A quick User Guide of WISDOM++

Contents

1. Installing WISDOM++2. The Graphical User Interface of WISDOM++3. The File Menu4. The Operate Menu5. The View Menu6. The Parameters Menu7. The Learning Menu8. The Management Menu

9. The Help Menu

10. Processing a document

1. Installing WISDOM++

1.1 System Requirements for WISDOM++

WISDOM++ can be installed on a computer with the following minimal system requirements:

1.2 Running the installation program

Whilst in Windows, click on the icon of the setup.exe file. After a few seconds, the installation program's welcome screen will be displayed, where upon you just follow the on-screen instructions.
The installation of WISDOM++ is automatic: at each stage you will typically be presented with a dialog containing three buttons, labelled  "Back", "Next " and "Cancel". If you click on the "Next" button or press the <enter> key, you will be taken on to the next stage of installation. Should you wish to return to an earlier screen to correct a previous entry, this can be achieved by pressing the "Back" button.You can choose to abort installation at any point by clicking on the "Cancel" button or pressing the <esc> key. If you choose to abort, you can start the installation process again whenever you want to.
When WISDOM++ is transferred to your hard disk, it will need to be assigned a home directory in which to keep all its files. Within the Choose Destination Location dialog, you will be asked for the name you want to use for this directory. The recommended name c:\wisdomis pre-filled in the text entry box. You can accept the default name or you can choose another name simply by typing it into the edit box, using the cursor keys and mouse as required to avoid any mistakes. Should you browse your hard disk, a Browse option is provided from where your choice of home directory can also be made. Click on the "Next" button or press the <enter> key when you are ready to go ahead.
Notice that the installation program will overwrite exisitng files or directories of the same name.
As well as a home directory, Windows applications usually have at least one shortcut associated with them, normally pointing to the application's principle executable file. In order to access WISDOM++ effectively, the shortcut Wisdom++ 2.0 will be created automatically. Within the Folder Selection dialog, you should select either an existing or a new folder (group) where you wish such shortcuts to appear. The default name is Wisdom Application.
Once you have entered all the required details, the installation program will copy all files in the proper sub-directories of the home directory you have choosen. Throughout this process, you will be shown the names of the files and sub-directories as they are being created.
Once all files have been successfully copied into the destination location,you will be presented with a final dialog. At this step, you can't abort the installation procedure any more. Simply click on the "OK" button or press the <enter> key once you have read the message. The installation program will then terminate, leaving you in Windows.

The directory Wisdom has the following sub-directories:
<
Wisdom\AtreRecognizer based on the learning system ATRE
Wisdom\Atre\AtreApplicationEmbedded version of ATRE which can be run in background
Wisdom\Atre\sp38 and Wisdom\Atre\AtreApplication\sp38SICStus Prolog working archive
BinOCR archive
Wisdom\BlocksBlock classifier based on the decision tree learning system ITI 
Wisdom\Users\rootArchive of the user root
Wisdom\Users\<User>\Doc<user>'s own archive of documents 
Wisdom\Users\<User>\Rules<user>'s own base of rules
Wisdom\Users\<User>\Rules\Classif<user>'s own archive for document classification descriptions used by ATRE 
Wisdom\Users\<User>\Rules\Layout<user>'s own archive for document layout descriptions used by ATRE
Wisdom\Users\<User>\Rules\Under<user>'s own archive for document understanding descriptions used by ATRE
Wisdom\Users\<User>\Rules\Under\Tpami<user>'s own archive for document understanding descriptions for tpami documents
Wisdom\Users\<User>\Rules\Under\<class><user>'s own archive for document understanding descriptions for <class> documents
Wisdom\Users\<User>\Xml-html<user>'s own class dependent .css and .dtd files
 
Before running the system, the user has to set the database connection in order to store and retrieve all information necessary for the document processing. WISDOM++ is able to interface two different DataBase Management Systems (DBMS), namely Microsoft Access (97 or 2000) and Oracle (8i or 9i). To interface with the Access Database the following setting steps are necessary: Star/settings/control Panel. Open the ODBC Data Sources. Open the tab users DNS. Click on "Add…" button and select the "driver for Microsoft Access" click on Finish. At this point the configuration window will appear. In the DataSource Name specify "WisdomDataBase" (pay attention to the upper/lower case). Click on the "select" button in order to specify the database. The default database is "C:\wisdom\WisdomDataBase.mdb". Click on OK.
Moreover, the user should run the wisdom.reg file and if the user is working with Windows XP has to select the application compatibility with Windows 98.
 

