Below, I released code/software I implemented.


SinoBerryPicker is a Chinese coreference resolution tool. Given raw text, an end-to-end coreference system retrieves coreference chains in the text. For text preprocess and feature generation, SinoBerryPicker relies on the following two tools:

1. Stanford CoreNLP Chinese Package
2. CRF++

Above two tools are also included in our software package for your convenience. Click here for the code.

Linguistically Aware Coreference Evaluation Tool

As all the commonly-used evaluation metrics for entity coreference resolution are linguistically agnostic, treating the mentions to be clustered as generic rather than linguistic objects. We propose a framework for incorporating linguistic awareness into commonly-used coreference evaluation metrics.

We implemented the Linguistically-Aware version for MUC, B3 and CEAF. Click here for the code.

SinoCoreferencer: An End-to-End Chinese Event Coreference Resolver

SinoCoreferencer is an end-to-end Chinese event coreference resolver. Given raw document(s) written in Chinese, it finally produces event coreference chains within each document. Additional, it also outputs some by-products: (1) entity mentions; (2) semantic type and subtype of entity mentions; (3) time and value expressions; (4) entity coreference chains; (5) events including triggers, trigger type, arguments and arguments role; and (6) the polarity, modality, genericity and tense of the events. Click here for the code.

For any questions or bugs, you are welcome to contact yzcchen at hlt dot utdallas dot edu.
[Home Page]