cancel
Showing results for 
Search instead for 
Did you mean: 
Reply
davidkr
Advocate I
Advocate I

Questions regarding invoice model training

Hello,

 

I have started trying AI Builder to process invoices. I have tried training a custom model and have some questions. I hope you could help me in this matter.

 

 

1. When I create a custom model by using "invoice" document type, does that mean that the custom model that I create will be a "more knowledgeable" version of the prebuilt model? I mean, does it build more knowledge of new invoice patterns (that are provided for the training) on top of the knowledge that the prebuilt model already has? The reason why I ask this is because sometimes, the line items that are extracted with the prebuilt model are more accurate than the ones that are extracted with the custom model even though the custom model was trained with that specific invoice layout. 

 

Example:

 

- How I tagged tha line items during training

Romi training.png

- The line items extracted with the prebuilt model

davidkr_0-1694509492448.png

 

- The line items extracted with the custom model that was trained with this specific invoice layout using 5 samples

davidkr_1-1694509569428.png

 

 

2. Is currency not a default field in "invoice" document type? I can choose it while working on a flow but it doesn't exist in the "invoice" document type. I noticed that about half of the time, the prebuilt model couldn't detect the currency on the invoice as well. Is this being worked on?

 

 

3. If a model can be trained with multiple invoice layouts using collections, what would be the reasons or scenerios to have multiple models? why not always put every invoice layouts in one model?

 

 

4. Would a custom model that was trained with one specific invoice layout deliver more accurate results when using to extract data from invoices with that layout compared to another custom model that was trained with several invoice layouts including the layout?

 

 

5. Sometimes a text on an invoice can be two things (exp. "customer name" and "costomer address recipient", "customer address" and "billing address") but it is not possible to tag the same text with more than one fields. Would this maybe change in the future?

 

 

6. Line items table of some invoices has more columns than the default line items table of the "invoice" document type but it is not possible to add more columns to the default line items table. It is not possible to tag the line items table more than once as well. In this case, if I want to extract the line items table with more columns than the available default columns, I have to define a new line items table with all the columns that I need, disable the default line items table during the tagging and just use the custom-defined line items table during the tagging?

 

 

7. Is there a best practice in tagging when an invoice has multiple pages with the same header and footer on every page and I want to extract data from the header/footer? I tried to always just tag everything on the first page but after the training, when I went back to edit the model, some tagged positions changed and were not on the first page anymore. The text on the header of the second page could be tagged. Is this normal, because I thought that tagged positions would always stay at the same positions?


Thank you in advance for your answers!

1 ACCEPTED SOLUTION

Accepted Solutions
Antrod
Power Apps
Power Apps

Hi @davidkr ,

 

Thanks for reaching out, I'll answer as best as possible to all these interesting questions, hope that'll help you in your AI Builder journey!

 

1- The custom model built starting with "Invoices" shouldn't be more or less knowledgeable that the Structured/Unstructured custom one. See that as a way to bootstrap tagging in a faster way. We recently did an update to the custom model, do you still see discrepancies with the Invoice prebuilt model?

 

2- Indeed currency symbol detection is not included. There are plans to include it in the future but no committed date on that.

 

3- Sometimes multiple models can be better if you have to do a conditional routing in a flow for instance. Otherwise yes building a unique model could perfectly do the job.

 

4- It shouldn't, especially if you add the deferent in separate collections.

 

5- Yes, each field should point at a different tag and there's no possibility to change that right now. However, if you have 2 fields that sometimes  point at different text and sometimes at the same field, it would be possible to add a post-processing logic saying that if a field is empty, set the value of it with the field that is not empty.

 

6- Yes, that would be the right process.

 

7- When you have multi-page tables, you should tag all the pages otherwise you will have text not properly recognized. About the footer you could try to include it as a line in your table or perhaps  tag it as a separate field.

 

View solution in original post

1 REPLY 1
Antrod
Power Apps
Power Apps

Hi @davidkr ,

 

Thanks for reaching out, I'll answer as best as possible to all these interesting questions, hope that'll help you in your AI Builder journey!

