Are there multiple graphemes for the same combination of grapheme_root, vowel_diacritic, and consonant_diacritic?
For example, in train.csv\n\nimage_id=378 : (64, 3, 2) : র্তী\nimage_id=1482 : (64, 3, 2) : র্ত্রী\n\nThere are 1295 types of grapheme in train.csv, but there are only 1292 combinations.\nPlease tell me about Bengali Grapheme.
I've reported this problem and got a response from one of competetion hosts saying that a fix is expected recently.\nJust check this link: (hope it help)\nhttps://www.kaggle.com/c/bengaliai-cv19/discussion/123002#728098
", "rawMarkdown": "I've reported this problem and got a response from one of competetion hosts saying that a fix is expected recently.\nJust check this link: (hope it help)\nhttps://www.kaggle.com/c/bengaliai-cv19/discussion/123002#728098", "votes": 1, "replies": [ { "id": 730425, "postDate": "2020-01-27T13:24:23.183Z", "content": "Hi AllenChangTW.\nThank you for the information. I was missing valuable information.
\nBengali characters are very complicated, but 漢字(kanji) used in Japan and China is also a character that combines 'hen', 'kanmuri', 'tsukuri', etc. It has similarities with Bengali characters and has a close affinity.
You are right there are 1295 distinct graphemes but only 1291 combinations of the 3 components. \nBut you can notice that all the 3 components are kind of the same, just a little different shape? Maybe there was some sort of a change in the fonts they used while preparing the dataset.
\n\n| | image_id | grapheme_root | vowel_diacritic | consonant_diacritic | grapheme | mix |\n|-----:|:-----------|----------------:|------------------:|----------------------:|:-----------|-------:|\n| 387 | Train_387 | 64 | 3 | 2 | র্তী | 64_3_2 |\n| 1482 | Train_1482 | 64 | 3 | 2 | র্ত্রী | 64_3_2 |\n| 577 | Train_577 | 64 | 7 | 2 | র্তে | 64_7_2 |\n| 1874 | Train_1874 | 64 | 7 | 2 | র্ত্রে | 64_7_2 |\n| 532 | Train_532 | 72 | 0 | 2 | র্দ্র | 72_0_2 |\n| 1117 | Train_1117 | 72 | 0 | 2 | র্দ | 72_0_2 |
\n\n| | component_type | label | component |\n|----:|:--------------------|--------:|:------------|\n| 64 | grapheme_root | 64 | ত |\n| 72 | grapheme_root | 72 | দ |\n| 171 | vowel_diacritic | 3 | ী |\n| 175 | vowel_diacritic | 7 | ে |\n| 168 | vowel_diacritic | 0 | 0 |\n| 181 | consonant_diacritic | 2 | র্ |
", "rawMarkdown": "You are right there are 1295 distinct graphemes but only 1291 combinations of the 3 components. \nBut you can notice that all the 3 components are kind of the same, just a little different shape? Maybe there was some sort of a change in the fonts they used while preparing the dataset.\n\n| | image_id | grapheme_root | vowel_diacritic | consonant_diacritic | grapheme | mix |\n|-----:|:-----------|----------------:|------------------:|----------------------:|:-----------|-------:|\n| 387 | Train_387 | 64 | 3 | 2 | র্তী | 64_3_2 |\n| 1482 | Train_1482 | 64 | 3 | 2 | র্ত্রী | 64_3_2 |\n| 577 | Train_577 | 64 | 7 | 2 | র্তে | 64_7_2 |\n| 1874 | Train_1874 | 64 | 7 | 2 | র্ত্রে | 64_7_2 |\n| 532 | Train_532 | 72 | 0 | 2 | র্দ্র | 72_0_2 |\n| 1117 | Train_1117 | 72 | 0 | 2 | র্দ | 72_0_2 |\n\n| | component_type | label | component |\n|----:|:--------------------|--------:|:------------|\n| 64 | grapheme_root | 64 | ত |\n| 72 | grapheme_root | 72 | দ |\n| 171 | vowel_diacritic | 3 | ী |\n| 175 | vowel_diacritic | 7 | ে |\n| 168 | vowel_diacritic | 0 | 0 |\n| 181 | consonant_diacritic | 2 | র্ |\n", "votes": 1 }, { "id": 728979, "postDate": "2020-01-25T14:31:08.313Z", "content": "Are there multiple graphemes for the same combination of grapheme_root, vowel_diacritic, and consonant_diacritic?
For example, in train.csv\n\nimage_id=378 : (64, 3, 2) : র্তী\nimage_id=1482 : (64, 3, 2) : র্ত্রী\n\nThere are 1295 types of grapheme in train.csv, but there are only 1292 combinations.\nPlease tell me about Bengali Grapheme.
Hi Datta!\nর্তী and র্ত্রী are different typefaces, but they are the same character.\nThank you very much.
