Our favicon is adapted from Taxonomy icons created by RanuKumbolo.lab - Flaticon
Synoptic Key of Life — Semantic Search for Mycological Taxonomy
SKOL is a Synoptic Key. If you've used a nature guide, you've likely encountered a dichotomous key. This is a list of paired choices that lead the user to identify a species based on observable characteristics. A synoptic key, on the other hand, allows users to start their search from any characteristic they observe, making it more flexible and user-friendly.
There are good synoptic keys on the Internet for various groups of organisms, including some good ones for fungi. SKOL aims to be a synoptic key for all fungi described in the open access mycological taxonomic literature. We want the machines to do the hard work of reading and understanding the literature, so that human users can simply describe what they see and find matching species descriptions.
SKOL guides you through creating a technical description of the organism you have in front of you. Your description becomes part of the synoptic key, accessible to other users searching for similar organisms. Along the way, SKOL connects you to relevant taxonomic literature, helping you learn more about the species you are studying.
SKOL applies machine learning and natural language processing techniques to the challenge of searching and navigating open mycological taxonomic literature. The project aims to make the vast corpus of fungal taxonomy publications more accessible to researchers, students, and enthusiasts through semantic search capabilities.
Our Goal: Enable researchers to find relevant taxonomic descriptions by searching with natural language descriptions rather than exact keyword matches, leveraging the power of modern text embeddings and semantic similarity.
Mycological taxonomy literature spans centuries of scientific publications across numerous journals and books. Traditional keyword-based searches often fail to connect researchers with relevant descriptions because:
SKOL uses a multi-stage machine learning pipeline to process and index taxonomic literature:
The search functionality is powered by MycoSearch, a fork of Dr. Draft's SOTA Literature Search system. MycoSearch provides:
SKOL started by indexing taxonomic literature from major mycological journals:
The training dataset includes manually labeled journal issues with paragraph-level annotations, enabling supervised learning for the classification models.
View detailed statistics about all ingestion sources, including publication information and record counts, on our Sources page.
SKOL originated in 2019 as a personal research project by La Monte H.P. Yarroll, without a clear path to final implementation. La Monte started a Masters program at the Syracuse University School of Information Studies a few years later to gain the necessary skills. Each class included a group project. For several classes, La Monte recruited fellow students to work on aspects of SKOL. As a capstone, La Monte did an independent study with Dr. Gregory Block to create this website.
Key milestones include the development of paragraph classification models, integration of modern transformer-based embeddings, and the creation of this web interface for public access to the semantic search capabilities.
SKOL is open source software released under the GNU General Public License v3 (GPLv3). The source code is available on GitHub: