Release notes - IBM SPSS Text Analysis for Surveys 4.0.1

Content

Release notes - IBM SPSS Text Analytics for Surveys 4.0.1

IBM® SPSS® Text Analytics for Surveys 4.0.1 is available. Installation and other getting-started issues are addressed.

Contents

• Description
• System requirements and installation
• Known issues

Description

For a list of new features in the latest SPSS Text Analytics for Surveys release, see the topic “What's New” in the online help.

System requirements and installation

For information on system requirements and installation, see the installation documents.

For information on downloading the product, see the Download Document.

Known issues

At time of publication, the following issues were known.

Problem: Saving a project where extractions are out of date may result in out of synchronisation extraction results.
Solution: If you have changed the extraction parameter before saving the project then extraction results may not be updated. To resolve this issue extract after you reopen the project.

Problem: When using the Import Predefined Categories option, from the Categories menu, code frames do not replace all existing categories.
Solution: If you want to replace some existing categories with categories imported via the Import Predefined Categories option, delete the existing categories before importing.

Problem: cls_appli.exe crashes when building categories after changing the template. The following error message is displayed: "cls_appli.exe has encountered a problem and needs to close."
Solution to work around this:

1) Save the current project.

2) Exit from TAfS.

3) Restart TAfS, and re-open the saved project.

4) When the project has been loaded and extraction has been completed, click Build in the Categories panel to build a set of categories.

Problem: A TAP created with TAfS 4.0.1 cannot be imported into Text Analytics.
Solution: Currently there is no resolution for this issue.

Problem: Breaking up large projects. When working with data sources with many text and reference variables, TAfS will sometimes perform very sluggishly.
Solution: If you plan on analyzing more than 6 open end text questions and associated reference variables, it is recommended that you create multiple projects.

Problem: No version control for .TAS files.
Solution: Project files (.TAS) were introduced in TAfS 2.0. The product does not support version control for these files. If two or more users access the same .TAS file it is possible to undo changes made by another user. One option for tracking changes is to place comments within the Project Properties Annotation field. Additionally, if a user attempts to open a file that is currently in the process of being saved by another user a warning stating "Unexpected end of ZLIB stream" will appear. If this happens, make sure the save process has completed before attempting to open. This scenario may happen when multiple users access a file on a shared network drive.

Problem: Restricted characters when naming TAfS files.
Solution: When naming your .TAS file, do not use the following characters: \ / * ? " < > |. Naming files with these characters can cause problems and you may lose work you have saved.

Problem: Active content blocking and TAfS help.
Solution: Most Help features in this application use technology based on Microsoft Internet Explorer. Some versions of Internet Explorer (including the version provided with Microsoft Windows XP, Service Pack 2) will, by default, block what it considers to be "active content" in Internet Explorer windows on your local computer. This default setting may result in some blocked content in Help features. The following steps will enable TAfS Help if it is being blocked. 1) From the Internet Explorer menus choose: Tools > Internet Options... 2) Click the Advanced tab. 3) Scroll down to the Security section. 4) Select (check) Allow active content to run in files on My Computer.

Problem: Publishing libraries in a saved state.
Solution: Under some conditions libraries published will not be saved with a project. If you open an existing project and immediately publish a library or, if you start a new project do some work, save and then publish a library you will be unable to save. Publishing a library in a saved or "clean" state will not enable you to save your project until you perform some other action (e.g., creating a new library, extracting, creating categories).

Problem: UNC Paths in File Chooser dialogs.
Solution: If you wish to read or write to a file located on your network, there is a limitation that you cannot directly enter just the network device name preceded by 2 back slashes (i.e. \\mynetworkcomputer ) in the "File Name" text box of the dialogs. You can enter a qualified path (i.e. \\mynetworkcomputer\somedirectory ) or you can navigate using the "My Network Places" folder from the "Look in" drop-down.

