As a newcomer, I noticed that recommendation systems and quantitative competitions are much more popular from a team count perspective compared to other NLP and CV competitions, even the one run by CVPR, why is this? Thank you for explaining.
", "messages": [ { "id": 1779014, "postDate": "2022-05-06T02:13:41.633Z", "content": "As a newcomer, I noticed that recommendation systems and quantitative competitions are much more popular from a team count perspective compared to other NLP and CV competitions, even the one run by CVPR, why is this? Thank you for explaining.
", "rawMarkdown": "As a newcomer, I noticed that recommendation systems and quantitative competitions are much more popular from a team count perspective compared to other NLP and CV competitions, even the one run by CVPR, why is this? Thank you for explaining.", "votes": 11 }, { "id": 1779549, "postDate": "2022-05-06T15:14:53.263Z", "content": "I agree with the previous comments, and I think an interactive recommendation system will be one of the more interesting research directions in the future.
\nThe current recommendation results are mostly dominated by the platform and algorithm, the user is more passive and can only silently receive the recommendation results delivered by the recommendation engine. The recommended ones are too single or uninteresting, and users have no right to adjust and feedback. The design of an interactive recommendation system can not only update the recommendation algorithm through data iteration, but also help the recommendation algorithm to sense people's emotions through HCI, so that the recommendation system can not only infer users' preferences for items, but also sense the users' tuning direction for the recommendation algorithm itself.
\nIf the user does not take the initiative, the recommendation system is a blind guess.
", "rawMarkdown": "I agree with the previous comments, and I think an interactive recommendation system will be one of the more interesting research directions in the future.\n\nThe current recommendation results are mostly dominated by the platform and algorithm, the user is more passive and can only silently receive the recommendation results delivered by the recommendation engine. The recommended ones are too single or uninteresting, and users have no right to adjust and feedback. The design of an interactive recommendation system can not only update the recommendation algorithm through data iteration, but also help the recommendation algorithm to sense people's emotions through HCI, so that the recommendation system can not only infer users' preferences for items, but also sense the users' tuning direction for the recommendation algorithm itself.\n\nIf the user does not take the initiative, the recommendation system is a blind guess.", "votes": 6, "replies": [ { "id": 1783175, "postDate": "2022-05-10T07:13:18.183Z", "content": "Isn’t it a prediction problem than a recommendation system? I can understand it being the starting point of an interactive recommendation system but right now, sales have happened and effort is to predict them based on history and trends.
", "rawMarkdown": "Isn’t it a prediction problem than a recommendation system? I can understand it being the starting point of an interactive recommendation system but right now, sales have happened and effort is to predict them based on history and trends." } ] }, { "id": 1779568, "postDate": "2022-05-06T15:28:34.020Z", "content": "I think it's because you can work with just tabular data.
\nCouldn't agree more with \"less resource intensive for a decent result\"
", "rawMarkdown": "Couldn't agree more with \"less resource intensive for a decent result\"", "votes": 5 } ] }, { "id": 1782249, "postDate": "2022-05-09T11:36:39.547Z", "content": "I also agree with the previous comments, you know, there are so many ways to realize a recommendation systems, and so both as a competition participant and as a competition launcher it should be a good project.
\nBy the way, I believe the value of a technology depends on its business value and its application. Perhaps it's just because recommendation systems is really profitable hah.
", "rawMarkdown": "I also agree with the previous comments, you know, there are so many ways to realize a recommendation systems, and so both as a competition participant and as a competition launcher it should be a good project.\n\nBy the way, I believe the value of a technology depends on its business value and its application. Perhaps it's just because recommendation systems is really profitable hah.", "votes": 1 }, { "id": 1780712, "postDate": "2022-05-07T18:12:02.060Z", "content": "Looking forward to participate in that CVPR CV competition
", "rawMarkdown": "Looking forward to participate in that CVPR CV competition", "votes": 2 }, { "id": 1779135, "postDate": "2022-05-06T06:20:27.960Z", "content": "I think it's because there are different ways to approach a recommendation problem, so it's more open to creative solutions. Plus, with the right data, a recommendation system can be very powerful and useful in the real world.
", "rawMarkdown": "I think it's because there are different ways to approach a recommendation problem, so it's more open to creative solutions. Plus, with the right data, a recommendation system can be very powerful and useful in the real world.", "votes": 1, "replies": [ { "id": 1779343, "postDate": "2022-05-06T10:13:57.857Z", "rawMarkdown": "", "votes": 1, "isDeleted": true }, { "id": 1779357, "postDate": "2022-05-06T10:30:25.330Z", "content": "Also true to me, the possibility of different strategies to generate candidates is the most interesting part of this competition from my point of view. You just can't do it enough.
