INCEpTION
Open-source, machine-assisted text annotation from TU Darmstadt's UKP Lab.
About
INCEpTION is an open-source annotation platform developed by the Ubiquitous Knowledge Processing (UKP) Lab at TU Darmstadt and funded by the German Research Foundation. It combines text annotation with machine assistance: recommenders suggest annotations inline, active learning picks what to label next, and entity and fact linking build a knowledge base as you annotate. It imports plain text, PDF, HTML, and TEI, and exports UIMA CAS XMI, JSON, or CoNLL-U, so corpora stay usable elsewhere. It is widely used across NLP, linguistics, digital humanities, and clinical research, including the GeMTeX German medical-text project across six university hospitals. The GitHub repo is active under Apache-2.0, with example project templates and a maintained hosted demo. It is one of the strongest free options for semantic annotation.
What it can do
- Inline ML recommenders that suggest annotations
- Active learning for label prioritization
- Entity and fact linking with knowledge-base building
- Custom annotation layers and editors
- Import of text, PDF, HTML, TEI; export to UIMA XMI, JSON, CoNLL-U
- Example project templates
Setup
Download the release and run the standalone JAR, or use Docker; hosted demo available.
Prerequisites
Java 17+ for self-hosting. Nothing for the hosted demo.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (metrics tier, metrics-v1). Evidence: 725 GitHub stars and 171 forks; commits as of October 2026; DFG-funded university lab; deployed in the six-hospital GeMTeX project. Strengths: fully open Apache-2.0 code, unique recommender plus knowledge-base design, deep academic adoption. Docked for: (1) research software with academic support rhythms; (2) Java-based self-hosting is heavier than SaaS alternatives. License: Apache-2.0.
★ 725 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 9.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · UKP Lab, TU Darmstadt; DFG-funded; inception-project/inception with 725 stars, 171 forks; commits current as of Oct 2026.
- · Recommenders, active learning, entity/fact linking with knowledge-base building; UIMA XMI/JSON/CoNLL-U export.
- · Deployed in GeMTeX across six German university hospitals.
- · Manual review: one of the strongest free semantic-annotation options; Java self-hosting weight noted.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.