[Most-ai-contest] 建議回覆

[闍怵羅]

各位好，在MSPE_v18_b27.py的版本中

針對蘇老師給的建議中，已經解決(劃掉線的部分)。

剩下的部分，透過“Sentence Scope”與  “Word-Matching
Indicator”，還在撰寫比較roughly的Rule，不希望針對各種題目去寫太specific的Rule出來。



Training-Set Errors:



1.      “問題：莎士比亚的喜剧作品有哪些?(在文本中)”, “問題：莎士比亚的悲剧作品有哪些?(在文本中)”, “
問題：莎士比亚的悲喜剧作品有哪些?(在文本中)”, page 7: “Angle bracket pair” formatting
information and external knowledge about the names of those scripts should
be helpful.



2.      “問題：试举出文本中五项作为生质燃料原料的粮食作物”, page 9: “Sentence Scope” information
should be helpful.



3.      “問題：生物医学所使用的奈米结晶金属是哪两种?(在本文中)”, page 10: Answer tends to be a
complete NP constitute.



4.      “問題：有哪些大众运输工具可以到达 2019 年台湾宝可梦活动会场呢?”, page 13: NE “中和新芦线” external
knowledge should be helpful



5.      “問題：根据文本的数据，男性癌症前三名是哪三种？”, page 14: “Word-Matching Indicator” (such
as 男性 and 癌症) should be helpful



6.      “問題：这起事故的肇事者所使用的交通工具为何?”, “問題：这起事故的肇事者其姓氏分别为何?”, page 15: “Sentence
Scope” information should be helpful.



7.      “問題：三级座舱设计的波音 747 客机的舱等是哪三个?”, page 15: Benchmark error.



8.      “問題：澎湖旅行团总共安排了几个景点？”, “問題：澎湖旅行团安排了哪五个景点？”, page 17: Need “Item
formatting information” (Items could be easily identified from visual
information, but not NLP).



9.      “問題：考古学家从遗址发掘中发现了人类会用什么工具来捕捞溪中的鱼蟹？”, page 18: Benchmark error.



10.  “問題：山羊老师教大家哪两种类型的垃圾需要压扁或踩扁后回收丢到资源回收筒里？”, page 19: “Sentence Scope”
information should be helpful.



11.   “問題：目前在台湾出现的肠病毒有哪几型?”, page 21: Adopted rules could be modified to
cover this case.





Test-Set Errors:



1.      “問題： 在「乌台诗案」中，有哪些人出面救了苏东坡？”, page 22: Adopted rules could be
modified to cover this case.



2.      “問題： 高雄市被列为国定古迹的建筑物是哪些?”, page 23: Adopted rules could be modified
to cover this case.



3.      “問題： 屏东县双层巴士观光服务行经路线停靠哪些景点?”, page 23: Adopted rules could be
modified to cover this case.



4.      “問題： 全民健保实施之前的健康保险有哪些?”, page 24: Adopted rules could be modified
to cover this case.



5.      “問題： 癌症分期是根据哪些原则来判断?”, page 24: Need “Item formatting information”
(Items could be easily identified from visual information, but not NLP).



6.      “問題： 小阿姨说，等春天来临时，哪些花儿会相继绽放，那十的植物元变成了万紫千红的锦绣世界？”, page 25: Adopted
rules could be modified to cover this case (e.g., matching more Q-strings).





Development-Set Errors:



1.      “問題： 「阿拉伯之春」运动发生在哪些区域？”, page 28: Adopted rules could be modified
to cover this case.



2.      “問題： 鲸鲨曾经在哪些海域出现过?”, page 30: Adopted rules could be modified to
cover this case.



3.      “問題： 菲律宾的鲸鲨保育政策为何?”, page 30: Adopted rules could be modified to
cover this case (e.g., matching more Q-strings).



4.      “問題： 开车出门前要记得做哪些车辆检查以减少交通事故发生?”, page 31: Adopted rules could be
modified to cover this case.



5.      “問題： 世界上最大的教堂有哪些建筑师与艺术家的设计?”, page 32: Adopted rules could be
modified to cover this case.



6.      “問題： 世界上最大的教堂内部有哪些艺术家的大作?(本文中)”, page 32: Benchmark error.



7.      “問題： 内政部消防署提出防范纵火方法「三从四得」，请问是哪「三从」 ?”, “問題：
内政部消防署提出防范纵火方法「三从四得」，请问是哪「四得」 ?”, page 32: Need “Item formatting
information”.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.iis.sinica.edu.tw/pipermail/most-ai-contest/attachments/20200403/3ba2f366/attachment-0001.html>