Resources & Demos

Open-source software

Hy-NLI is a hybrid NLI engine that computes inference in a hybrid way by employing the symbolic system GKR4NLI and the deep learning model BERT.
XplaiNLI focuses on the explainability of the Hy-NLI system and provides sketches of explanations for the decisions made by each component of Hy-NLI .
GKR4NLI is a symbolic NLI engine that computes the inference relation between a pair of sentences based on Natural Logic and GKR .
GKR parser transforms a given sentence into a layered semantic graph, suitable for the tasks of Natural Language Inference (NLI) and semantic similarity. The graphs can also be integrated in any application as semantic representations.
SICK processing contains all work undertaken to (semi-automatically) correct and re-annotate the SICK corpus.


GKR Demo gives a taste of what GKR looks like.
Hy-NLI Demo offers an overview of the hybrid inference system.
XplaiNLI Demo shares the visualization interface of XplaiNLI.
ParHistVis is an online tool for the visualization of parallel multilingual historical corpora.

Freely-available Corpora

CoUSBi consists of all enacted, legislative bills related to education for the years 2007 to 2015 from the US states North Carolina and New Mexico. The bills are formatted according to the TEI standard and are also annotated for Named Entities.
KRoQ : a parallel, multilingual corpus of French, Spanish and Greek Bible translations. The questions included in the text are annotated for their type, i.e. whether they are information-seeking or non-information seeking (e.g. rhetorical). The corpus can be used for training a question-type classifier or for the theoretical study of the annotated questions.

* * * Check out the relevant publications for more details on these resources and demos. * * *