Answer

The focus of this documentation is how to use Add-on Wizard and Add-on Installer to add existing custom annotators to eDiscovery Analyzer. This document will not explain how to develop such an annotator in Java or using a rule language. This documentation is primarily intended for IBM development, support or services personnel, business partners, or advanced administration personnel.

An annotator (actual text analysis component) can be provided by some third party (IBM Software Group, IBM Research, IBM Global Business Services, a business partner, custom made, open source) and delivered ready for inclusion into eDiscovery Analyzer. The Add-on Wizard and Add-on Installer packages and installs this annotator.

The prerequisite for adding a custom annotator to eDiscovery Analyzer is an existing, unmodified eDiscovery Analyzer installation that will be extended with additional analytic capabilities. Modifying the eDiscovery Analyzer installation requires write privileges to the installation directory.

Overview

The Add-on Wizard and Add-on Installer is an add-on package creation wizard and add-on package installer for additional annotators or text analysis packages for eDiscovery Analyzer.

There are five different types of annotator technologies supported by the Add-on Wizard and Add-on Installer:

1. PEAR package custom annotator: a full annotator with Java code packaged as UIMA PEAR. (see http://incubator.apache.org/uima/index.html for a description of the PEAR package format and how to create such a package).
2. File bundle custom annotator: a full annotator with Java code packaged as set of JAR and XML files. (see http://incubator.apache.org/uima/index.html for a description of how to implement UIMA annotators in Java).
3. RegEx extensions for the built-in Apache RegEx annotator: a RegEx rule file and descriptor or type system. (see http://incubator.apache.org/uima/annotators.html#regex.annotator for details on how to write rules for this annotator).
4. LanguageWare semantic dictionaries: new concepts delivered binary DIC files. (see http://www.ibm.com/software/globalization/topics/languageware/index.jsp for a description of the LanguageWare system and how to create semantic dictionaries with it).
5. System T: new concepts delivered as System T AGL and DIC files plus descriptor or type system. (see http://www.alphaworks.ibm.com/tech/systemt for a description of System T).

1. eDiscovery Analyzer Add-On Wizard
The add-on wizard helps a person who has been given access to an existing annotator in one of the above formats with the task to package this annotator up for distribution to the administrators of one or more eDA systems. Typically that person will be a solution builder for eDiscovery Analyzer (business partner, GBS, etc.) who wants to prepare additional analysis technology to be deployed to actual customers of eDiscovery Analyzer. The add-on wizard is therefore used by a person with some administrative knowledge of eDiscovery Analyzer, some basic knowledge of what the annotator components do (but not necessarily how they do it), some basic knowledge of UIMA (but not necessarily knowing the details of UIMA descriptor configuration or UIMA Java programming or LanguageWare or System T).

The add-on wizard packages either an individual annotator (delivered in one of the supported text analysis technologies) or packages of several annotators. The add-on wizard:

• Auto-detects the specific text analysis technology in which the annotator is delivered.
• Determines what concepts the annotator creates; for example what UIMA output types the annotator declares.
• Asks the user to specify which of the concepts created by the annotator should be used and which aspects of them to use; for example, which UIMA types and features to actually use.
• Validates an existing eDiscovery Analyzer installation and dynamically uses this installation as the base line for the installation.
• Asks the user to specify the display labels to use in the eDiscovery Analyzer user interface and creates the necessary changes to the eDiscovery Analyzer configuration and messages.

The add-on wizard creates an edaaddon extension package file similar to a compressed ZIP file, with the annotator contents packaged inside. This output is delivered to other people who can use add-on installer for installation of the .edaaddon package on different machines.

2. eDiscovery Analyzer Add-On Installer

The add-on installer installs the annotator package that was created by the add-on wizard into an existing eDiscovery Analyzer installation.

The add-on installer:

• Copies all new files to their required locations.
• Modifies eDiscovery Analyzer configuration files.
• Creates a change log that lists each copied or modified file.
• Creates backup files of the modified files.
• Creates an uninstaller for the add-on package.
• Verifies that the installation results in a working UIMA aggregate.

To package annotators:

GUI mode:

1. Run the add-on wizard program. On Windows, run the addon_packager.exe file. On AIX, run the addon_packager.bin file.
2. On the first panel, enter the package details:
• The default eDiscovery Analyzer installation path is shown for Windows (C:\Program Files\IBM\eDiscoveryAnalyzer) or for AIX (opt/IBM/eDiscoveryAnalyzer). If eDiscovery Analyzer is not installed in the default path and user doesn’t change the default path, a warning message is shown as invalid eDiscovery Analyzer installation”
• Specify the annotator package (text analysis component) to convert into an edaaddon file. The edaaddon file is the input for add-on installer.
• Specify the output directory for the edaaddon file.
3. On the second input panel, enter the display option details:
• Specify the facet name to be shown in the eDiscovery Analyzer search interface. Only alphabetical characters, numerals, underscore and spaces are valid.
• Select the UIMA type
• Select the feature to be indexed
4. If the add-on is successfully packaged, the output edaaddon file is available in the output directory that was specified in the first panel.
5. To package this annotator again in future simply run the wizard in silent mode and use the response file created in the log directory of the eDiscovery Analyzer installation. The file is named <facetname>.addon.packager.properties.

