16.45-17.45 KTM Advanced Technical Track

Bo Jin

Sr. Solution Architect

Fatih Karaoglu


Kofax Transformation Modules –

Advanced Track & What’s new in

KTM

Technical Track

Agenda

Clustering Utility

Benchmarking

Separation

Classification

Extraction

Project Merge Tool

Localisation

Thin Client Enhancements

Q&A

2

Technology Enhancements

Productivity

Enhancements

Design Time

Productivity

Enhancements

Users

Clustering Utility

Technology Enhancements

New Utility for Clustering Unknown Documents

What it does

Requirements

Step-by-step

Importing into KTM

4

What does the Kofax Clustering Utility do?

When configuring KTM content classification, the customer needs

to provide samples for each class.

What KTM requires:

5


When configuring KTM content classification, the customer needs

to provide samples for each class.

What customers usually provide:

6


presorting a document set into clusters of similar documents

User labels some of these clusters

Utility learns from labeling and pre-sorts again

Several iterations of labeling and pre-sorting

Export of sorted documents as learn-set for KTM content

classification

7


new KTM project

Customer uses Utility to provide KPSG or partner with sorted

documents

KPSG or partner uses Utility to sort documents from customer

Understanding what are the biggest subsets of documents in a

customer’s monthly mailroom volume

enhancing a KTM project

Customer adds new classes to project and needs samples for

classification

8

Requirements

Kofax Clustering Utility works with XDocuments

XDocuments must be created with KTM OCR Server tool

KTM (5.5 SP2) must be installed to use Clustering Utility.

9

Using the KTM OCR Server reduces the KTM base volume count

Eval licenses supported

Hardware requirements same as for KC/KTM

Files to be clustered should be local for performance

Need write access to file location

Requirements

10

Step by Step – KTM OCR Server

Configuring the KTM OCR Server:

Select path to unsorted images

Enable „Save XDoc files“ and

„Save text files“

Under OCR Settings, select

proper language

Leave rest at default

Running the KTM OCR Server:

Simply press the Start button

11

Step by Step – Kofax Clustering Utility

1. Import

Point „Import directory“ to same directory of unsorted documents

For each document, an .xdc file and a .txt file must exist

Select „Start Discovery“.

Takes a while, ~0.5 sec per document

Converts XDocs into internal format

Identifies initial clusters

12


2. Discovery

Label initial 3 clusters

You see the most representative document of each cluster

Provide a name for each cluster, will be used as class name in

KTM

13


2. Discovery

14


2. Discovery

You can stop discovery when 80-90% of the documents are

discovered or continue until all documents are discovered

At 80-90% the most common document types are often known,

remaining documents are likely in very small clusters

Click „Review“ to continue to next step

15


3. Review

Sort by categories (labels)

Examine the categories for consistency

Confirm some documents if you want to cluster again

16


3. Review

17


4. Export

Select any directory for export

Sub directories will be created for each category/label

.txt files (and tif/xdoc for reference) will be exported, since only .txt

files are used to train KTM content classification later

18

Importing into KTM

In Project Builder, point New Project dialog Content Classifier

settings to exported directory

Select „Discovered documents“ sub directory

19

Importing into KTM

A class is created in Project Builder for each category

Training documents are imported

Select „Train“ in Project Builder main menu

Verify in Classification Benchmark (Result Matrix)

20

Importing into KTM

21

Setting this up manually

and finding/organizing the

proper training documents

takes hours or days.

With the Kofax Clustering

Utility, this example took

20 minutes.

Benchmarking

Productivity Enhancements – Design Time

KTM 5.5 – Benchmarking

Separation Benchmarking

Classification Benchmarking

Extraction Benchmarking

23

KTM 5.5 – Separation Benchmarking

Separation Benchmark

24

Document Separation Test



25


Golden Batch

Golden Files – Extraction Benchmarking



How can a Golden Batch be created?

Kofax Capture (before Export Connector)

KTM Project Builder

26



27

Separation Benchmark – Quality?


28

Correct Documents

Rejected Documents

Incorrect Documents

Incorrectly classified

Additional splits

Missed splits

But confidently

Document Review...?

