Shared Task on Short Text Conversation
This repository provides the datasets for NTCIR STC (Short Text Conversation) Japanese subtask. For the details of the shared task, refer to the STC homepage.
Provided datasets
taskdata: Data from which tweets are to be retrieved. Provided tweets were randomly sampled from all the tweets in 2014 (1 million tweets; .5 million tweet pairs).devset: Development data. For each input tweet, a baseline system retrieved tweets fromtaskdata. Each tweet pair is labeled by 10 annotators.
Tweets are provided as tweet IDs (id_str). Download original texts using twitter API, or buy the data by contacting the following address:
- Email: info
<at>nazuki-oto.com - Dataset name: NTCIR STC Japanese subtask data set
Schedule
- Release of the Twitter data and starts accepting registration: Nov 01, 2015
- Release of the development data:
Nov 20, 2015Nov 24, 2015 - Registration deadline of STC Japanese task: Jan 15, 2016
- Release of the test data: Feb 15, 2016
- Formal run deadline: Feb 22, 2016
- Distribution of evaluation results: Mar 10, 2016
- Paper draft deadline: Mar 20, 2016
- (brief review of the draft papers)
- Camera ready deadline: May 1, 2016
- NTCIR-12 conference: Jun 7-10, 2016
Change log
- Nov. 1, 2015: Twitter data (task data) released
- Nov. 24, 2015:
- Task data fixed (some pairs are randomly removed so that the task data contains exactly one million tweets)
- Development data released
- Feb. 8, 2016:
- The list of deleted tweets released
- The rules for formal run submission released
- Feb. 15, 2016: Test data released
- Jun. 27, 2016:
- Test data and annotated labels released