I have a random idea: when I tried playing with QuickDraw, it asked me to draw Penguin, but in my head I was thinking Dolphin (don't know why I often mix up these 2 words). As a non-native speaker, I made that simple mistake. Which makes me wonder if people from other non-English-speaking countries might have the same misunderstanding problem, especially with difficult categories like: asparagus, chandelier, fire hydrant, hedgehog, hockey puck, octagon, popsicle, string bean (I have no idea what that is) etc., so maybe we should give more weight to samples from native English speaking countries or countries with high level of English? Just a thought.
", "messages": [ { "id": 415475, "postDate": "2018-11-05T07:39:48.947Z", "content": "I think Embedding will be your answer.
Tried that. Adding country to the model didn't help at all. :-/
", "rawMarkdown": "Tried that. Adding country to the model didn't help at all. :-/", "votes": 1 }, { "id": 415496, "postDate": "2018-11-05T08:22:07.093Z", "content": "I have the same result as you. I can explain that maybe my base model is too strong and it does not need country code features. Why dont try with the weaker model? I belive that it will help.
", "rawMarkdown": "I have the same result as you. I can explain that maybe my base model is too strong and it does not need country code features. Why dont try with the weaker model? I belive that it will help.", "votes": 1 } ] }, { "id": 415425, "postDate": "2018-11-05T05:23:55.790Z", "content": "Pants have a very different meaning in the UK & US. I'll be trying to use this info in my model if I have time!
", "rawMarkdown": "Pants have a very different meaning in the UK & US. I'll be trying to use this info in my model if I have time!", "votes": 1 }, { "id": 415422, "postDate": "2018-11-05T05:21:04.657Z", "content": "I have a random idea: when I tried playing with QuickDraw, it asked me to draw Penguin, but in my head I was thinking Dolphin (don't know why I often mix up these 2 words). As a non-native speaker, I made that simple mistake. Which makes me wonder if people from other non-English-speaking countries might have the same misunderstanding problem, especially with difficult categories like: asparagus, chandelier, fire hydrant, hedgehog, hockey puck, octagon, popsicle, string bean (I have no idea what that is) etc., so maybe we should give more weight to samples from native English speaking countries or countries with high level of English? Just a thought.
", "rawMarkdown": "I have a random idea: when I tried playing with QuickDraw, it asked me to draw Penguin, but in my head I was thinking Dolphin (don't know why I often mix up these 2 words). As a non-native speaker, I made that simple mistake. Which makes me wonder if people from other non-English-speaking countries might have the same misunderstanding problem, especially with difficult categories like: asparagus, chandelier, fire hydrant, hedgehog, hockey puck, octagon, popsicle, string bean (I have no idea what that is) etc., so maybe we should give more weight to samples from native English speaking countries or countries with high level of English? Just a thought. ", "votes": 2 }, { "id": 416314, "postDate": "2018-11-06T14:29:57.153Z", "content": "yep, I also thought about this idea before, but it did not improve the LB score.
", "rawMarkdown": "yep, I also thought about this idea before, but it did not improve the LB score." } ], "comments": [ { "id": 415475, "author_name": "cab", "author_url": "", "post_date": "2018-11-05T07:39:48.947000", "content": "I think Embedding will be your answer.
Tried that. Adding country to the model didn't help at all. :-/
", "votes": 1, "replies": [] }, { "id": 415496, "author_name": "cab", "author_url": "", "post_date": "2018-11-05T08:22:07.093000", "content": "I have the same result as you. I can explain that maybe my base model is too strong and it does not need country code features. Why dont try with the weaker model? I belive that it will help.
", "votes": 1, "replies": [] } ] }, { "id": 415425, "author_name": "datasaurus", "author_url": "", "post_date": "2018-11-05T05:23:55.790000", "content": "Pants have a very different meaning in the UK & US. I'll be trying to use this info in my model if I have time!
", "votes": 1, "replies": [] }, { "id": 416314, "author_name": "Gary", "author_url": "", "post_date": "2018-11-06T14:29:57.153000", "content": "yep, I also thought about this idea before, but it did not improve the LB score.
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "415475": "I think `Embedding` will be your answer.", "415425": "Pants have a very different meaning in the UK & US. I'll be trying to use this info in my model if I have time!", "415422": "I have a random idea: when I tried playing with QuickDraw, it asked me to draw Penguin, but in my head I was thinking Dolphin (don't know why I often mix up these 2 words). As a non-native speaker, I made that simple mistake. Which makes me wonder if people from other non-English-speaking countries might have the same misunderstanding problem, especially with difficult categories like: asparagus, chandelier, fire hydrant, hedgehog, hockey puck, octagon, popsicle, string bean (I have no idea what that is) etc., so maybe we should give more weight to samples from native English speaking countries or countries with high level of English? Just a thought. ", "416314": "yep, I also thought about this idea before, but it did not improve the LB score." } }