Task 1 and Task 2 Output Format examples

https://github.com/WangYao-GoGoGo/ntcir19_fehu/tree/master/dataset/evaluation_test_data_examples/pred 

Task-1 output

• Task-1a: <guid, actor, l2_value>

  • guid: article id 

  • actor: actors in the news

  • l2_value: level-2 human value label

• Task-1b: <guid, actor, l1_value, direction>

  • guid: article id 

  • actor: actors in the news

  • l1_value: level-1 human value label

  • direction: aligned with/ contradictory to

Task-2 output

• Task-2a: <guid, subevent_id, actor, l2_value>

  • guid: article id 

  • subevent_id

  • actor: actors in the news

  • l2_value: level-2 human value label

• Task-1b: <guid, subevent_id, actor, l1_value, direction>

  • guid: article id 

  • subevent_id

  • actor: actors in the news

  • l1_value: level-1 human value label

  • direction: aligned with/ contradictory to

Submission：

Send one ZIP + one PDF.  

- ZIP: FEHU_[TeamID].zip; its folder structure, output-file contents and naming follow the previous instructions.  

- PDF: FEHU_[TeamID]_report.pdf; 1–2 pages outlining approach and key results.  

output.zip

     task1/

  pred_task1a.json

  pred_task1b.json      

     task2/

  pred_task2a.json

  pred_Task2b.json  

Email both to oh.gyou.tkb_gf@u.tsukuba.ac.jp.

CC:

Zhuochen Liu: s2426082@u.tsukuba.ac.jp

Jiankang Chen: s2526100@u.tsukuba.ac.jp 

*Haitao Yu: yuhaitao@slis.tsukuba.ac.jp