google/sling
A neural-network-based natural language parser that produces frame semantic representations for text, used for knowledge base completion from Wikipedia articles.

SLING is a natural language frame semantic parser that uses neural networks to directly produce frame semantic representations of text without explicit intermediate linguistic structures. It includes a JIT compiler for efficient parsing, a frame store for representing knowledge graphs, and a pipeline (Wiki flow) for processing Wikidata and Wikipedia dumps. The system is designed to read and understand Wikipedia articles in multiple languages to extract and complete knowledge bases.
Frequently asked
- What is google/sling?
- A neural-network-based natural language parser that produces frame semantic representations for text, used for knowledge base completion from Wikipedia articles.
- Is sling open source?
- Yes — google/sling is open source, released under the Apache-2.0 license.
- What language is sling written in?
- google/sling is primarily written in C++.
- How popular is sling?
- google/sling has 1.9k stars on GitHub.
- Where can I find sling?
- google/sling is on GitHub at https://github.com/google/sling.