Time and Event Annotation
Our proposed annotation standard for time, events, and temporal relations (sequencing and duration) differs from various existing standards in its simplicity and efficiency, both for annotators and machine learning approaches.
The main Knowedge Graph related project that we are engaged in currently is:
- Unstructured Information to High-quality Knowledge Graph Mapping
- Mapping of Relational Databases to High-quality Knowledge Graphs using R2RML and RML, as well as JRML
- Linking of Large Heterogenous Knowledge Graphs for cross-domain and common sense Reasoning
- Temporal Graph Neural Networks for event and temporal logic modeling (see also Tie-ML)
There are various sub-projects related to this goal and some code repos are available on GitHub.
- JSON-NLP Schema
- Open Information Extraction (OpenIE)
- High Performance NLP
- Free Linguistic Environment
- JSON Probabilistic Unifier (C++, Python, Java, Rust)
- NLP RESTful Microservices (Java, Python, Rust) requires credentials, contact us
- Hoosier Semantic Information ExtractoR (HOOSIER)
- Food and Nutrients AI related to Health Knowledge Graphs
Turkic Language Resources
We are interested in corpora, NLP technologies, and Knowledge Representations for the following languages:
Information Retrieval from Medical Reports
- Mental Health NER and general NLP technologies in Medical and Health applications
- Food and Nutrients Knowledge Graphs related to Knowledge and Semantics technologies
During the summer of 2020 we engaged in an international Datathon and Hackathon on focusing on antisemitic tweets and the development of corpora and machine learning algorithms to classify tweets as antisemitic or not. See the project page for more information on this project.