ICC Line Item Recognition White Paper

Invoice Capture Center
Line Item recognition and PO Line Item Mapping

Gisela Hammann, SAP Solutions Group
March 2012
Abstract
This document gives insight in how line item recognition works in Invoice Capture Center (ICC),
and how mapping with Purchase Order/Goods Receipt Data is working in detail. This document
applies for versions > ICC 5.2 SP5, and > ICC 6.0 SP1.
Highlights:
– Line item recognition
– Mapping with downloaded PO/GR data
– Customizing settings
Contents
1. Introduction .................................................................................................... 3
2. OCR invoice item recognition....................................................................... 4
3. DOKuStar Invoice Items ................................................................................ 6
4. PO line item Mapping .................................................................................... 7
Evaluation of customer business scenario ............................................................ 7
5. ICC LineItemMapping Algorithm ................................................................. 8
Data set used for line item mapping ....................................................................... 8
OCR data ......................................................................................................... 8
SAP download data ......................................................................................... 8
Line item mapping algorithm .................................................................................. 9
Determine relevant POItems for Mapping ....................................................... 9
Assignment mapping ..................................................................................... 10
SnapMatch mapping and Assignment mapping ............................................ 10
Mapping result ............................................................................................... 10
6. Tips for successful use of invoice line item mapping ............................. 13
About Open Text ....................................................................................................... 14
w w w. o p e n t e x t . c o m
For more information about Open Text products and services, visit www.opentext.com. Open Text is a publicly traded company on both NASDAQ (OTEX) and the TSX (OTC).
Copyright © 2009 by Open Text Corporation. Open Text and The Content Experts are trademarks or registered trademarks of Open Text Corporation. This list is not exhaustive. All other
trademarks or registered trademarks are the property of their respective owners. All rights reserved. SKU#_EN
1. Introduction
Line item recognition is a feature which is considered as crucial for Invoice
Capture Center (ICC), as it has an impact on automation rate of the end to end
system.
In baseline, Vendor Invoice Management (VIM) expects OCR line item data
delivered with PO number and PO line item number.
ICC line item mapping has been conceived to provide automation support for
standard invoices with few line items and with good relation to PO. Other options
should be considered, if relationship between PO data and invoice data is not
obvious, due to specific business scenarios running in the customer environment.
Alternative VIM customizing options exist, which will not at all require invoice line
item data delivered by recognition, or which will do PO line item mapping using
VIM methods. For more details please refer to VIM 6.0 Configuration Guide
chapter 5.1.7.
2. OCR invoice item recognition
In general we discern invoice header and invoice item data.
Invoice header data in general are fields like invoice date, invoice number, which
occur once, mainly on the first page of the invoice, and which are determined by
searching for structured data related to keywords and phrases.
Invoice item data recognition is more complex, as multiple line items may exist on
an invoice, and line items very often have not a regular layout, but are printed in
different variations.
The following parameters can be set for defining how line item recognition should
work in a specific customer environment:
 Extract line items: Invoice item recognition can be switched on and off.
 Line items for PO invoices / line items for non-PO invoices: ICC will
recognize line items, but only transmit line item data to VIM for the
option which is checked. E.g. if line items for nonPO data would be
unchecked, ICC would not transmit data for invoices where no PO
number has been found.
 Check line items against downloaded data: This parameter activates

line item mapping with Purchase Order/Goods Receipt detail data
downloaded from SAP system. For mapping of amounts, an absolute or
relative tolerance can be defined.
 If no PO found determine PO number from DB (available since ICC

5.2 SP7 and since ICC 6.0 SP2): This parameter is not activated by
default in a newly created application. Because of compatibility reasons it
is activated when importing older applications, but it is highly
recommended to deactivate this parameter. If the parameter is not
activated, it stops LineItemMapping algorithm, if no PO number has been
found on invoice. If the parameter is activated, ICC LineItemMapping
tries to map with all PO items available for the current vendor. This may
take a long time, if many PO items exist in PO download data for the
current vendor, and it may cause erroneous mapping results, especially
same vendor sends PO and NonPO invoices.
3. DOKuStar Invoice Items Recognition
The underlying DOKuStar KnowledgeBase method for table recognition is generic
and works out of the box. For specific customer business scenarios it may be
appropriate to extend line item recognition by customizing, if e.g. line item mapping
should rely on additional line item information like material numbers etc for providing
better mapping results. For more detailed information about customizing for line items
please refer to ICC Customizing Guide.
Invoice line item recognition has been elaborated by analyzing a big volume of
various invoices from different vendors and different countries. Invoice line item
recognition works independently from PO download data, PO download data are not
considered for optimizing recognition result. Invoice line item recognition considers
local writing styles for all supported countries. The DOKuStar method for line item
recognition takes text lines found by pure OCR as a starting point
These text lines are analyzed for occurrences of syntactical structures like table
header keywords, amounts, quantities, units of measure, phrases typical for
summarizing lines at the end of the table a.s.o. Results of this first step are used to
find table header and end of table as well as horizontal position of lines and vertical
position of columns. Logical checks like calculation of unit price x quantity = item
price are also used in this second step. All this results in a list of line items, each line
item consisting of one or more text lines.
For invoices, some line items currently are filtered out and not considered as valid
line items:
 Line items with negative value (except for credit memos). For credit memos,
negative line items are not filtered out.
 Line items with keywords like “expenses, discount”, showing incidental costs,
discounts or additional costs
For ICC 7.0 it is planned to develop a business scenario framework providing
customizing options for handling line items currently filtered out.
4. PO line item Mapping
Line item recognition using the table recognition algorithm, described in the
precedent chapter, will have very good recognition results. Nevertheless, by
default, VIM requires for PO invoices, for every line item PO number and PO line
item number. The fact is, that PO line item number is not printed on the invoice,
but can only be determined from PO data. Also for some invoice items, especially
multiline invoice items on invoices referring to multiple POs, it is difficult to assign
PO numbers to line items by means of optical character recognition. This is the
reason why data mapping with downloaded PO and GR data is used in order to
complete data to be delivered to VIM.
Evaluation of customer business scenario

Before deciding to activate PO line item mapping, it is crucial to test, if this
generic feature will fit to specific customer environment, if customizing is needed
for adapting to customer business scenario, or if PO line item mapping cannot be
used in ICC, but must be used in VIM. ICC and VIM interaction is very flexible, so
in some cases it may be better to do line item mapping in VIM.
Currently, before activating PO line item mapping in ICC, the following business
scenarios should be checked:
 Discount handling on invoice line item level vs. availability of

discount data in PO download
 Handling of additional costs on invoice line item level vs. availability
of additional cost data in PO download
 If multiple invoices refer to one PO, check availability of GR data in
download data. Without GR data, line item mapping will not work.
5. ICC LineItemMapping Algorithm
ICC LineItem mapping has been built using innovative generic SnapMatch
algorithm. This is an OCR full text based pattern matching algorithm, working
language independent. ICC LineItem mapping works based on PO details data
downloaded from SAP and stored in ICC internal data base. ICC LineItem
Mapping tries to map data contained in data base with DOKuStar invoice item
data.
Data set used for line item mapping
OCR data
ICC invoice line item mapping uses data extracted by DOKuStar invoice items
table recognition. Line item mapping considers structured table data for mapping
and evaluates against PO details download data. Depending on parameter
settings, in some cases, line item mapping considers additional OCR data, which
have not been acknowledged by table recognition, as valid line candidates, with
lower priority.
SAP download data
For line item mapping, data listed below are downloaded from SAP. For
description of the ABAP download report in SAP please refer to VIM
Configuration Guide.
PO Header data
PO item data (PO details)
There are 2 different item types:
 „1“ PO item
 „2“ GR item (posted)
The column “reserve1” is intended to be used for implementing custom specific

extensions, by adding an additional line item field to be considered for mapping,
e.g. material number. If line item table recognition will be extended to recognize
e.g. material number in item table (for description how to do this, please refer to
ICC Customizing Guide: Adding a Custom Column to the Invoice Items Custom
Field), line item mapping will work for this new column, if download report fills
material numbers in column “reserve1”.
Line item mapping will not work, if one of the table columns will be missing in
download data, but it will work, if additional columns exist. Only reserve1 column
will be considered for mapping, additional custom fields will not be considered.
Line item mapping algorithm

Line item mapping algorithm is working in subsequent steps:
Determine relevant POItems for Mapping
Based on determined vendor, Company Code and PO numbers (if available in

download: Delivery note numbers) from the invoice, download data are filtered for
relevant items. Line item mapping is dependent on good PO/DN number
recognition.
For getting a good data quality for mapping, the algorithm refines PO download
data as follows:
 If in PO item QUANTITY * UNITPRICE != AMOUNT, then a clone of the
POItem with the correct amount value QUANTITY * UNITPRICE is
inserted, and also considered for mapping with OCR data.
 If PO items and GR items with identical QUANTITY, UNITPRICE and
AMOUNT, ORDERID and POSITION are found, PO items are deleted for
not generating duplicated mapping.
With this set of relevant PO and GR items, mapping with OCR data is set up.
Assignment mapping
Assignment mapping is applied, if the sum of the extracted invoice item amounts
equals the net amount of the invoice. Then only the corresponding PO Numbers
and PO Item Numbers need to be assigned,
This assignment is based on the criteria listed below (and will be applied in the
mentioned order)
1. Quantity, UnitPrice, ItemAmount on POItems of type 1 & 2
2. ItemAmount of type 1 & 2
Based on these criteria, groups of POItems and InvoiceItems are created.
If another criteria-identical group is found, the mapping algorithm assigns

POItems of the group with best fitting invoice items in regard to Description,
ItemUnit and Reserve1. For mapping, the absolute or relative tolerance, if defined
in customizing, is considered for the amount fields ItemUnitPrice and
ItemAmount.
After an item group was mapped, its POItems and InvoiceItems will be removed
from the list of mapping candidates.
SnapMatch mapping and Assignment mapping
SnapMatch mapping and Assignment mapping is applied in combination, if the

sum of the extracted invoice item amounts does not equal the net amount of the
invoice.
Mapping result
The result of the mapping is, that the columns ORDERID, POSITION, DELIVERY,
are written to the invoice item result data.
Example
The following example shows how mapping algorithm is working:
Invoice data:
Download data:
ORDERID POSITION ITEMTYPE AMOUNT DESCRIPTION QUANTITY UNITPRICE UNIT

4500017277 00001 1 204.60 Bremse 2.000 102.30 ST
4500017277 00002 1 40.00 Schrauben 1.000 40.00 ST
4500017277 00003 1 46.00 Lager 5.000 9.20 ST
4500017277 00004 1 204.60 22 Antrieb xyz234 2.000 102.30 ST
4500017277 00005 1 40.00 Rohling Welle 1.000 40.00 ST
4500017277 00006 1 24.00 Muttern 20.000 1.20 ST
4500017277 00007 1 204.60 Batterie 2.000 102.30 ST
Results:
 Invoice item 30 maps for QUANTITY, UNITPRICE, AMOUNT completely

with PO item 00003 and is assigned without considering DESCRIPTION.
 Download items 00001, 00004, 00007 and Download items 00002 and
00005 are grouped as they show identical triples QUANTITY,
UNITPRICE, AMOUNT.
 The triple 2.000, 102.30, 204.60 is found three times on the invoice
(invoice items 10, 40, 60). Assignment is done using item description
text: 10 ->00004, 40->00007, 60->00001. For assignment it is sufficient
to find the same words in PO descriptions and invoice descriptions,
independent of the word sequence.
In the same way positions 20 and 50 are assigned to PO items 00005 and 00002.
Specific exceptions:
No Company Code determined
If no CC could be determined, PO line item mapping will not work for the invoice.
Depending on settings for line item recognition, line items will or will not be
exported to VIM. If exported to VIM, fields PO item number and PO number will
be empty.
No vendor determined
If no vendor could be determined, PO line item mapping will not work for the
current invoice. Depending on settings for line item recognition, line items will or
will not be exported to VIM. If exported to VIM, fields PO item number and PO
number will be empty.
No PO number found
If no PO numbers have been determined for the invoice, all PO/DN items for all
PO/DN numbers of the relevant vendor may be considered for mapping. This is
the default setting, which is recommended to deactivate, especially if in customer
environment same vendor sends PO and nonPO invoices.
6. Tips for successful use of invoice line
item mapping
Understanding how line item mapping is working, for every customer environment
it should be checked, if ICC line item mapping can be used successfully. This is
important, as line item mapping may abuse validation resources in case it doesn’t
work properly.
 Is PO number recognition working with high recognition rate? Good

PO/DN recognition will enable filtering of the right items from data base
for mapping. Same for vendor recognition.
 Does a good relation between PO/GR data and invoice data exist to
enable line item mapping? If PO/GR data do not map at all with invoice
data, line item mapping will not work. This may be the case for invoices
with discounts or additional costs, as, in standard, discount and expense
items are currently not considered for mapping.
 For framework contracts, data mapping will only succeed, if goods

receipt data exist.
 It should also be considered how many data make sense to be

downloaded using the PO download report: Define carefully selection
parameters for download report: for most business scenarios selection
parameters should be applied as follows:
 Open POs: download only PO data from PO which are

not completed
 Delta report: The first report should run for complete PO

data of the selected period, later on delta download
should be applied for several weeks. A full download
should be applied regularly e.g. once a month for
consolidating data in VIM staging table and ICC data
base.
 Time period: the time period should be selected as short

as possible – ideally several months, not several years.
This reduces running time for collecting PO data in VIM
staging table, and also reduces time to import PO data
from staging table into ICC data base.
About Open Text
Open Text is a leader in Enterprise Content Management (ECM). With two
decades of experience helping organizations overcome the challenges
associated with managing and gaining the true value of their business content,
Open Text stands unmatched in the market.
Together with our customers and partners, we are truly The Content Experts,™
supporting 46,000 organizations and millions of users in 114 countries around the
globe. We know how organizations work. We have a keen understanding of how
content flows throughout an enterprise, and of the business challenges that
organizations face today.
It is this knowledge that gives us our unique ability to develop the richest array of
tailored content management applications and solutions in the industry. Our
unique and collaborative approach helps us provide guidance so that our
customers can effectively address business challenges and leverage content to
drive growth, mitigate risk, increase brand equity, automate processes, manage
compliance, and generate competitive advantage. Organizations can trust the
management of their vital business content to Open Text, The Content Experts.
www.opentext.com/[solution group URL]
Sales: [email address]

[phone number]
Support: [email address]

[phone number]

ICC Line Item Recognition White Paper

Uploaded by

Copyright:

Available Formats

ICC Line Item Recognition White Paper

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ICC Line Item Recognition White Paper

Uploaded by

Copyright:

Available Formats

Invoice Capture Center

Line Item recognition and PO Line Item Mapping

– Line item recognition

– Mapping with downloaded PO/GR data

 Check line items against downloaded data: This parameter activates

 If no PO found determine PO number from DB (available since ICC

Evaluation of customer business scenario

 Discount handling on invoice line item level vs. availability of

Data set used for line item mapping

SAP download data

There are 2 different item types:

 „2“ GR item (posted)

The column “reserve1” is intended to be used for implementing custom specific

Line item mapping algorithm

Determine relevant POItems for Mapping

Based on determined vendor, Company Code and PO numbers (if available in

1. Quantity, UnitPrice, ItemAmount on POItems of type 1 & 2

2. ItemAmount of type 1 & 2

Based on these criteria, groups of POItems and InvoiceItems are created.

If another criteria-identical group is found, the mapping algorithm assigns

SnapMatch mapping and Assignment mapping

SnapMatch mapping and Assignment mapping is applied in combination, if the

The following example shows how mapping algorithm is working:

ORDERID POSITION ITEMTYPE AMOUNT DESCRIPTION QUANTITY UNITPRICE UNIT

 Invoice item 30 maps for QUANTITY, UNITPRICE, AMOUNT completely

No Company Code determined

 Is PO number recognition working with high recognition rate? Good

 For framework contracts, data mapping will only succeed, if goods

 It should also be considered how many data make sense to be

 Open POs: download only PO data from PO which are

 Delta report: The first report should run for complete PO

 Time period: the time period should be selected as short

www.opentext.com/[solution group URL]

Sales: [email address]

Support: [email address]

You might also like