", "rawMarkdown": "Hi Datta!\nর্তী and র্ত্রী are different typefaces, but they are the same character.\nThank you very much.", "replies": [ { "id": 730603, "postDate": "2020-01-27T17:13:29.823Z", "content": "I thought so too, but have a look here. https://www.kaggle.com/c/bengaliai-cv19/discussion/123859
", "rawMarkdown": "I thought so too, but have a look here. https://www.kaggle.com/c/bengaliai-cv19/discussion/123859" } ] }, { "id": 729000, "postDate": "2020-01-25T15:04:04.910Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 730068, "author_name": "AllenChangTW", "author_url": "", "post_date": "2020-01-27T04:00:34.220000", "content": "I've reported this problem and got a response from one of competetion hosts saying that a fix is expected recently.\nJust check this link: (hope it help)\nhttps://www.kaggle.com/c/bengaliai-cv19/discussion/123002#728098
", "votes": 1, "replies": [ { "id": 730425, "author_name": "K.Amano", "author_url": "", "post_date": "2020-01-27T13:24:23.183000", "content": "Hi AllenChangTW.\nThank you for the information. I was missing valuable information.
\nBengali characters are very complicated, but 漢字(kanji) used in Japan and China is also a character that combines 'hen', 'kanmuri', 'tsukuri', etc. It has similarities with Bengali characters and has a close affinity.
You are right there are 1295 distinct graphemes but only 1291 combinations of the 3 components. \nBut you can notice that all the 3 components are kind of the same, just a little different shape? Maybe there was some sort of a change in the fonts they used while preparing the dataset.
\n\n| | image_id | grapheme_root | vowel_diacritic | consonant_diacritic | grapheme | mix |\n|-----:|:-----------|----------------:|------------------:|----------------------:|:-----------|-------:|\n| 387 | Train_387 | 64 | 3 | 2 | র্তী | 64_3_2 |\n| 1482 | Train_1482 | 64 | 3 | 2 | র্ত্রী | 64_3_2 |\n| 577 | Train_577 | 64 | 7 | 2 | র্তে | 64_7_2 |\n| 1874 | Train_1874 | 64 | 7 | 2 | র্ত্রে | 64_7_2 |\n| 532 | Train_532 | 72 | 0 | 2 | র্দ্র | 72_0_2 |\n| 1117 | Train_1117 | 72 | 0 | 2 | র্দ | 72_0_2 |
\n\n| | component_type | label | component |\n|----:|:--------------------|--------:|:------------|\n| 64 | grapheme_root | 64 | ত |\n| 72 | grapheme_root | 72 | দ |\n| 171 | vowel_diacritic | 3 | ী |\n| 175 | vowel_diacritic | 7 | ে |\n| 168 | vowel_diacritic | 0 | 0 |\n| 181 | consonant_diacritic | 2 | র্ |
", "votes": 1, "replies": [] }, { "id": 729058, "author_name": "K.Amano", "author_url": "", "post_date": "2020-01-25T17:14:34.183000", "content": "Hi Datta!\nর্তী and র্ত্রী are different typefaces, but they are the same character.\nThank you very much.
", "votes": 0, "replies": [ { "id": 730603, "author_name": "datta", "author_url": "", "post_date": "2020-01-27T17:13:29.823000", "content": "I thought so too, but have a look here. https://www.kaggle.com/c/bengaliai-cv19/discussion/123859
", "votes": 0, "replies": [] } ] }, { "id": 729000, "author_name": "", "author_url": "", "post_date": "2020-01-25T15:04:04.910000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "730068": "I've reported this problem and got a response from one of competetion hosts saying that a fix is expected recently.\nJust check this link: (hope it help)\nhttps://www.kaggle.com/c/bengaliai-cv19/discussion/123002#728098", "729028": "You are right there are 1295 distinct graphemes but only 1291 combinations of the 3 components. \nBut you can notice that all the 3 components are kind of the same, just a little different shape? Maybe there was some sort of a change in the fonts they used while preparing the dataset.\n\n| | image_id | grapheme_root | vowel_diacritic | consonant_diacritic | grapheme | mix |\n|-----:|:-----------|----------------:|------------------:|----------------------:|:-----------|-------:|\n| 387 | Train_387 | 64 | 3 | 2 | র্তী | 64_3_2 |\n| 1482 | Train_1482 | 64 | 3 | 2 | র্ত্রী | 64_3_2 |\n| 577 | Train_577 | 64 | 7 | 2 | র্তে | 64_7_2 |\n| 1874 | Train_1874 | 64 | 7 | 2 | র্ত্রে | 64_7_2 |\n| 532 | Train_532 | 72 | 0 | 2 | র্দ্র | 72_0_2 |\n| 1117 | Train_1117 | 72 | 0 | 2 | র্দ | 72_0_2 |\n\n| | component_type | label | component |\n|----:|:--------------------|--------:|:------------|\n| 64 | grapheme_root | 64 | ত |\n| 72 | grapheme_root | 72 | দ |\n| 171 | vowel_diacritic | 3 | ী |\n| 175 | vowel_diacritic | 7 | ে |\n| 168 | vowel_diacritic | 0 | 0 |\n| 181 | consonant_diacritic | 2 | র্ |\n", "728979": "Are there multiple graphemes for the same combination of `grapheme_root`, `vowel_diacritic`, and `consonant_diacritic`?\n\nFor example, in train.csv\n```\nimage_id=378 : (64, 3, 2) : র্তী\nimage_id=1482 : (64, 3, 2) : র্ত্রী\n```\nThere are **1295** types of grapheme in train.csv, but there are only **1292** combinations.\nPlease tell me about Bengali Grapheme.", "729058": "Hi Datta!\nর্তী and র্ত্রী are different typefaces, but they are the same character.\nThank you very much.", "729000": "" } }