Links

(a regular column in intelligence magazine)

Syed S. Ali
University of Wisconsin - Milwaukee

Authoring Tools for AI

An authoring tool is software that simplifies difficult or tedious data or document creation. Such software can run the range of text editors (emacs, vi) to software development environments (Microsoft’s Visual Studio). However, a search for authoring tools on the World Wide Web suggests that the majority of authoring tools are HTML editors, courseware creation tools, and multimedia presentation editors. Multimedia tools are tools that are used to create applications that use a variety of media (e.g. music, video, graphics, text, speech, etc...) to communicate their message. Courseware is software used in computer-based training to enhance or replace student-teacher interaction. Document production tools include HTML editors and XML tools which are often used to create document formats that are portable to different platforms or to enhance a document's semantic or structural information.

Authoring tools are important for artificial intelligence because much of the "intelligence" in an application lies in its use of content or data (i.e., its knowledge; see Links column in intelligence 10(2), Summer 1999 or at http://tigger.cs.uwm.edu/~syali/links.html). For example, my current research in building intelligent dialog systems for a tutoring system (see intelligence 10(1) 14-23; or at http://tigger.cs.uwm.edu/~mcroy/Publications/intelligence.pdf) requires the development of rich representations of domain knowledge. While this knowledge can be (and often is) handcrafted, this is a time-consuming process; authoring tools can partially automate the process. I am currently developing an authoring tool for adding semantic markup (using XML) to instructional documents in our domain. As part of this work, members of my research group surveyed available authoring tools; a portion of the survey appears here.

 

Criteria for Choosing an Authoring Tool

Generally, selecting the best authoring tool depends on the task to which it will be applied. For example, one might want to create new content from scratch or to mark-up an existing document so that its content will be usable in an AI system. We have found the following criteria helpful in evaluating an authoring tool’s appropriateness for a task:

• Automatic processing of text is the ability of the software to take a raw text or word-processing document and do some sort of initial markup or processing of the information contained within to capture structural or semantic information.
• Readily usable output is the software's ability to produce output in a non-proprietary format (e.g. ASCII text).
• Tools for document structure processing allow the software to preserve a representation of the original document's structure. Additionally, they may provide some features that allow indication of relationships between document elements as well as definition of ambiguous elements (e.g. ordered vs. unordered lists).
• The ability to create course layout is the ability to specify a course's structure and curriculum by the user in the course design process. This often includes a graphical representation of the course structure (as a tree, graph, etc.).
• Ability to import course layout is the ability to import standard designs (a training course for example) which has been created using other software and use the design in structuring content.
• Inclusion of document-structure relationships refers to the specific ability of the software to indicate relationships between various elements of the document (section-subsection, between paragraphs, etc.).
• The ability to insert semantic data refers to the software's ability to include semantic information about text elements while processing a document. For the most part, this refers to the insertion of semantic tags into the source text as defined by the user.
• Insertion of tags into output text is the ability of the tool to embed information gathered during authoring by tagging the text (using XML for example) with this information.
• The ability to perform queries, link elements, etc. is the tool’s ability to do some special processing of content beyond the initial design process. This might include the ability to take some sort of tagged output text and link similar elements or perform queries on added semantic data.
• Tools for insertion of test questions are tools that are provided by the software for authoring test questions and inserting them in specific areas of the designed course.
• A user-adaptable or extendable program allows the user to add functionality to the authoring tool.
• The ability to create glossary files is a feature that is provided by many of the courseware packages which allows the addition of terms and their definitions.
• Automatic index generation takes all of the entered content and creates an index that points to each term.
• Ease of use is a subjective assessment based upon how difficult software is to use and the quality of support and documentation.
• Demonstration/evaluation copies should be available.

After I evaluated many authoring tools on these criteria, I decided that none was entirely appropriate to our task of constructing the content for our intelligent dialog system. Basic research often encounters this problem; your luck may be better.

Where to Find Authoring Tools

Authoring tools are popular; as a consequence there is a large number of systems that are available. Here are some links to a variety of authoring tools; hopefully, one or more of them will be useful to your task.

• ADEPT8: http://www.arbortext.com/Products/ADEPT_Series/ADEPT8/adept8.html
• Authorware Attain 5: http://www.macromedia.com/software/authorware/
• Blackboard: http://www.blackboard.net/
• CyberProf: http://loner.ccsr.uiuc.edu/index.html
• Designer's Edge: http://www.allencomm.com/software/designer/
• Digital Chisel: http://www.digitalchisel.com
• Digital Trainer: http://www.micromedium.com/
• Dreamweaver Attain: http://www.macromedia.com/software/dreamweaverattain/
• Epic: http://www.arbortext.com
• IconAuthor: http://www.asymetrix.com/products/iconauthor/
• Instructional Toolkit: http://toolkit.virginia.edu
• Knowledge Annotator: http://www.cs.umd.edu/projects/plus/SHOE/KnowledgeAnnotator.html
• Lotus LearningSpace: http://www.lotus.com/home.nsf/welcome/learnspace
• LT XML: http://www.ltg.ed.ac.uk/software/XML/
• MetaCard: http://www.metacard.com
• Quest: http://www.allencomm.com/software/quest
• SHOE: http://www.cs.umd.edu/projects/plus/SHOE/
• Toolbook II Assistant: http://www.asymetrix.com/products/toolbook2/assistant/
• Toolbook II Instructor: http://www.asymetrix.com/products/toolbook2/instructor/
• Video Test: http://www.frax.com/Pages/FSVS340.htm
• VisualBridge Edition: http://www.visualbridge.com/
• Web Course in a Box: http://www.madduck.com/wcbinfo/wcb.html
• Web CT: http://homebrew.cs.ubc.ca/webct/
• XED: http://www.cogsci.ed.ac.uk/~ht/xed.html
• XML Pro: http://www.vervet.com/
• Xpose: http://www.intravenous.com/

 

Writing Next Issue's Column

The next issue of Links will be on Reasoning About Function. Function is a source of device knowledge that relates behavior to human notions of utility. For example, the behavior of an electric switch is to open or close a circuit, while its function is to turn a light on or off. Function is an important source of knowledge in AI applications such as diagnosis, design, and explanation generation.

If you have any comments on this column (or suggestions for topics for future ones) or interesting resources, please email me at: syali@usa.net. This article (and updates to it) is available online at http://tigger.cs.uwm.edu/~syali/links.html/ with links to all the resources mentioned.

Acknowledgements

Thanks to Krista Bennett for her help compiling the list of authoring tools. This work was supported by the National Science Foundation, under grants IRI-9701617 and IRI-9523666 and by a gift from Intel Corporation.