Problem: Forbidden characters for Excel worksheets.
Solution: Excel does not allow : ,, ; '' ? ! characters in their worksheet titles. TAfS uses question labels to brand the Excel worksheets. If your question label contains one of these characters you will receive a notification - 'Repairs to "filename".xls'. Your worksheet will be re-named to an acceptable default.

Problem: Controls in Excel column headings.
Solution: The driver used by TAfS for importing Excel files cannot handle importing worksheets where the column headings have controls (for example, drop-down control) embedded in them. In order to import such data sources, please remove the control from all header columns.

Problem: Importing pre-'97 versions of Excel.
Solution: TAfS cannot import Microsoft Excel files from versions prior to Office '97. It is recommended that you open the .xls file in a more current version of Excel and resave it. Then import the new file into TAfS.

Problem: Importing Excel files exported from OpenOffice.Org.
Solution: Under certain circumstances - such as records with a high degree of duplication - TAfS takes a long time to open Excel (.xls) files that were exported from OpenOffice.org, the freeware personal productivity suite. In some cases, the application may freeze and have to be closed via Windows Task Manager. In these circumstances, you have one of two options: 1) Open the file in TAfS via ODBC 2) Open the file in Office 97 or later version of Excel, copy/paste all contents to a new Excel file, and resave. Simply opening and saving is not sufficient.

Problem: Formula cells in Excel.
Solution: In Excel, if a "=" appears before text data the application will believe it is a formula and the cell will contain the words '#NAME?' and have a tool tip that says "The formula contains unrecognized text." In order for this text to be analyzed in TAfS you must remove the "=" otherwise it will be ignored.

Problem: Importing .SAV files from SPSS Statistics or Data Collection.
Solution: TAfS can import .SAV files generated from SPSS 12.0 - 16, Statistics 17 or greater, Data Entry 3.0.3 or greater, and Data Collection. The maximum size of a string field in an imported .SAV file must be less than 4000 characters.

Problem: Changing variable names in source data.
Solution: TAfS does not store source data (i.e., the .sav, .xls, ODBC or Data Collection data where your text data is stored) in its projects. Instead, it stores a reference to that data source on your machine. If someone changes one or more of the variable names (or column headers) in the source data that were imported into a TAfS project, a warning will appear that the data is not found the next time a user attempts to open that project. The user will have to re-import the data and match the "new" questions with the original variable names to continue working with that project. Alternatively, the user can place the data back in the original folder location. In general, it is not recommended that you rename variables or column headers in your source data after you've imported them into TAfS.

Problem: Inflection error with multi-word terms in user-defined types.
Solution: The TAfS extraction engine will not generate the inflected forms of multi-word terms containing at least one word with fewer than 3 letters in a user-defined type dictionary. For example, "type of cars" and "type of car" will be recognized as the same thing if naturally extracted. However, if users create a new type and add the term "type of car", then "type of cars" or "types of car" or "types of cars" will not be automatically recognized and extracted. The user can either add these variations as terms to this type or include them as synonyms.

Problem: Reserved type name.
Solution: Certain type names are reserved. Do not use "Name" as a type name. If you do so, it will overwrite the Person dictionary in the Core library.

Problem: Terms not included in Extraction results.
solution: Given the complexities and the nuances of natural language, extraction results will never be perfect. You may encounter situations where a term in a given record is not included in the extraction results and you believe it should. Below are some known examples:

1) The extractor attempts to identify complex terms representing specific ideas. Therefore, a multi-word term will be extracted rather than single word terms if they share words. For example, "apple pie" rather than "apple" and "pie". However, you can group terms together during categorization or create synonyms to group the terms together automatically during the extraction process.

2) Following the same logic as the previous point, a longer pattern will be extracted rather than a pattern with a common term, but overall fewer words.

3) In order to reduce the amount of noise returned in an extraction, some patterns may be identified as "null." A null pattern may contain a term that was extracted either as an individual term or as part of a pattern. Users can manually join these responses during the categorization process.

