1 個回答
- 最新
- 最多得票
- 最多評論
0
Amazon Personalize supports for languages enables customers to unlock the information trapped in their product descriptions, reviews, movie synopses or other unstructured text to generate highly relevant recommendations for users. So as you mentioned, it is about contextualizing the unstructured text in the data set. You can see this in this reference documentation.
已回答 1 年前
相關內容
- 已提問 6 個月前
AWS 官方已更新 2 年前
Thank you for the clarification. To confirm, does this mean that AWS Personalize is capable of extracting meaning from the provided unstructured text data? Additionally, since we do not currently have plans to translate our data into English, would it be advisable to simply Latinize the data and publish it in that form?
Yes, it uses NLP behind the scene to extract key elements from the metadata. can you elaborate what you mean by Latinizing ?
Apologies for any confusion caused by my previous posts. This is what I mean by Latinizing:
In my language, the word for "sky" is represented by the characters "ცა." By Latinizing, I am referring to the process of replacing each of these characters with their Latin counterparts. In this particular example, the Latinized version of "ცა" would be "Tsa."
If your data includes any non-ASCII encoded characters, your CSV file must be encoded in UTF-8 format : https://docs.aws.amazon.com/personalize/latest/dg/data-prep-formatting.html
So I don't think you need to Latinize your data.
Also, here is another reference to know more about NLP part of Amazon Personalize : https://aws.amazon.com/blogs/machine-learning/unlock-information-in-unstructured-text-to-personalize-product-and-content-recommendations-with-amazon-personalize/