The worst of all three categories

False Postive

KTM 5.5 – Classification Benchmarking

Classification Benchmark

29



30



31



32



33

Extraction Benchmark

KTM 5.5 – Extraction Benchmarking

Slide 34 34

Extraction Benchmark

KTM 5.5- Extraction Benchmarking

EV = Extracted Value GFV = Golden File Value (perfect file)

EV = GFV Work

EV ≠ GFV Work

EV ≠ GFV False positives

EV = GFV Super

Project quality

Project design

Slide 35 35


Extraction Benchmark - Comparison

36


37

Extraction Benchmark - Enhancements

Selection List

Sorting

By Column Content

By Status

Open in Document Viewer

Re-arrange columns

Project Merge Tool

Productivity Enhancements – Design Time

Multiple Users – One Project

KTM 5.5 – Project Merge Tool

39


Project Master

40


Copy the Project Master for each aditional user

41


Project Master

42


Copy 1

43


Copy 2

44


Merge Copy 1

45


Source and Destination projects

46


Select Classes

47


Summary

48


Save changes to destination project (Project Master)

49


Merge Copy 2

50


Source and Destination projects

51


Select Classes

52


Summary

53


Save changes to destination project (Project Master)

54


Project Master after merging

55


Elements that can be merged...

56

Classes

Fields

Locators

Validation Rules

Script

Localization

Validation Forms


57


Elements

58


59


60


61


62


63


64


Summary

65


Save changes

66


The merged project

67

Localisation

Productivity Enhancements – Users

KTM 5.5 – Localisation

KTM Languages

English

German

69


Additional KTM Languages

70

# Language Pack Language ID

1 Brazilian pt-BR

2 Chinese zh-CN

3 Czech cs

4 French fr

5 Italian it

6 Japanese ja

7 Polish pl

8 Russian ru

9 Spanish es

10 Swedish sv-SE

Additional KTM Languages

Graphic User Interface

Project Builder and runtime modules

Component based messages

KTM Server

Documentation (runtime modules and Userguide.pdf)

1. Document Review 2. Correction 3. Validation 4. Verification


71


Project Settings - Localization

72


Project Settings - Localization

73


.Net concept

Primary language

Secondary language

74

English en

English (United Kingdom) en-UK

English (United Stated) en-US


Fall back principle

75


Fall back principle

76

Localise

End

Primary – Secondary

language translation?

Yes

No

Primary language

translation?

Use default value for display name Use translation value for display name

Yes

No


KTM GUI, Server and Active Language

77


KTM GUI Language, Server and Active Language

78

The Project.ActiveLanguage overrides the Region and Language settings


Summary

KTM Graphic User Interface language

KTM Server language

Project language (Project.ActiveLanguage)

79

-

-

-


What can be localised?

80

KTM Element Yes/No Note

Fields

Table Columns

Formatting Methods Component messages used

Validation Methods Regular Expression only

Component messages used

Validation Form Tab captions

Field label

Simple label

Button captions

DB button captions

Group captions

Script Resources


Fields

81


Tables

82


Project – Script Resources

83


Project – Script Resources

84

Project.Resources.GetString("Error_Example")

KTM project folder structure

Default language in *.fpr file

Additional languages


85

Document Review

Default language

Localised languages


Localisation.xml

External editor

Language ID

Example: Field • Default value • Localised translation

86


XML Update

87


Project design language

88

Thin Client Enhancements

Productivity Enhancements – Users

New and Improved Functionality Inside KTM TC 5.5

Validation Form Layouts

Annotations

Additional Batch Editing Operations

User Settings

Advanced Login Capabilities

Combo-boxes With Descriptions

Combo-boxes Inside Tables

Other “Small” Things

KTM TC 5.5 Improvements

90

Support Validation Form Layouts

Different font types and sizes

Mini-viewers

Custom buttons

Location of fields

Anchoring

Layout localization


91

Support Annotations

Display annotations created by KTM modules

Create new annotations inside Thin Client

Edit annotations

Delete annotations

Move annotations

Hide/Display annotations


92

Additional Batch Editing Operations

Delete pages

Move, merge, delete documents

Move, merge, delete, split, create folders


93

Preserve User Settings

User name at login screen

Batch Open dialog box: size, columns, sorting settings

Panels: size, expanded states

Zoom settings: fit width, fit height, custom zoom

Annotation settings: hide/display annotations


94

Advanced Login Capabilities

Domain login for linked users

Single sign-on support for Active Directory users


95

Combo-boxes Inside Tables, Items With Descriptions

Display descriptions, values or both

Support empty strings consistently for all combo-boxes

Paging control for over 100 items

Type-ahead filtering capabilities

New script events to initialize scripted combo-boxes


96

Other “Small” Things…

Batch loading performance improvements (project caching)

PDF support

Reject/Unreject documents – support scripting on the server

Allow to install Thin Client Server on top of previous version

Propagate user changes in config files to a new version


97

Q&A

Fatih Karaoglu


Phone: +41 41 799 82 36

Email: [email protected]

For further information, please contact:

Bo Jin


Phone: +41 41 799 82 30

Email: [email protected]