1.3 Uninstalling WISDOM++

During the installation process, a shortcut called Uninstall WISDOM++ 2.0 is created which, when executed, allows the application to be uninstalled.Only directories (if empty) that were created as a result of the installation process will be deleted, where as your own files and directories will remain intact.

Up to the contents

2. The Graphical User Interface of WISDOM++

WISDOM++ is a SDI (Single Document Interface) application developed withthe MFC (Microsoft Foundation Classes) library.
 

2.1 The Main Window

When WISDOM++ starts up, it initially displays a graphical window showing the LACAM (Laboratorio di Acquisizione della Conoscenza e Apprendimentonelle Macchine) logo, which remains on the screen until the mouse cursor is moved to the banner and the left button is pressed.
Once the banner window has been replaced with the main window, the user is required to enter a login and a password in order to be identified by the system. Currently, the following user profiles are active:
 
LoginPasswordClass
rootrootA
<User-name><User-password>U
where A stands for user administrator and U stands for final user. Administrator users can access to documents and stored information of all the final users as well as to learning operations management.
The main window is the command centre since operations are launched from the menu bar or, in case of the most relevant ones among them, from the tool bar.
 

2.2 The Menu Bar

The menu bar consists of seven menus which are briefly described here.
The File menu provides a set of functions for file management, such as the opening and saving of document images.
The Operate menu contains the commands to process documents, such as the classification and understanding of documents.
The View menu contains the commands to customize the display mode within the SDI window like facilities for showing/hiding and zooming.
The Parameters menu allowsyou to tailor the WISDOM++ environment to suit your own particular needs, such as settings for the transformation of a document image into a XML file.
The Learning menu contains the commands to prepare and start up learning/recognition processes.
The Management menu allows the user administrator to manage the database of the authorised users.
The Help menu provides the standard Windows functions for accessing the WISDOM++ help file. 
 

Up to the contents

3. The File Menu

The File menu provides a number of options concerning file management as well as some options particularly related to the application domain of WISDOM++.
 

3.1 The Open Option

This option is used to load existing files. Since WISDOM++ is a SDI application,it is possible to open only one file at a time.
When you select the File/Openoption, the list of all user documents, stored in database, appears. The user selects the document that he/she wants to manage, and then clicks on OK button.The first page of the selected document is displayed. If the document is multi-page, the user can navigate through it by four buttons corresponding to then “first page”, “previous page”, “next page” and “last page” options. These buttons are on the right side of the toolbar. The system WISDOM++ can manage 300dpi black-and-white document images whose sizes are at most 210.00´297.00 mm (A4-format). To perform a conversion of document images, WISDOM++ uses an integrated library that automatically transforms document images into a format satisfying its requirements.
 

3.2 The Save Image Option

This option is used to save the document image (.tif file) currently opened in the SDI window. If no document image is opened or the document image currently opened has not been processed yet, then this option is greyed out.
 

3.3 The Save Image As Option

This option is similar to File/SaveImage except that you are always prompted for a file name.If no document image  is opened or the image document  currently opened has not been processed yet, then this option is greyed out.
When you select the  File/SaveImage As option, you will be presented with the SaveImage As dialog. Just like the Opendialog (see section 3.1), you can select file names or directories from the Files and Paths boxes respectively, or you can simply type a new name into the File box.
 

3.4 The Save Layout Option

This option is used to save in the database the layout of the document image currently opened in the SDI window. If the layout of the document image currently opened has not been analysed yet, then this option is greyed out.
 

3.5 The Import/Export Layout Options

This option is used to save/load in/from the database the layout of a document image to select. This option is enabled only for administrator users,otherwise it is greyed out.
 

