google/sling

Health v2 · maintenance only

Flags: archived

Adoption not part of the score

1930 stars · 264 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

SLING is a natural language frame semantics parser that annotates text with frame semantic graph representations using bi-directional LSTMs and a transition-based decoder. It includes a scalable frame store for knowledge representation, a Wiki flow pipeline for processing Wikidata and Wikipedia dumps, and a neural network JIT compiler for fast training and parsing.

Use cases

• parse text into frame semantic representations
• extract facts from Wikipedia articles for knowledge base completion
• convert Wikidata dumps into an in-memory frame graph
• map entity names and phrases to knowledge base entities
• train a custom semantic parser on annotated documents
• upload extracted facts to Wikidata via a bot

When to choose

• you need frame semantic parsing with direct graph output from raw text
• you are doing knowledge base completion with Wikidata or Wikipedia data
• you want an efficient in-memory frame store for graph traversal and reasoning
• you need a self-contained neural parser with a JIT compiler for fast inference

When to avoid

• you need a general-purpose production NLP pipeline with broad language support
• you want a maintained project with active development and support
• you need simple off-the-shelf NER or dependency parsing rather than frame semantics
• your stack requires a pure Python or managed-language solution

Facets

library · maturity maintenance

nlp machine-learning parser llm-training data-science machine-learning cpp python frame-semantics knowledge-base-completion wikidata wikipedia neural-network-parser frame-store jit-compiler natural-language-processing knowledge-graphs linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="google/sling")