Tips for Using the Machine Aided Indexing Application

Adding and Selecting Input Text

• The MAI text processor will accept any character-based text that you copy&paste (or type) into the workscreen--from any source, including simple-text and word-processing documents, text from web pages, and character-based text copied from PDF documents.

• How much text can be input? The application is designed to handle at least an average-sized full-text report or article. A limitation may be imposed by your web browser (e.g., a 30,000 character limit was noticed for some older versions of Netscape). It is recommended that you use the latest available version of your browser.

• How little text can be input? Good results will be obtained from short summaries and abstracts. In most cases, entering simply a word, phrase, or short sentence will not provide sufficient information to get good results.

• Texts from different subject areas. The MAI knowledge base uses the NASA Thesaurus as its subject domain, which has a very broad coverage in all of the natural and applied sciences, engineering, and many specialized technologies. MAI works best in these areas, but keep in mind that good results can be obtained for some non-technical topics as well.

• In some cases, better quality output will result if you omit those parts of a document that do not address the theme of your document (such as a preface or acknowledgements section).

• Processing more than one abstract or short document at a time. Processing multiple abstracts, or multiple brief documents or web pages at once can be helpful when you want to get a quicklook overview of concepts addressed by those items as a collection. The output will also provide a frequency ranking for each concept term.

Selecting Terms

The MAI application was designed as a machine-aided rather than a fully automatic indexing tool. For this reason it is recommended that you review the list of output terms and select those you judge to be appropriate for a given document. Additional terms can also be identified by browsing the NASA Thesaurus using the thesaurus search function.

Output terms are ranked according to the frequency with which each particular term was suggested from text expressions in a document. Particularly in cases where a full-text document was input, this ranking can be helpful in reviewing the term list and making term selections.

Additional information:

• Instructions
• What is Machine Aided Indexing?
• Why Use Machine Aided Indexing?