3.6 The Import/Export Rules Options

This option is used to save/load in/from the database the task specific (layout correction,classification and understanding) rules learned by means of ATRE. It requires that the user selects both the user and the task.This option is enabled only for administrator users,otherwise it is greyed out.
 

3.7 The Close Option

This option is used to dismiss the currently loaded document image/layout file. If no document image/layout file is opened then this option is greyed out. If you attempt to close a SDI window which has not been saved since the last update, the application  will offer to save the associated disk file before closing it.
 

3.8 The Create&Modify Multipage Option

When you select this option, you will be presented with the Create&ModifyMultipage Document dialog. In the text box, the user writes the name of the new document and clicks on the “Ok New Document” button. At this point, it is necessary to list the pages that form the document. If the page has already been scanned, it can be selected from file system through “Add Page from File” button. Selected pages can be of any image format because once chosen they will be transformed by WISDOM++ in its tiff format and saved in the "Doc" folder of the user. Otherwise, if the page is in a paper format yet, it can be selected through the “Add Page from Scanner” button. In this case, the scanner acquires the page, transforms it in a digital format and adds it to the new document. The selection has to be repeated for each document page. In this session the user defines the sequence of pages into document. At the end of the document creation, the user clicks on the OK button and the entire document is stored into database. If the user wants to modify a document (by deleting or adding pages), he/she has to select the name of the document from the text box. All document pages appear into the list box. The page(s) to delete has(have) to be selected, and then the user clicks on “Delete Selected Pages” button. Deleted pages disappear from the list box, and they are deleted from database. If the user deletes all listed pages the whole document will be deleted from database.
 

3.9 The Acquisition Option

This option is provided to allow the user to acquire a document by means of a scanner. It is possible to select the source and to setup some parameters related to the embedded Twain Manager.
 

3.10 The Exit Option

This option is provided  to allow you to exit from WISDOM++ at the end of a session.
 

Up to the contents

4. The Operate Menu

The Operate menu contains the commands to process documents.
 

4.1 The Evaluate Skew Option

This option allows the user to evaluate the skew angle (further details) of the currently loaded document image.
Once the skew angle has been evaluated, you will be presented with the Skew Information dialog showing you the value of the skew angle as well as the spread factor.
 

4.2 The Rotate Option

This option allows the user to rotate (further details) the currently loaded document image with respect to the skew angle.
 

4.3 The Analyze Layout Option

This option allows the user to trigger off the analysis of the currently loaded document image as regards the segmentation of the document image into blocks (further details), the classification of the blocks (further details) and the detection of structures among the blocks (further details).
When you select this option, you will be presented with the Spreadfactor and Thresholds dialog for the setting of  the parameters required by the segmentation step. Each parameter has an edit box showing a default value. You can either accept the default values or edit the desidered values or ask the system to recompute the spread factor and estimate the smoothing thresholds according to the spread factor. You can also run the layout analysis in the interactive mode which allows the user to modify the results of the layout analysis. During the blocks classification step, a progress bar is displayed.
Once the layout structure has been extracted, the currently loaded document image will be replaced by its layout structure showing basic blocks enclosing any kind of content.
A tool bar in the right side of the SDI window allows the user to set the layout level and the content type of the blocks to be displayed by the layout structure.
Options for the levels of the layout structure are the following:
 
Button labelLayout level
BBbasic blocks
LNlines
SLset of lines
F1frame 1
F2frame 2
 

Options for the types of content are the following:
 
Button labelContent type
TXtext
HLhorizontal line
VLvertical line
IMpicture
GRgraphic
MXmixed
ALLall content types
 

4.4 The Update Block Classifier Option

This option allows the user to train the block classifier on-line in order to update it according to the correct classification of the blocks which have been misclassified by the current decision tree (further details). If the layout of the currently loaded document image has not been yet analysed then this option is greyed out.
To classify correctly a misclassified block, double-click on the block.You will be presented with a dialog which allows you to assign the right physical type to the block.
 

