Biomedical Text Mining and Ontologies

The paper presents a method and tool for constructing personalized knowledgebases from scientific literature, enabling researchers to generate hypotheses and explore data efficiently.

The paper constructs a visual atlas of gene-pathology relationships by mining PubMed abstracts and using NLP to identify tissue-specific gene roles in disease.

LitRev identifies common and important biological phrases from PubMed to aid in quick literature reviews.

LEXAS is a web tool that curates biomedical experiments and suggests next experiments on genes by analyzing literature and training a machine learning model.

This paper introduces GraphPrompt, a graph-based prompt learning method for biomedical entity normalization, improving performance on zero-shot and few-shot settings using a curated dataset of entity-synonym pairs.

The paper describes the development and implementation of a vocabulary extension pipeline for the SPARC program to improve FAIR data practices in autonomic nervous system research.

The paper describes a two-stage approach combining fine-tuned BioBERT for chemical entity recognition with semantic approximate search in MeSH and PubChem for entity linking, addressing challenges in chemical text-mining.

The paper introduces an interpretable visualization approach for scientific hypotheses using semantic concept interconnectivity, network-based methods, and topic modeling, validated with a pipeline evaluated by experts.

The paper develops a hybrid system for chemical identification and indexing in full-text articles using BERT-based methods and dictionary matching, achieving high F1 scores on chemical NER and normalization.

The paper develops an ensemble model combining BERT, Sentence BERT, and T5 to predict drug-protein interactions from text, achieving an F-1 score of 0.753 on a benchmark dataset.

The paper introduces GoMi, a manually curated gold standard corpus for miRNA Named Entity Recognition, to train and test machine learning and rule-based approaches.

The paper integrates three tools to enhance automated model extension from literature, addressing issues like machine reading errors and high literature volume.

This paper introduces a contrastive approach to Literature-Based Discovery (LBD) to improve relation identification between concepts in biomedical literature, demonstrating its effectiveness through case studies on neurodegenerative diseases.

This paper develops a deep learning NERL system and a parallel oncology corpus for Portuguese and Spanish, addressing the lack of language support in biomedical text mining tools.

This study combines manual annotation and deep learning to improve name entity recognition in biomedical texts, achieving high F1-scores for gene, variant, disease, and species entities.

Klarigi identifies compositional and discriminatory ontology classes for groups of biomedical entities, providing semantic explanations that extend existing knowledge.

The paper creates a corpus for extracting diagnostic information from eye disease literature, showing strong agreement in entity annotation and potential for new insights using BioBERT.

The paper develops an NLP-ML approach to annotate -omics samples with tissue and cell-type labels using free-text metadata, outperforming existing methods and demonstrating biological relevance.

The paper describes a crowdsourcing initiative, CACAO, where undergraduates contribute gene function annotations to the Gene Ontology, enhancing coverage and enabling community-driven curation.

The paper develops a text mining pipeline to extract gene-phenotype associations from autism-related literature, creating a database that can aid in understanding autism mechanisms and treatment.

ChexMix extracts bioentity identifiers from literature to construct taxonomic trees and identify co-occurring medical subject headings.

The paper evaluates how well machine-learning models capture ontology relationships and suggests ways to improve curation using these models.

The paper introduces two hypothesis generation systems for COVID-19 research using graph mining and transformer models, achieving high-quality predictions and discovering new research findings.

The paper demonstrates the use of natural language processing to analyze plant phenotype descriptions, generating similarity networks and providing a web tool for querying gene-phenotype associations.

The paper develops word2vec-based embeddings from PubMed abstracts to capture biological relationships and improves metastatic event prediction using graph convolutional networks.

The paper develops a framework to extend ontologies using clinical text, improving their ability to characterize and classify medical data.

MACE2K extracts variant-disease association evidence from biomedical literature using machine learning models, achieving high recall and precision in identifying relevant sentences.

This paper introduces NERO, a biomedical named-entity ontology with an annotated corpus, and demonstrates its utility through text embedding for discovering associations.

DOUBLER learns unified representations of biological entities and documents to predict protein-disease relationships, outperforming existing methods when additional data modalities are informative.

SKiM is a literature-based discovery system that rediscovered biomedical knowledge and identified potential drug candidates by analyzing PubMed literature.