 

1- The custom model built starting with "Invoices" shouldn't be more or less knowledgeable that the Structured/Unstructured custom one. See that as a way to bootstrap tagging in a faster way. We recently did an update to the custom model, do you still see discrepancies with the Invoice prebuilt model?

 

2- Indeed currency symbol detection is not included. There are plans to include it in the future but no committed date on that.

 

3- Sometimes multiple models can be better if you have to do a conditional routing in a flow for instance. Otherwise yes building a unique model could perfectly do the job.

 

4- It shouldn't, especially if you add the deferent in separate collections.

 

5- Yes, each field should point at a different tag and there's no possibility to change that right now. However, if you have 2 fields that sometimes  point at different text and sometimes at the same field, it would be possible to add a post-processing logic saying that if a field is empty, set the value of it with the field that is not empty.

 

6- Yes, that would be the right process.

 

7- When you have multi-page tables, you should tag all the pages otherwise you will have text not properly recognized. About the footer you could try to include it as a line in your table or perhaps  tag it as a separate field.

 

Helpful resources

Announcements

Community will be READ ONLY July 16th, 5p PDT -July 22nd

Dear Community Members,   We'd like to let you know of an upcoming change to the community platform: starting July 16th, the platform will transition to a READ ONLY mode until July 22nd.   During this period, members will not be able to Kudo, Comment, or Reply to any posts.   On July 22nd, please be on the lookout for a message sent to the email address registered on your community profile. This email is crucial as it will contain your unique code and link to register for the new platform encompassing all of the communities.   What to Expect in the New Community: A more unified experience where all products, including Power Apps, Power Automate, Copilot Studio, and Power Pages, will be accessible from one community.Community Blogs that you can syndicate and link to for automatic updates. We appreciate your understanding and cooperation during this transition. Stay tuned for the exciting new features and a seamless community experience ahead!

Summer of Solutions | Week 4 Results | Winners will be posted on July 24th

We are excited to announce the Summer of Solutions Challenge!    This challenge is kicking off on Monday, June 17th and will run for (4) weeks.  The challenge is open to all Power Platform (Power Apps, Power Automate, Copilot Studio & Power Pages) community members. We invite you to participate in a quest to provide solutions to as many questions as you can. Answers can be provided in all the communities.    Entry Period: This Challenge will consist of four weekly Entry Periods as follows (each an “Entry Period”)   - 12:00 a.m. PT on June 17, 2024 – 11:59 p.m. PT on June 23, 2024 - 12:00 a.m. PT on June 24, 2024 – 11:59 p.m. PT on June 30, 2024 - 12:00 a.m. PT on July 1, 2024 – 11:59 p.m. PT on July 7, 2024 - 12:00 a.m. PT on July 8, 2024 – 11:59 p.m. PT on July 14, 2024   Entries will be eligible for the Entry Period in which they are received and will not carryover to subsequent weekly entry periods.  You must enter into each weekly Entry Period separately.   How to Enter: We invite you to participate in a quest to provide "Accepted Solutions" to as many questions as you can. Answers can be provided in all the communities. Users must provide a solution which can be an “Accepted Solution” in the Forums in all of the communities and there are no limits to the number of “Accepted Solutions” that a member can provide for entries in this challenge, but each entry must be substantially unique and different.    Winner Selection and Prizes: At the end of each week, we will list the top ten (10) Community users which will consist of: 5 Community Members & 5 Super Users and they will advance to the final drawing. We will post each week in the News & Announcements the top 10 Solution providers.  At the end of the challenge, we will add all of the top 10 weekly names and enter them into a random drawing.  Then we will randomly select ten (10) winners (5 Community Members & 5 Super Users) from among all eligible entrants received across all weekly Entry Periods to receive the prize listed below. If a winner declines, we will draw again at random for the next winner.  A user will only be able to win once overall. If they are drawn multiple times, another user will be drawn at random.  Individuals will be contacted before the announcement with the opportunity to claim or deny the prize.  Once all of the winners have been notified, we will post in the News & Announcements of each community with the list of winners.   Each winner will receive one (1) Pass to the Power Platform Conference in Las Vegas, Sep. 18-20, 2024 ($1800 value). NOTE: Prize is for conference attendance only and any other costs such as airfare, lodging, transportation, and food are the sole responsibility of the winner. Tickets are not transferable to any other party or to next year’s event.   ** PLEASE SEE THE ATTACHED RULES for this CHALLENGE**   Week 1 Results: Congratulations to the Week 1 qualifiers, you are being entered in the random drawing that will take place at the end of the challenge.   Community MembersNumber SolutionsSuper UsersNumber Solutions Deenuji 9 @NathanAlvares24  17 @Anil_g  7 @ManishSolanki  13 @eetuRobo  5 @David_MA  10 @VishnuReddy1997  5 @SpongYe  9JhonatanOB19932 (tie) @Nived_Nambiar  8 @maltie  2 (tie)   @PA-Noob  2 (tie)   @LukeMcG  2 (tie)   @tgut03  2 (tie)       Week 2 Results: Congratulations to the Week 2 qualifiers, you are being entered in the random drawing that will take place at the end of the challenge. Week 2: Community MembersSolutionsSuper UsersSolutionsPower Automate  @Deenuji  12@ManishSolanki 19 @Anil_g  10 @NathanAlvares24  17 @VishnuReddy1997  6 @Expiscornovus  10 @Tjan  5 @Nived_Nambiar  10 @eetuRobo  3 @SudeepGhatakNZ 8     Week 3 Results: Congratulations to the Week 3 qualifiers, you are being entered in the random drawing that will take place at the end of the challenge. Week 3:Community MembersSolutionsSuper UsersSolutionsPower Automate Deenuji32ManishSolanki55VishnuReddy199724NathanAlvares2444Anil_g22SudeepGhatakNZ40eetuRobo18Nived_Nambiar28Tjan8David_MA22   Week 4 Results: Congratulations to the Week 4 qualifiers, you are being entered in the random drawing that will take place at the end of the challenge. Week 4:Community MembersSolutionsSuper UsersSolutionsPower Automate Deenuji11FLMike31Sayan11ManishSolanki16VishnuReddy199710creativeopinion14Akshansh-Sharma3SudeepGhatakNZ7claudiovc2CFernandes5 misc2Nived_Nambiar5 Usernametwice232rzaneti5 eetuRobo2   Anil_g2   SharonS2  