4.5 The Correct Global Analysis Option

The layout analysis is performed in two steps one at a global level and another at local level (further details). Layout components resulting from the global analysis are columns and sections. The user can see sections or columns by choosing “View/Document Sections” option or “View/Document Columns” option in the main menu. The areas (columns/section) of the document are superimposed to the layout structure extracted by the local analysis.The user can correct the results of the global analysis by selecting the “Operate/Correct Global analysis” option in the main menu. To apply a correction, the user selects an area with a double click. It becomes light green and then the user can either vertically split, or horizontally split or group together with another area. The cut point in the two splitting operations is automatically determined, while grouping is possible only if some overlapping/nesting conditions are not satisfied. Immediately after each splitting/grouping operation, WISDOM++ computes the result of the local analysis process, so that the user can immediately perceive the final effect of the requested corrections and can decide whether to confirm the correction or not.
 

4.6 The Classify Option

This option allows the user to trigger off the document classification step on the currently loaded document image. This step applies some classification rules induced off-line by a first-order rule learning system (further details). If the layout of the currently loaded document image has not been yet analysed or no document layout has been loaded from the database then this option is greyed out.
Once the document has been classified, the class name will be displayed in the status bar of the SDI window.
 

4.7 The Understand Option

This option allows the user to trigger off the document understanding step on the currently loaded document image. This step applies some understanding rules induced off-line by a first-order rule learning system (further details). If the currently loaded document has not been yet classified then this option is greyed out.
Once the document has been understood, its layout structure will be replaced by its logical structure showing the labels assigned to the layout components.
 

4.8 The OCR Option

This option allows the user to trigger off the optical character recognition step on the logical components of interest of the currently loaded document image. If the layout of the currently loaded document image has not been yet analysed or no document layout has been loaded from the database then this option is greyed out.
 

4.9 The Generate HTML Option

This option allows the user to trigger off the transformation step of thecurrently loaded document image into a web-accessible format. If the layout of the currently loaded document image has not been yet analysed or no document layout has been loaded from the database then this option is greyed out.
 

Up to the contents

5. The View Menu

The View menu contains the commands to customize the display mode within the SDI window like facilities for showing/hiding and zooming.
 

5.1 The Image Option

This option allows the user to display only the image of the document currently opened in the SDI window. The option is disabled if the layout of the currently loaded document has not been yet analysed or no document layout has been loaded from the database then this option is greyed out. Moreover, it's mutually exclusive with the Layout(see section 5.2) and Background(see section 5.3) options.
 

5.2 The Layout Option

This option allows the user to display only the layout structure of the document currently opened in the SDI window. The option is disabled ifthe layout of the currently loaded document has not been yet analysed or or no document layout has been loaded from the database. Moreover, it's mutually exclusive with the Image (see section 5.1) andBackground (see section 5.3) options.
 

5.3 The Background Option

This option allows the user to display both the image and the layout structure of the document currently opened in the SDI window. The option is disabled if the layout of the currently loaded document has not been yet analysed or no document layout has been loaded from the database. Moreover, it's mutually exclusive with the Image (see section 5.1)and Layout (see section 5.2) options.
 

5.4 The Zoom Option

This option allows the user to zoom in/out the document layout/image file currently opened in the SDI window.
 

5.5 The Document's Sections Option

This option allows the user to view sections of the document layout/image file currently opened in the SDI window.
 

5.6 The Document's Columns Option

This option allows the user to view columns of the document layout/image file currently opened in the SDI window.
 

5.7 The Hide columns/sections Option

This option allows the user to hide columns/sections of the document layout/image file currently opened in the SDI window.
 

5.8 The Show Label Option

This option is used to show/hide the labels associated to the layout components of the document layout/image file currently opened in the SDI window. The option is disabled if the document has not been yet understood.
 

5.9 The Image Info Option

This option allows the user to get information about the document image file currently opened in the SDI window.
 

5.10 The Layout Info Option

This option allows the user to get information about the layout structure of the document currently displayed in the SDI window. The option is disabled if the layout structure of the current document has not been yet analysed or no layout document file is currently opened.
 

