Guys This task is not related to this competition but i thought this discussion panel would help me.\nI have a task to create text classification system for Indonesian Tweets. 2 things i want to know:-\n1. Can I use Bert Base Uncased multilingual for creating embeddings and should i Use it or anything else should i use?\n2. How can I extract features from this data(my professor wants) me to do this additionally.\ni have 2 columns ( comment , sentiment) for 8000 tweets.
", "messages": [ { "id": "825300", "postDate": "04/28/2020 23:13:12", "content": "Guys This task is not related to this competition but i thought this discussion panel would help me.\nI have a task to create text classification system for Indonesian Tweets. 2 things i want to know:-\n1. Can I use Bert Base Uncased multilingual for creating embeddings and should i Use it or anything else should i use?\n2. How can I extract features from this data(my professor wants) me to do this additionally.\ni have 2 columns ( comment , sentiment) for 8000 tweets.
", "rawMarkdown": "Guys This task is not related to this competition but i thought this discussion panel would help me.\nI have a task to create text classification system for Indonesian Tweets. 2 things i want to know:-\n1. Can I use Bert Base Uncased multilingual for creating embeddings and should i Use it or anything else should i use?\n2. How can I extract features from this data(my professor wants) me to do this additionally.\ni have 2 columns ( comment , sentiment) for 8000 tweets.", "votes": null }, { "id": "825324", "postDate": "04/28/2020 23:53:22", "content": "Try to use at first TF-IDF it works with most languages.
", "rawMarkdown": "Try to use at first TF-IDF it works with most languages.", "votes": null }, { "id": "825334", "postDate": "04/29/2020 00:10:59", "content": "OK\nAnd how to tokenise Indonesian @hamditarek
", "rawMarkdown": "OK\nAnd how to tokenise Indonesian @hamditarek", "votes": null }, { "id": "825902", "postDate": "04/29/2020 10:11:35", "content": "Hi Sahib,\nYes, you can use Bert Base Uncased Multilingual as it was trained with Indonesian text data too. If you want a more powerful model, you could also use the latest multilingual model XLM-R.
", "rawMarkdown": "Hi Sahib,\nYes, you can use Bert Base Uncased Multilingual as it was trained with Indonesian text data too. If you want a more powerful model, you could also use the latest multilingual model XLM-R.", "votes": null }, { "id": "825921", "postDate": "04/29/2020 10:28:17", "content": "Buddy thank you so much @ilhamfp31\nCan you suggest any tutorial for the same \nThat will be very helpful
", "rawMarkdown": "Buddy thank you so much @ilhamfp31\nCan you suggest any tutorial for the same \nThat will be very helpful", "votes": null }, { "id": "825928", "postDate": "04/29/2020 10:34:17", "content": "My current undergraduate final project is on Indonesian text classification using multilingual language model like XLM-R and mBERT! I'll publish it on GitHub tomorrow. Right now I'm still finishing it :D
\n\nIn the meantime, you can look at kernel such as this one.
", "rawMarkdown": "My current undergraduate final project is on Indonesian text classification using multilingual language model like XLM-R and mBERT! I'll publish it on GitHub tomorrow. Right now I'm still finishing it :D\n\nIn the meantime, you can look at kernel such as [this one](https://www.kaggle.com/xhlulu/jigsaw-tpu-xlm-roberta).", "votes": null }, { "id": "826022", "postDate": "04/29/2020 11:49:51", "content": "Thank you so much @ilhamfp31 ill be looking forward to your reply
", "rawMarkdown": "Thank you so much @ilhamfp31 ill be looking forward to your reply", "votes": null }, { "id": "914639", "postDate": "07/04/2020 05:41:05", "content": "Hi! Sorry for the late reply, here is my GitHub source code: https://github.com/ilhamfp/indonesian-text-classification-multilingual
", "rawMarkdown": "Hi! Sorry for the late reply, here is my GitHub source code: https://github.com/ilhamfp/indonesian-text-classification-multilingual", "votes": null }, { "id": "915851", "postDate": "07/05/2020 06:31:52", "content": "thanks bro @ilhamfp31
", "rawMarkdown": "thanks bro @ilhamfp31", "votes": null } ], "comments": [ { "id": 825324, "author_name": "hamditarek", "author_url": "", "post_date": "04/28/2020 23:53:22", "content": "Try to use at first TF-IDF it works with most languages.
