awsdocs / amazon-comprehend-developer-guide Goto Github PK

The Python script file has errors from line 18-21.
File "bin/comprehend-ssie-annotation-tool-cli.py", line 18, in
from utils.s3_helper import S3Client
ModuleNotFoundError: No module named 'utils.s3_helper'

Could there be missing steps in the Uploading a PDF to an S3 bucket section?
Or is current script out dated?

Tested and script work, perhaps update instructions to first run - pipenv shell before executing script, missing instructions on AWS page

Batch processing examples for aws cli still use aws insight which doesn't seem to work anymore.

Hi
Going through the documents , it seems like we have a lot of steps can we please add a tutorial where we take a use case and walk the customer through each step.
These steps specially were not very easy to implement: https://github.com/awsdocs/amazon-comprehend-developer-guide/blob/master/doc_source/analysis-jobs.md

Cheers
Mahesh Yadav

In the Documentation "Training a Custom Classifier" (in how-document-classification-training.md) there is a mention to escaped commas in the training dataset CSV file but it is for labels only and no example is given.

I tried several typical escaping characters one can usually use with CSV files such as \ or " but none of them has worked.

I ended with replacing comma with the html version of it: ",".

What is the official escaping method for comma in the document or label? Could you please provide an example.

You should expand the LDA.

You should expand the abbreviation "LDA" to "Latent Dirichlet Allocation."

There are changes to the documentation that are not represented here. Specifically the SageMaker Ground Truth information here.

https://docs.aws.amazon.com/comprehend/latest/dg/cer-annotation.html

I couldn't find where this file is since it's an announcement.

But you guys will probably know where it is:

https://aws.amazon.com/about-aws/whats-new/2018/07/amazon-comprehend-now-supports-syntax-analysis/

“The customers moved quickly when doors opened” allows a developer to understand the nouns (customers, doors) and the verb (quickly).

Quickly is not a verb. I'll leave the irony of this issue to you guys. :)

Regarding the usage of 10-20% of the documents for testing purposes, shouldn't it say
"we ensure that there are no labels in the test set which we have seen before"
instead of
"we ensure that there are no labels in the test set which we have never seen before"