5.11 The Document Description Option

This option allows the user to access in reading mode the symbolic description of the document currently displayed in the SDI window. Such description is written according to a 0-order representation language when the document has to be input to the block classification step (this is loadable after the layout analysis step) whereas according to a 1-order representation language when the document has to be input to the document classification and understanding steps (both of them are loadable after using the classification and the understanding options, respectively).
The option is disabled if the layout structure of the current document has not been yet analysed or no layout document file is currently opened.
 

5.12 The Rule Description Option

This option allows the user to access in reading mode the description of the rules which can be applied to the document currently displayed in the SDI window. Such description is a decision tree in case of the block classification step (this is loadable after the layout analysis step) whereas is a set of first-order rules in case of the document classification and understanding steps (both of them are loadable after using the classification and the understanding options, respectively).
The option is disabled if the layout structure of the current document has not been yet analysed or no layout document file is currently opened.
 

Up to the contents

6. The Parameters Menu

The Parameters menu contains the commands to set up parameters related to several system components.
 

6.1 The Set Up OCR Option

This option allows the user to choose the OCR executable to be run as wellas the kind of logical components which the OCR has to be applied to for transforming the document image into a web-accessibile format.
 

6.2 The Set Up Training Option

This option allows the user to choose the decision tree to be trained during the block classification step, to enable the test of predictive accuracy and to set up the pruning.
 

6.3 The Set Up XML Option

This option allows the user to set up the parameters (such as formatting and rendering) required by the generator of XML code. You can also setup the layout of the XML page to be generated by means of the Cascade Style Sheets (creating, loading and saving of .css files).
 

Up to the contents

7. The Learning Menu

The Learning menu contains the commands to set up parameters related to the learning systems. 

7.1 Blocks Learning: Training ITI off-line

1. Choose the block classifier to be trained (click on the "Parameters/SetUp Trainer" option, see 6.1)
2. Train the chosen block classifier selecting one of the available modes (Batch, Incremental, Error Correction and Test, see bibliography on block classification functionality for details) with respect to a selected data set.
 

7.2 Start Learning Process: Training ATRE

This functionality allows to set and start the training of the ATRE system. Choose the user, the learning task, the data set (by selecting documents already processed the system contemporaneously generates training descriptions for the specific task), the output location (report and rule files), the possible background knowledge, the set of concepts to learn and finally a window, which summurizes all selected settings, will appear. At this moment, the user can choose to go back to correct some settings or to start the learning session by clicking the "Start Process..." button. Before running the learning system, the user should tune learning parameters and set the process priority.
 

7.3 Display Processes: Monitoring ATRE

By clicking on this menu option the user can both view the history of terminated processes and monitor the active processes (by double click on the process of interest a window with current status of ATRE will appear). Indeed, when the user starts up a learning session, ATRE runs in background mode.
 

7.4 Start Recognition Process: Testing ATRE

This functionality allows to set and start the testing of the ATRE system. Choose the user, the learning task, the data set (by selecting documents already processed the system contemporaneously generates testing descriptions for the specific task), the rule set to test, the output location, the possible background knowledge and finally a window, which summurizes all selected settings, will appear. At this moment, the user can choose to go back to correct some settings or to start the testing session by clicking the "Start Process..." button. Before running the learning system, the user can check the testing descriptions, while after the end of the process (a message will appear in the summary box) the user can view the final testing report.
 

Up to the contents

8. The Management Menu

This menu currently contains only the User Administration option which allows the user administrator to manage the database of the authorised users. The suboptions Add, Delete and Modify supply the usual operators for adding, deleting and updating respectively and are accessible only by the admin user.
 

Up to the contents

9. The Help Menu

This menu provides the standard Windows functions for accessing the WISDOM++ help file.
The "Help Topics" option is currently under construction.
The "About WISDOM++" option is used to display version and license information about your copy of WISDOM++.
 

Up to the contents

10. Processing a document

Some sample sessions of document processing with WISDOM++ will be illustrated in the following.
When WISDOM++ starts up, it initially displays a graphical window showing the LACAM (Knowledge Acquisition &Machine Learning Lab) logo, which remains on the screen until the user clicks on the logo. A message box advising that the user has 5 days of to use the embedded OCR appears. The main window appears and the user inserts his/her Login and Password in order to be identified by the system. If the data are right, the user can manage his/her documents working with system, otherwise WISDOM++ returns an error message and then the user has to start up again the system.
 