", "votes": null, "replies": [ { "id": 825334, "author_name": "sahib12", "author_url": "", "post_date": "04/29/2020 00:10:59", "content": "OK\nAnd how to tokenise Indonesian @hamditarek
", "votes": null, "replies": [] } ] }, { "id": 825902, "author_name": "ilhamfp31", "author_url": "", "post_date": "04/29/2020 10:11:35", "content": "Hi Sahib,\nYes, you can use Bert Base Uncased Multilingual as it was trained with Indonesian text data too. If you want a more powerful model, you could also use the latest multilingual model XLM-R.
", "votes": null, "replies": [ { "id": 825921, "author_name": "sahib12", "author_url": "", "post_date": "04/29/2020 10:28:17", "content": "Buddy thank you so much @ilhamfp31\nCan you suggest any tutorial for the same \nThat will be very helpful
", "votes": null, "replies": [] }, { "id": 825928, "author_name": "ilhamfp31", "author_url": "", "post_date": "04/29/2020 10:34:17", "content": "My current undergraduate final project is on Indonesian text classification using multilingual language model like XLM-R and mBERT! I'll publish it on GitHub tomorrow. Right now I'm still finishing it :D
\n\nIn the meantime, you can look at kernel such as this one.
", "votes": null, "replies": [] }, { "id": 826022, "author_name": "sahib12", "author_url": "", "post_date": "04/29/2020 11:49:51", "content": "Thank you so much @ilhamfp31 ill be looking forward to your reply
", "votes": null, "replies": [] }, { "id": 914639, "author_name": "ilhamfp31", "author_url": "", "post_date": "07/04/2020 05:41:05", "content": "Hi! Sorry for the late reply, here is my GitHub source code: https://github.com/ilhamfp/indonesian-text-classification-multilingual
", "votes": null, "replies": [] }, { "id": 915851, "author_name": "sahib12", "author_url": "", "post_date": "07/05/2020 06:31:52", "content": "thanks bro @ilhamfp31
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "825300": "Guys This task is not related to this competition but i thought this discussion panel would help me.\nI have a task to create text classification system for Indonesian Tweets. 2 things i want to know:-\n1. Can I use Bert Base Uncased multilingual for creating embeddings and should i Use it or anything else should i use?\n2. How can I extract features from this data(my professor wants) me to do this additionally.\ni have 2 columns ( comment , sentiment) for 8000 tweets.", "825324": "Try to use at first TF-IDF it works with most languages.", "825334": "OK\nAnd how to tokenise Indonesian @hamditarek", "825902": "Hi Sahib,\nYes, you can use Bert Base Uncased Multilingual as it was trained with Indonesian text data too. If you want a more powerful model, you could also use the latest multilingual model XLM-R.", "825921": "Buddy thank you so much @ilhamfp31\nCan you suggest any tutorial for the same \nThat will be very helpful", "825928": "My current undergraduate final project is on Indonesian text classification using multilingual language model like XLM-R and mBERT! I'll publish it on GitHub tomorrow. Right now I'm still finishing it :D\n\nIn the meantime, you can look at kernel such as [this one](https://www.kaggle.com/xhlulu/jigsaw-tpu-xlm-roberta).", "826022": "Thank you so much @ilhamfp31 ill be looking forward to your reply", "914639": "Hi! Sorry for the late reply, here is my GitHub source code: https://github.com/ilhamfp/indonesian-text-classification-multilingual", "915851": "thanks bro @ilhamfp31" }, "source": "meta" }