles relevancy, we present and release an expert-curated benchmark set for the task, and we analyze the performance of several state-of-the-art machine learning classifiers that may distinguish the relevant from the non-relevant COVID-19 literature.

Distributed representation of words, or word embeddings, have motivated methods for calculating semantic representations of word sequences such as phrases, sentences and paragraphs. Most of the existing methods to do so either use algorithms to learn such representations, or improve on calculating weighted averages of the word vectors. In this work, we experiment with spectral methods of signal representation and summarization as mechanisms for constructing such word-sequence embeddings in an unsupervised fashion. In particular, we explore an algorithm rooted in fluid-dynamics, known as higher-order Dynamic Mode Decomposition, which is designed to capture the eigenfrequencies, and hence the fundamental transition dynamics, of periodic and quasi-periodic systems. It is empirically observed that this approach, which we call EigenSent, can summarize transitions in a sequence of words and generate an embedding that can represent well the sequence itself. To the best of the authors’ knowledge, this is the first application of a spectral decomposition and signal summarization technique on text, to create sentence embeddings. We test the efficacy of this algorithm in creating sentence embeddings on three public datasets, where it performs appreciably well. Moreover it is also shown that, due to the positive combination of their complementary properties, concatenating the embeddings generated by EigenSent with simple word vector averaging achieves state-of-the-art results.

The rapid growth of documents across the web has necessitated finding means of discarding redundant documents and retaining novel ones. Capturing redundancy is challenging as it may involve investigating at a deep semantic level. Techniques for detecting such semantic redundancy at the document level are scarce. In this work we propose a deep Convolutional Neural Networks (CNN) based model to classify a document as novel or redundant with respect to a set of relevant documents already seen by the system. The system is simple and do not require any manual feature engineering. Our novel scheme encodes relevant and relative information from both source and target texts to generate an intermediate representation which we coin as the Relative Document Vector (RDV). The proposed method outperforms the existing state-of-the-art on a document-level novelty detection dataset by a margin of ∼5% in terms of accuracy. We further demonstrate the effectiveness of our approach on a standard paraphrase detection dataset where paraphrased passages closely resemble to semantically redundant documents.

In this paper we present a solution for tagging funding bodies and grants in scientific articles using a combination of trained sequential learning models, namely conditional random fields (CRF), hidden markov models (HMM) and maximum entropy models (MaxEnt), on a benchmark set created in-house. We apply the trained models to address the BioASQ challenge 5c, which is a newly introduced task that aims to solve the problem of funding information extraction from scientific articles. Results in the dry-run data set of BioASQ task 5c show that the suggested approach can achieve a micro-recall of more than 85% in tagging both funding bodies and grants.