SCINOBO: a novel system classifying scholarly communication in a dynamically constructed hierarchical Field-of-Science taxonomy.

Classifying scientific publications according to Field-of-Science taxonomies is of crucial importance, powering a wealth of relevant applications including Search Engines, Tools for Scientific Literature, Recommendation Systems, and Science Monitoring. Furthermore, it allows funders, publishers, scholars, companies, and other stakeholders to organize scientific literature more effectively, calculate impact indicators along Science Impact pathways and identify emerging topics that can also facilitate Science, Technology, and Innovation policy-making. As a result, existing classification schemes for scientific publications underpin a large area of research evaluation with several classification schemes currently in use. However, many existing schemes are domain-specific, comprised of few levels of granularity, and require continuous manual work, making it hard to follow the rapidly evolving landscape of science as new research topics emerge. Based on our previous work of scinobo, which incorporates metadata and graph-based publication bibliometric information to assign Field-of-Science fields to scientific publications, we propose a novel hybrid approach by further employing Neural Topic Modeling and Community Detection techniques to dynamically construct a Field-of-Science taxonomy used as the backbone in automatic publication-level Field-of-Science classifiers. Our proposed Field-of-Science taxonomy is based on the OECD fields of research and development (FORD) classification, developed in the framework of the Frascati Manual containing knowledge domains in broad (first level(L1), one-digit) and narrower (second level(L2), two-digit) levels. We create a 3-level hierarchical taxonomy by manually linking Field-of-Science fields of the sciencemetrix Journal classification to the OECD/FORD level-2 fields. To facilitate a more fine-grained analysis, we extend the aforementioned Field-of-Science taxonomy to level-4 and level-5 fields by employing a pipeline of AI techniques. We evaluate the coherence and the coverage of the Field-of-Science fields for the two additional levels based on synthesis scientific publications in two case studies, in the knowledge domains of Energy and Artificial Intelligence. Our results showcase that the proposed automatically generated Field-of-Science taxonomy captures the dynamics of the two research areas encompassing the underlying structure and the emerging scientific developments.

A novel tool that allows interactive screening of PubMed citations showed promise for the semi-automation of identification of Biomedical Literature.

BACKGROUND AND OBJECTIVES: Systematic reviews form the basis of evidence-based medicine, but are expensive and time-consuming to produce. To address this burden, we have developed a literature identification system (Pythia) that combines the query formulation and citation screening steps. METHODS: Pythia incorporates a set of natural-language questions with machine-learning algorithms to rank all PubMed citations based on relevance, returning the 100 top-ranked citations for human screening. The tagged citations are iteratively exploited by Pythia to refine the search and re-rank the citations. RESULTS: Across seven systematic reviews, the ability of Pythia to identify the relevant citations (sensitivity) ranged from 0.09 to 0.58. The number of abstracts reviewed per relevant abstract number needed to read (NNR) was lower than in the manually screened project in four reviews, higher in two, and had mixed results in one. The reviews that had greater overall sensitivity retrieved more relevant citations in early batches, but retrieval was generally unaffected by other aspects, such as study design, study size, and specific key question. CONCLUSION: Due to its low sensitivity, Pythia is not ready for widespread use. Future research should explore ways to encode domain knowledge in query formulation to better enrich the questions used in the search.

Hierarchical bi-directional attention-based RNNs for supporting document classification on protein-protein interactions affected by genetic mutations.

In this paper, we describe a hierarchical bi-directional attention-based Re-current Neural Network (RNN) as a reusable sequence encoder architecture, which is used as sentence and document encoder for document classification. The sequence encoder is composed of two bi-directional RNN equipped with an attention mechanism that identifies and captures the most important elements, words or sentences, in a document followed by a dense layer for the classification task. Our approach utilizes the hierarchical nature of documents which are composed of sequences of sentences and sentences are composed of sequences of words. In our model, we use word embeddings to project the words to a low-dimensional vector space. We leverage word embeddings trained on PubMed for initializing the embedding layer of our network. We apply this model to biomedical literature specifically, on paper abstracts published in PubMed. We argue that the title of the paper itself usually contains important information more salient than a typical sentence in the abstract. For this reason, we propose a shortcut connection that integrates the title vector representation directly to the final feature representation of the document. We concatenate the sentence vector that represents the title and the vectors of the abstract to the document feature vector used as input to the task classifier. With this system we participated in the Document Triage Task of the BioCreative VI Precision Medicine Track and we achieved 0.6289 Precision, 0.7656 Recall and 0.6906 F1-score with the Precision and F1-score be the highest ranking first among the other systems.Database URL: https://github.com/afergadis/BC6PM-HRNN.