You can fine-tune your extraction results by editing your dictionaries, creating synonyms and creating custom libraries. (See User's Guide for more information). However, it is important to understand that no extraction will be perfect. All analysis will require some manual intervention. In general, when issues such as those described in the preceding bullets are encountered, we recommend you attempt to address them by using one of the different categorization techniques available to you.

Problem: Enter key doesn't activate dialog boxes with active focus.
Solution: In some instances some dialog boxes that are in focus do not respond when the user presses the 'Enter' key. In this case, the mouse should be used.

Problem: Unable to move nodes on Web Graph.
Solution: Currently it is not possible to move the nodes around on a web graph.

Problem: Error when restoring TAfS 3.0.1 Resources.
Solution: Export the templates and import into TAfS 4.0.1.

Problem: No extraction results visible after restoring Resources.
Solution: Don't use Backup; use Save Project instead.

Problem: Error: DMSException: SpssioBufferShort 49 - when importing a string variable with a width of 4000.
Solution: Export the .SAV file into an alternative format and use that as the input.

Problem: "Extend" on opinion concepts creates type descriptors.
Solution: A workaround is to manually remove such descriptors. To obtain text link descriptors with the Extend function, make sure that Type Patterns (rather than Types) are selected in the Build Categories dialog. To obtain concept descriptors with the Extend function, ensure that Types (rather than Type Patterns) are selected in the Build Categories dialog.

Problem: Change data source warning is displayed if there is no column name. This applies to a dataset without a column name and with "Column names in first row" unchecked.
Solution: Column names should be added to the data set.

Problem: ODBC data source won't connect when the source is already defined.
Solution: A workaround is to click the ODBC button to open the ODBC Data Source Administrator dialog; close the dialog and you will see the tables listed.

Problem: Invalid digital signature on installation. From release 4.0.1 onwards TAfS uses IBM-issued certification for digital signing. If you try to install TAfS on a machine that has no Internet connection, and that does not have the correct certificate installed, you see the following message from the installer: "Error 1330. A file that is required cannot be installed because the cabinet file filename has an invalid digital signature."
Solution: Use the following procedure to correct this problem:

1) Click OK to acknowledge the message.

2) Click Cancel to exit from the installer.

3) If the machine on which you want to install has no Internet connection, perform the next step on an Internet-connected machine and copy the .cer file to the machine where you want to install.

4) Follow the instructions at https://knowledge.verisign.com/support/code-signing-support/index?page=content&actp=CROSSLINK&id=AR1553 to download and save the root Certification Authority (CA) as a .cer file.

5) Double-click the .cer file.

6) On the General tab, click "Install Certificate".

7) Follow the instructions in the Certificate Import Wizard, using the default options and clicking Finish at the end.

8) Retry the installation.

TAfS 4.0.1 - Japanese specific issues

Problem: There are three Extraction options shown: "All sentiment", "Representative Sentiment Only", and "Conclusion only".
Solution: For Japanese, the Sentiment options "All sentiment" and "Conclusion only" are not available so "Representative Sentiment Only" will be performed whatever the option selected.

Problem: Some functions not enabled in the edit menu for the Resource Editor view.
Solution: Use context menu for access to disabled file menu functions. Alternatively use dragging & dropping to move to a different Type, or to add to a synonym. Keyboard shortcuts work too for copy/paste for example.

Problem: Find function doesn't work in the Resource Editor view.
Solution: Use Sort function instead.

Problem: Removing types from a group in the Extraction Settings dialog box does not remove the items from the group. The extraction settings are not applied as the items removed from the extraction groups are still included in the extracted results
Solution: Currently there is no resolution for this issue.

Problem: Language Weaver does not work when using TAfS in a Japanese user interface.
Solution: If you change the system locale to English, TAfS will work with Language Weaver.

Related information

Technical Support is available to maintenance customers. Customers may contact Technical Support for assistance in using IBM Corp. products or for installation help for one of the supported hardware environments. To reach Technical Support, see the IBM Corp. web site at http://www.ibm.com/support. We recommend that you check the support site for updates.

© Copyright IBM Corporation 2004, 2011