", "rawMarkdown": "Also true to me, the possibility of different strategies to generate candidates is the most interesting part of this competition from my point of view. You just can't do it enough.", "votes": 3 } ] }, { "id": 1782857, "postDate": "2022-05-09T23:49:56.123Z", "content": "Because it's very common business problem
", "rawMarkdown": "Because it's very common business problem" }, { "id": 1782958, "postDate": "2022-05-10T03:01:27.697Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1779549, "author_name": "Kz", "author_url": "", "post_date": "2022-05-06T15:14:53.263000", "content": "I agree with the previous comments, and I think an interactive recommendation system will be one of the more interesting research directions in the future.
\nThe current recommendation results are mostly dominated by the platform and algorithm, the user is more passive and can only silently receive the recommendation results delivered by the recommendation engine. The recommended ones are too single or uninteresting, and users have no right to adjust and feedback. The design of an interactive recommendation system can not only update the recommendation algorithm through data iteration, but also help the recommendation algorithm to sense people's emotions through HCI, so that the recommendation system can not only infer users' preferences for items, but also sense the users' tuning direction for the recommendation algorithm itself.
\nIf the user does not take the initiative, the recommendation system is a blind guess.
", "votes": 6, "replies": [ { "id": 1783175, "author_name": "AtulVerma", "author_url": "", "post_date": "2022-05-10T07:13:18.183000", "content": "Isn’t it a prediction problem than a recommendation system? I can understand it being the starting point of an interactive recommendation system but right now, sales have happened and effort is to predict them based on history and trends.
", "votes": 0, "replies": [] } ] }, { "id": 1779568, "author_name": "Clear n' Simple", "author_url": "", "post_date": "2022-05-06T15:28:34.020000", "content": "I think it's because you can work with just tabular data.
\nCouldn't agree more with \"less resource intensive for a decent result\"
", "votes": 5, "replies": [] } ] }, { "id": 1782249, "author_name": "Qianye", "author_url": "", "post_date": "2022-05-09T11:36:39.547000", "content": "I also agree with the previous comments, you know, there are so many ways to realize a recommendation systems, and so both as a competition participant and as a competition launcher it should be a good project.
\nBy the way, I believe the value of a technology depends on its business value and its application. Perhaps it's just because recommendation systems is really profitable hah.
", "votes": 1, "replies": [] }, { "id": 1780712, "author_name": "Dora", "author_url": "", "post_date": "2022-05-07T18:12:02.060000", "content": "Looking forward to participate in that CVPR CV competition
", "votes": 2, "replies": [] }, { "id": 1779135, "author_name": "", "author_url": "", "post_date": "2022-05-06T06:20:27.960000", "content": "I think it's because there are different ways to approach a recommendation problem, so it's more open to creative solutions. Plus, with the right data, a recommendation system can be very powerful and useful in the real world.
", "votes": 1, "replies": [ { "id": 1779343, "author_name": "", "author_url": "", "post_date": "2022-05-06T10:13:57.857000", "content": "", "votes": 1, "replies": [] }, { "id": 1779357, "author_name": "HAO", "author_url": "", "post_date": "2022-05-06T10:30:25.330000", "content": "Also true to me, the possibility of different strategies to generate candidates is the most interesting part of this competition from my point of view. You just can't do it enough.
", "votes": 3, "replies": [] } ] }, { "id": 1782857, "author_name": "Aleksey Romanov", "author_url": "", "post_date": "2022-05-09T23:49:56.123000", "content": "Because it's very common business problem
", "votes": 0, "replies": [] }, { "id": 1782958, "author_name": "", "author_url": "", "post_date": "2022-05-10T03:01:27.697000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1779014": "As a newcomer, I noticed that recommendation systems and quantitative competitions are much more popular from a team count perspective compared to other NLP and CV competitions, even the one run by CVPR, why is this? Thank you for explaining.", "1779549": "I agree with the previous comments, and I think an interactive recommendation system will be one of the more interesting research directions in the future.\n\nThe current recommendation results are mostly dominated by the platform and algorithm, the user is more passive and can only silently receive the recommendation results delivered by the recommendation engine. The recommended ones are too single or uninteresting, and users have no right to adjust and feedback. The design of an interactive recommendation system can not only update the recommendation algorithm through data iteration, but also help the recommendation algorithm to sense people's emotions through HCI, so that the recommendation system can not only infer users' preferences for items, but also sense the users' tuning direction for the recommendation algorithm itself.\n\nIf the user does not take the initiative, the recommendation system is a blind guess.", "1779568": "I think it's because you can work with just tabular data.\n\n- less resource intensive for a decent result\n- less daunting for people who aren't familiar with deep neural nets\n- more manual feature engineering, hence more room for creativity vs. technical expertise", "1782249": "I also agree with the previous comments, you know, there are so many ways to realize a recommendation systems, and so both as a competition participant and as a competition launcher it should be a good project.\n\nBy the way, I believe the value of a technology depends on its business value and its application. Perhaps it's just because recommendation systems is really profitable hah.", "1780712": "Looking forward to participate in that CVPR CV competition", "1779135": "I think it's because there are different ways to approach a recommendation problem, so it's more open to creative solutions. Plus, with the right data, a recommendation system can be very powerful and useful in the real world.", "1782857": "Because it's very common business problem", "1782958": "" } }