Check Out | 2024 Release Wave 2 Plans for Microsoft Dynamics 365 and Microsoft Power Platform

On July 16, 2024, we published the 2024 release wave 2 plans for Microsoft Dynamics 365 and Microsoft Power Platform. These plans are a compilation of the new capabilities planned to be released between October 2024 to March 2025. This release introduces a wealth of new features designed to enhance customer understanding and improve overall user experience, showcasing our dedication to driving digital transformation for our customers and partners.    The upcoming wave is centered around utilizing advanced AI and Microsoft Copilot technologies to enhance user productivity and streamline operations across diverse business applications. These enhancements include intelligent automation, AI-powered insights, and immersive user experiences that are designed to break down barriers between data, insights, and individuals. Watch a summary of the release highlights.    Discover the latest features that empower organizations to operate more efficiently and adaptively. From AI-driven sales insights and customer service enhancements to predictive analytics in supply chain management and autonomous financial processes, the new capabilities enable businesses to proactively address challenges and capitalize on opportunities.    

Updates to Transitions in the Power Platform Communities

We're embarking on a journey to enhance your experience by transitioning to a new community platform. Our team has been diligently working to create a fresh community site, leveraging the very Dynamics 365 and Power Platform tools our community advocates for.  We started this journey with transitioning Copilot Studio forums and blogs in June. The move marks the beginning of a new chapter, and we're eager for you to be a part of it. The rest of the Power Platform product sites will be moving over this summer.   Stay tuned for more updates as we get closer to the launch. We can't wait to welcome you to our new community space, designed with you in mind. Let's connect, learn, and grow together.   Here's to new beginnings and endless possibilities!   If you have any questions, observations or concerns throughout this process please go to https://aka.ms/PPCommSupport.   To stay up to date on the latest details of this migration and other important Community updates subscribe to our News and Announcements forums: Copilot Studio, Power Apps, Power Automate, Power Pages

Users online (648)