Silent mode:

1. From the Windows command line, run addon_packager.exe. From the AIX command prompt, run addon_packager.bin. The syntax for the package command is:
   addon_packager.exe -i silent -f <response file>
2. The response file must include the following input parameters:
• LICENSE_ACCEPTED Set to true to accept the terms of the license agreement.
• USER_INSTALL_DIR eDiscovery Analyzer installation path (default is C:\Program Files\IBM\eDiscoveryAnalyzer)
• ANNOTATOR_INPUT_DIR Path of the annotator to convert into an edaaddon file
• ANNOTATOR_OUTPUT_DIR Output directory for the edaaddon file
• FACET_NAME Name of the facet to be shown in the eDiscovery Analyzer search interface (e.g phone)
• UIMA_TYPE UIMA type of the facet (e.g. com.sample.namespace.PhoneNumber)
• FEATURE Feature to be indexed (e.g. /:coveredText())
3. If the add-on is successfully packaged, the output edaaddon file is available in the output directory that was specified in the first panel.

To install the annotator package:

GUI mode:

1. Stop the eDiscovery Analyzer server.
2. Run the add-on installation program. On Windows, run addon_installer.exe. On AIX, run addon_installer.bin.
3. On the first panel, enter the package details:
• The default eDA installation path is shown for Windows (C:\Program Files\IBM\eDiscoveryAnalyzer) or for AIX (opt/IBM/eDiscoveryAnalyzer). If eDiscovery Analyzer is not installed in the default path and you do not change the default path, a warning message is shown as invalid eDiscovery Analyzer installation”
• Select the edaaddon file that was generated by add-on wizard to be installed.
4. On the second panel the facet name that was specified for the edaaddon file is displayed. You can change the facet name in the panel before installation. If the facet name entered already exists a warning message is displayed. The facet name must be unique for each annotator installed on eDiscovery Analyzer.
5. On successful completion, annotator is installed on the existing eDiscovery Analyzer specified during installation.
6. Start the eDiscovery Analyzer server.
7. To install this annotator again in future simply run the installer in silent mode and use the response file created in the log directory of the eDiscovery Analyzer installation. The file is named <facetname>.addon.installer.properties.

Silent mode:

1. Stop the eDiscovery Analyzer server.
2. From the Windows command line, run addon_installer.exe. From the AIX command prompt, run addon_installer.bin. The syntax for the install command is:
   addon_installer.exe -i silent -f <response file>
3. The response file must include the following input parameters:
• LICENSE_ACCEPTED Set to true to accept the terms of the license agreement.
• USER_INSTALL_DIR eDiscovery Analyzer installation path (default is C:\Program Files\IBM\eDiscoveryAnalyzer)
• ANNOTATOR_FILE Path of the annotator edaaddon file
• FACET_NAME Name of the facet to be shown in the eDiscovery Analyzer search interface (e.g phone)
4. On successful completion, annotator is installed on the existing eDiscovery Analyzer specified during installation.
5. Start the eDiscovery Analyzer server.

Using the new annotator

After installing the new annotator, you can start eDiscovery Analyzer (stop it first if it is still running). Navigate to the Administration page, select Manage Default Preferences. On the Search Categories page you should see all the new user defined facets that were installed with the new annotator. Select the annotator to see in the search interface and save it.

To see the new facet in the search interface:

1. Create a new case, process some documents, and then check the search interface for the new case to see what the changes look like, or reindex an existing case after saving the new search categories.
2. You should see a new category (facet) section, similar to the categories that show phrases and names, but showing the new facet name that you specified during annotator installation.

To remove annotators

GUI mode:

1. Stop the eDiscovery Analyzer server.
2. Run the add-on uninstallation program in tadonis_Uninstall. On Windows, run uninstall_addon.exe. On AIX, run uninstall_addon.
3. On the first panel, select the facet name of the annotator to be uninstalled and Click 'Next.'
4. On the second panel, verify that the correct facet name was selected and click 'Uninstall.'
5. After the annotator is removed, start the eDiscovery Analyzer server, or stop and start the server if it is already running. You should not see the facet under search categories in Manage Default Preferences.

Silent mode:

1. Stop the eDiscovery Analyzer server.
2. Run the add-on uninstallation program located in the tadonis_Uninstall directory. On Windows, run uninstall_addon.exe. On AIX, run uninstall_addon.
3. In the response file, specify the following parameter:
• ADDON_SPEC_FILE Path of the annotator specification file. The file is named eDA_addOn_installationspec.xml and is located inside the tadonis_Uninstall directory.
4. After the annotator is removed, start the eDiscovery Analyzer server, or stop and start the server if it is already running. You should not see the facet under search categories in Manage Default Preferences.

Debugging and Logging

During annotator installation, the tadonis.log file is created in the log folder with all the details of add-on installation. The log file contains the details of the files created and files modified. In case of any error or exception, the log file shows the full stack trace and the specific error message.

Limitations

1. Cancel button limitation: If you click the Cancel button in the add-on wizard after the first input panel or any time during the add-on creation, you must manually clear the output directory. If you click the Cancel button in the add-on installer after the first input panel in the installer, you must manually clear the package directory to remove the extra files and folders other than edaaddon file.
2. You cannot generate the edaaddon file in the annotator package directory. The same path for output directory and the annotator package is not allowed. Choose a different output directory.
3. Facet name creation limitation: Only alphabetical characters, numerals, underscore, and space are valid.
4. You can use the Add-on Wizard and Add-on Installer only on eDiscovery Analyzer Version 2.1.1 or later.
5. A RegEx annotator with embedded type system is not supported. The annotator must have a separate type system file.
6. The new facet will not appear in the search interface unless e-mails or documents actually have corresponding content that is detected by the annotator.
7. The eDiscovery Analyzer server must be restarted, and existing cases must be reindexed to see the new facet in search interface.
8. License Agreement: if available, a license agreement should be included in the annotator package as a text file named license.txt. The license agreement will be displayed in the Add-on Installer during installation of the annotator.
9. No search features other than facets are supported.
10. Only English language graphical user-interface mode is supported for the installer.
11. On AIX, the speed of both the add-on wizard and installer is very slow. It can take several minutes to complete each of these processes.