Decipher Data Extraction Accuracy Level

[Shweta_Naik.]

Hi Team,

While Extracting table values from PDFs through Decipher, I have not yet achieved 100% accuracy. I believe that we can come closer to achieving our desired 100% accuracy level.

If you have any specific concerns or suggestions, please let me know as soon as possible.

Thanks,

Shweta

------------------------------
Shweta Naik
------------------------------

[chrispotvin]

Hi Sweta, 

How many documents is your training data set? As you continue to process documents through Decipher, your confidence rating should increase. The other thing to try is to increase your training data set. 

Thanks!

------------------------------
Christopher Potvin
Senior RPA Developer
SS&C Blue Prism
Canada
------------------------------

[LukasRamasauskas]

Hi, Shweta,

Could you please elaborate more what is that 100% as per your use-case. How complex the documents are (purely digital, or printed letters and then scanned physically where any graphical noise artifact may appear, etc.). Is it possible to reach 100% at all as per structure and fields? Do you set 100% as a confidence score for specific fields or whole document calculating it from separate fields?

------------------------------
Lukas Ramasauskas
RPA Software Engineer / Robotic Solutions Architect
Swedbank AB
Europe/Vilnius
------------------------------

[Shweta_Naik.]

Thanks for your response Christopher,

I'm training one document and Checking the Extracted values for the same formatted documents. Through Decipher I have trained one document (5 to 10 times). 

------------------------------
Shweta Naik
------------------------------

[Shweta_Naik.]

Thanks for your Response Lukas,

For the Overall Document I'm not getting 100% Accuracy, but for a few fields, I'm getting 100% Accuracy.

------------------------------
Shweta Naik
------------------------------