Document Type


Original Publication Date


Journal/Book/Conference Title

The 12th International Workshop on Semantic Evaluation

Date of Submission

October 2018


The Chrono System: Chrono is a hybrid rule-based and machine learning system written in Python and built from the ground up to identify temporal expressions in text and normalizes them into the SCATE schema. Input text is preprocessed using Python’s NLTK package, and is run through each of the four primary modules highlighted here. Note that Chrono does not remove stopwords because they add temporal information and context, and Chrono does not tokenize sentences. Output is an Anafora XML file with annotated SCATE entities. After minor parsing logic adjustments, Chrono has emerged as the top performing system for SemEval 2018 Task 6. Chrono is available on GitHub at

Future Work: Chrono is still under development. Future improvements will include: additional entity parsing, like “event”; evaluating the impact of sentence tokenization; implement an ensemble ML module that utilizes all four ML methods for disambiguation; extract temporal phrase parsing algorithm to be stand-alone and compare to similar systems; evaluate performance on THYME medical corpus; migrate to UIMA framework and implement Ruta Rules for portability and easier customization.

Is Part Of

VCU Computer Science Publications