Extraction of chemical structures and reactions from the literature

By Daniel M. Lowe. 2012

309 citations recorded by the source corpus.

Subfields: Physical and Theoretical Chemistry, Spectroscopy

View Paper: https://www.semanticscholar.org/paper/316189e419212794e74bf83442f65de96b5320ba

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Physical and Theoretical Chemistry

In Spectroscopy

Citations in the corpus

Improving Few- and Zero-Shot Reaction Template Prediction Using Modern Hopfield Networks (2022, methodology): “62 This dataset contains 1.8 million text-mined reaction equations in SMILES notation 21 and consists of reactions recorded in the years from 1976 to 2016.”
Modern Hopfield Networks for Few- and Zero-Shot Reaction Prediction (2021, methodology): “All data sets used in this study are derived from the publicly available U.S. patent literature (USPTO) data set by Lowe (2012).”
Molecular Transformer: A Model for Uncertainty-Calibrated Chemical Reaction Prediction (2018, methodology): “urrent model is trained end-to-end, fully data-driven and is free of chemical knowledge/rules. 3 Data Most of the publicly available reaction datasets were derived from the patent mining work of Lowe [35], where the chemical reactions were described using a text-based representation called SMILES [1”
Unsupervised Attention-Guided Atom-Mapping (2020, methodology): “The predominant use case for atom-mapping algorithms is to map heavily imbalanced reactions, such as those in patent reaction data sets (Lowe, 2012; 2017) or those predicted by data-driven reaction prediction models (Schwaller et al.”
Retrosynthetic Reaction Prediction Using Neural Sequence-to-Sequence Models (2017, methodology): “46 Given an input SMILES that represents the reactants, the model directly outputs a SMILES that represents the predicted products.”
Augmenting large language models with chemistry tools (2023, methodology): “• Train a random forest model to predict absorption max wavelength of molecules.”
“Found in Translation”: predicting outcomes of complex organic chemistry reactions using neural sequence-to-sequence models (2017, methodology): “All the openly available chemical reaction datasets were derived in some form from the patent text-mining work of Daniel M. Lowe [20].”
State-of-the-art augmented NLP transformer models for direct and single-step retrosynthesis (2020, methodology): “We used a training set filtered from USPTO database [36] containing 50k reactions classified into 10 reaction types.”
The development of models to predict melting and pyrolysis point data associated with several hundred thousand compounds mined from PATENTS (2016, methodology): “Considering that the procedure to eliminate duplicated records eliminated also molecules having ΔT ≤ 1 °C measured in different experiments, we corrected the observed distribution of values for ΔT = 0 and 1 by using the same number of counts as observed for ΔT = [2, 3] °C interval.”