Why isn't ITEM included in the Summary section in Textract Expense Analysis API? and it can't capture ITEM most of the time too.

Question

For Expense Analysis API such as invoice and receipts, I have tested the API in the past few days, and found out it misses ITEM (items purchased/sold) on the receipts or invoice most of the times, and it is NOT included in the Summary section, everything else in the Summary are pretty accurate so far.

The question, how to improve this and capture ITEM accurately? Without this key data point being accurate, this Expense Analysis API actually doesn't have too much value, just like the saying "the capacity of a barrel is determined not by the longest wooden bars, but by the shortest".

Please kindly advise, thanks.

Answer

Item data are returned in LineItemGroups section:
* SummaryFields - Contains all other information a receipt, such as header information or the vendors name
* LineItemGroups- A data set containing LineItems which store information about the lines of text, such as an item purchased and its price on a receipt.

See [AnalyzeExpense API](https://docs.aws.amazon.com/textract/latest/dg/API_AnalyzeExpense.html)

Have you reviewed [Best Practices for Amazon Textract](https://docs.aws.amazon.com/textract/latest/dg/textract-best-practices.html) to improve results produced by Textract?

You also may consider using Amazon Textract with Amazon Augmented AI for processing critical documents as described [here](https://aws.amazon.com/blogs/machine-learning/using-amazon-textract-with-amazon-augmented-ai-for-processing-critical-documents/)

Answer

Thanks for the follow up.

Even in LineItemGroups LineItems, it fails to capture ITEM most of the time... For a consumer facing product, we can not afford to ask users to follow specific format or guild lines, this is the shortest bar in this barrel which makes this product unusable most of the time. Is there any solution to fix this issue? The goal is to capture ITEM - items purchased/sold.

Appreciate your input!

Why isn't ITEM included in the Summary section in Textract Expense Analysis API? and it can't capture ITEM most of the time too.

Relevanter Inhalt