10.1 Acquiring a single page document

1. Select a source (click on the "File/Acquisition/Select source.." option, see 3.8)
2. Set up parameters related to the embedded Twain Manager (click onthe "File/Acquisition/Acquisition Setup .." option, see 3.8)
3. Acquire the document image (click on the "File/Acquisition/Acquire.." option, see 3.8)

10.2 Processing a single page document in training mode

1. Open an existing single page document (click on the "File/Open.." option, see 3.1) if the association between the document name and the tiff file has been already stored in the database, otherwise use option "File/Create&ModifyMultipage Document" (see 3.8).
2. Read the information about the document image (click on the "View/ImageInfo" option, see 5.9).
3. Evaluate the skew angle of the document image (click on the "Operate/EvaluateSkew" option, see 4.1).
4. Rotate the document image (click on the "Operate/Rotate"option, see 4.2).
5. Save the pre-processed document image
    5.a to overwrite the file: click on the "File/SaveImage .." option, see 3.2;
    5.b to create a new file: click on the "File/SaveImage As .." option, see 3.3.
6. Analyze the layout of the document image (click on the "Operate/AnalyzeLayout" option, see 4.3).
7. Read the information about the layout structure (click on the "View/Layout Info" option, see 5.10).
8. If the resulting layout doesn't allow to label individual logical components of interest, the user can:
    8.a train the block classifier on-line by double click directly on the misclassified basic blocks, selection of the correct class in the "Block Type" pop-up menu, and then clicking on the "Operate/Update Block Classifier" option, see 4.4) 
  8.b try to correct the global analysis result (click on the "Operate/Correct Global Analysis"option, see 4.5).
9. Classify manually the document by double click on the document image (not inside layout blocks), a dialog window will appear which allows the user to select the document class label.
10. Understand manually the document by double click on each individual layout block that corresponds to a logical components. A dialog window will appear which allows the user to select the logical labels belonging to the particular document class.
11. Apply the OCR by clicking on the "Operate/OCR"option (see 4.8) or by double click on the individual logical block and then clicking on the "Recognize" button (the user can view the result by double click again on the individual logical block and can correct possible OCR errors).
12. Save all extracted information (click on the "File/Layout .." option, see 3.4).
13. Generate the XML document (click on the "Operate/Generate XML"option, see 4.9).

10.3 Processing a single page document using the learned knowledge base

1. Open an existing single page document (click on the "File/Open.." option, see 3.1) if the association between the document name and the tiff file has been already stored in the database, otherwise use option "File/Create&ModifyMultipage Document" (see 3.8).
2. Read the information about the document image (click on the "View/ImageInfo" option, see 5.9)
3. Evaluate the skew angle of the document image (click on the "Operate/EvaluateSkew" option, see 4.1)
4. Rotate the document image (click on the "Operate/Rotate"option, see 4.2)
5. Save the pre-processed document image
    5.a To overwrite the file: click on the "File/SaveImage .." option, see 3.2;
    5.b To create a new file: click on the "File/SaveImage As .." option, see 3.3;
6. Analyze the layout of the document image (click on the "Operate/AnalyzeLayout" option, see 4.3)
7. Read the information about the layout structure (click on the "View/Layout Info" option, see 5.10)
8. Correct the global analysis result (click on the "Operate/Correct Global Analysis"option, see 4.5)
9. Classify the document (click on the "Operate/Classify"option, see 4.6)
10. Understand the document (click on the "Operate/Understand"option, see 4.7)
11. Apply the OCR by clicking on the "Operate/OCR"option (see 4.8) or by double click on the individual logical block and then clicking on the "Recognize" button (the user can view the result by double click again on the individual logical block and can correct possible OCR errors).
12. Save all extracted information (click on the "File/Layout .." option, see 3.4).
13. Generate the XML document (click on the "Operate/Generate XML"option, see 4.9).
 

Up to the contents

 

Back to home page

berardi@di.uniba.it