Papers by Alex van Grootel
Zach Jensen, Alex van Grootel MIT We present a data set of over 1 million titles from materials s... more Zach Jensen, Alex van Grootel MIT We present a data set of over 1 million titles from materials science journal articles. The data set consists of the title, journal, ISSN, year, number of times cited, DOI, and machine learned synthesis labels for each word in the sentence. This data is extracted with a natural language processing pipeline utilizing state of the art word embedding and recurrent neural network algorithms. This data can be used to analyze high level trends in materials science such as promising new research areas. We present analysis on the correlations between journals, citations, and years. We also present analysis comparing the most common target materials in the synthesis with synthesis techniques. Finally, we build a deep neural network to predict the scientific impact a paper will have based on its title.
Open source data set of materials science journal article titles and metadata.
Uploads
Papers by Alex van Grootel