HyperPIE: Hyperparameter Information Extraction from Scientific Publications
This work addresses the gap in making hyperparameter data machine-readable for tasks like academic search and knowledge graph construction, representing an incremental advance in information extraction for computer science publications.
The paper tackles the problem of automatically extracting hyperparameter information from scientific publications, formalizing it as an entity recognition and relation extraction task, and achieves a 29% F1 improvement over a state-of-the-art baseline for fine-tuned models and a 5.5% F1 improvement in entity recognition for large language models using YAML output.
Automatic extraction of information from publications is key to making scientific knowledge machine readable at a large scale. The extracted information can, for example, facilitate academic search, decision making, and knowledge graph construction. An important type of information not covered by existing approaches is hyperparameters. In this paper, we formalize and tackle hyperparameter information extraction (HyperPIE) as an entity recognition and relation extraction task. We create a labeled data set covering publications from a variety of computer science disciplines. Using this data set, we train and evaluate BERT-based fine-tuned models as well as five large language models: GPT-3.5, GALACTICA, Falcon, Vicuna, and WizardLM. For fine-tuned models, we develop a relation extraction approach that achieves an improvement of 29% F1 over a state-of-the-art baseline. For large language models, we develop an approach leveraging YAML output for structured data extraction, which achieves an average improvement of 5.5% F1 in entity recognition over using JSON. With our best performing model we extract hyperparameter information from a large number of unannotated papers, and analyze patterns across disciplines. All our data and source code is publicly available at https://github.com/IllDepence/hyperpie