Inorganic Materials Synthesis Planning with Literature-Trained Neural Networks

Kim, Edward; Jensen, Zach; Grootel, Alexander van; Huang, Kevin; Staib, Matthew; Mysore, Sheshera; Chang, Haw-Shiuan; Strubell, Emma; McCallum, Andrew; Jegelka, Stefanie; Olivetti, Elsa

doi:10.1021/acs.jcim.9b00995

Cited by 103 publications

(114 citation statements)

References 39 publications

Supporting

Mentioning

112

Contrasting

Unclassified

Order By: Relevance

“…For instance, Hawizy et al define a set of 21 types of so-called action phrases for experimental procedures from patents 16 . In the context of materials science, Huo et al interpret topics extracted by a latent Dirichlet allocation as categories of experimental steps 24 , and Kim et al cluster actions into a set of 50 categories in an automated procedure 22 .…”

Section: Resultsmentioning

confidence: 99%

Automated extraction of chemical synthesis actions from experimental procedures

et al. 2020

View full text Add to dashboard Cite

Experimental procedures for chemical synthesis are commonly reported in prose in patents or in the scientific literature. The extraction of the details necessary to reproduce and validate a synthesis in a chemical laboratory is often a tedious task requiring extensive human intervention. We present a method to convert unstructured experimental procedures written in English to structured synthetic steps (action sequences) reflecting all the operations needed to successfully conduct the corresponding chemical reactions. To achieve this, we design a set of synthesis actions with predefined properties and a deep-learning sequence to sequence model based on the transformer architecture to convert experimental procedures to action sequences. The model is pretrained on vast amounts of data generated automatically with a custom rule-based natural language processing approach and refined on manually annotated samples. Predictions on our test set result in a perfect (100%) match of the action sequence for 60.8% of sentences, a 90% match for 71.3% of sentences, and a 75% match for 82.4% of sentences.

show abstract

Section: Resultsmentioning

confidence: 99%

Automated extraction of chemical synthesis actions from experimental procedures

et al. 2020

View full text Add to dashboard Cite

show abstract

“…We believe the dataset will enable the development of robust supervised entity tagging models and is suitable for evaluating models trained to extract shallow semantic structures. This is evidenced by the adoption of the dataset by work contemporaneous with this work (Kim et al, 2018;Tamari et al, 2019).…”

Section: Discussionmentioning

confidence: 96%

“…We believe that the dataset we release fills an important gap in the existing work on extraction of inorganic materials synthesis procedures, by allowing exploration into extraction at a scale not attempted before. Parallel with this work, work by Kim et al (2018) and Tamari et al (2019) adopt the dataset released here to aid extraction of structured representations from synthesis procedures and with Kim et al presenting early experiments in synthesis planning from extracted synthesis.…”

Section: Materials Science and Chemistrymentioning

confidence: 99%

The Materials Science Procedural Text Corpus: Annotating Materials Synthesis Procedures with Shallow Semantic Structures

Mysore¹,

Jensen²,

Kim³

et al. 2019

Proceedings of the 13th Linguistic Annotation Workshop

Self Cite

View full text Add to dashboard Cite

Materials science literature contains millions of materials synthesis procedures described in unstructured natural language text. Largescale analysis of these synthesis procedures would facilitate deeper scientific understanding of materials synthesis and enable automated synthesis planning. Such analysis requires extracting structured representations of synthesis procedures from the raw text as a first step. To facilitate the training and evaluation of synthesis extraction models, we introduce a dataset of 230 synthesis procedures annotated by domain experts with labeled graphs that express the semantics of the synthesis sentences. The nodes in this graph are synthesis operations and their typed arguments, and labeled edges specify relations between the nodes. We describe this new resource in detail and highlight some specific challenges to annotating scientific text with shallow semantic structure. We make the corpus available to the community to promote further research and development of scientific information extraction systems.

show abstract

“…16 In the context of materials science, Huo et al interpret topics extracted by a latent Dirichlet allocation as categories of experimental steps, 24 and Kim et al cluster actions into a set of 50 categories in an automated procedure. 22 The actions we selected are listed in Table 1. Each action type has a set of allowed properties.…”

Section: Synthesis Actionsmentioning

confidence: 99%

“…21 More recently, data extracted with the same tools allowed machine-learning models to learn to predict the precursors and sequence of actions to synthesize inorganic materials. 22 Mysore et al apply text-mining tools to convert synthesis procedures to action graphs. 23 The nodes of the action graphs represent compounds, actions, or experimental conditions, and they are connected by edges that represent the associations between the nodes.…”

Section: Introductionmentioning

confidence: 99%

Automated Extraction of Chemical Synthesis Actions from Experimental Procedures

Vaucher¹,

Zipoli²,

Geluykens³

et al. 2020

Preprint

View full text Add to dashboard Cite

<div> <div> <div> <p>Experimental procedures for chemical synthesis are commonly reported in prose in patents or in the scientific literature. The automatic extraction of the details necessary to reproduce and validate a synthesis in a chemical laboratory is quite often a tedious task, requiring extensive human intervention. We present a method to convert unstructured experimental procedures written in English to structured synthetic steps (action sequences) reflecting all the operations needed to successfully conduct the corresponding chemical reactions. To achieve this, we design a set of synthesis actions with predefined properties and a deep-learning sequence to sequence model based on the transformer architecture to convert experimental procedures to action sequences. The model is pretrained on vast amounts of data generated automatically with a custom rule-based natural language processing approach and refined on a smaller set of manually annotated samples. Predictions on our test set resulted in a perfect (100%) match of the action sequence for 60.8% of sentences, a 90% match for 71.3% of sentences, and a 75% match for 82.4% of sentences. </p> </div> </div> </div>

show abstract

Inorganic Materials Synthesis Planning with Literature-Trained Neural Networks

Cited by 103 publications

References 39 publications

Automated extraction of chemical synthesis actions from experimental procedures

Automated extraction of chemical synthesis actions from experimental procedures

The Materials Science Procedural Text Corpus: Annotating Materials Synthesis Procedures with Shallow Semantic Structures

Automated Extraction of Chemical Synthesis Actions from Experimental Procedures

Contact Info

Product

Resources

About