The example used was the Dyna Text system. This short piece showed the different ways that could be used, including wildcards (*,?), to assist in searching textual data. (Data in the form of words, strings, paragraphs etc.)

Dyna Text demonstrated the use of boolean logic to assist in searches.

Boolean searches can be extended using proximity searches using phrases such as: after, before, of, near, within, word, words e.g.

• jack within 2 words before jill
• jill within 2 words after jack
• jack fell down within 6 words before jill came tumbling
• jack within 8 words of jill

Dyna Text also demonstrated the use of SGML (Standard Generalised Markup Language) to do a context search in different type of textual data - e.g. titles. This is similar to the use of tags in HTML documents where elements of textual data are tagged into different types of data with HTML tags.

Using different methods or a combination of methods, it is possible to formulate search queries to assist in searching textual data.

Dyna Text does not read the documents. The occurrences of the textual data are recorded into an Index. It is this indexing that makes it possible to search for the data.

Extremely common words such as: - the, and are not indexed. Indexing and searching for these words would be costly in processor time. It is also highly unlikely, IMHO, that a user would wish to search for such words.

Information sought is rarely the same, nor does it use the same queries. Effective searching requires iteration. The results obtained, and the methods used, will vary from person to person dependent upon their individual outlook, experience etc.

One problem with searching a document information system is that the results returned will include all instances of the word/phrase looked for. In a large system this could possibly run into hundreds of results.

Comments, suggestions, ideas to
Stuart Banner