* Using textract databases - a search guide

A note on database structure

 Textract databases are full-text-retrieval databases - every word or number in the database can be used as a search term. The database is made up of 'records' which are the smallest complete units of information - for example, in a company directory database, each company will be a separate record; in a newspaper database, each article will be a separate record. Each record may be further subdivided into 'fields' - this will vary from database to database except for a heading/title field, which all databases will have. Searches can be run in specific fields, as well as across the whole database.

How searches work

The database software will take your search query and match it against every record in the database in turn. It will then tell you how many records it has found that contain your search query. For example, if your search query is the word apples it will check every database record to see if that word is in each one. The number of 'hits' it returns to you is the total of all the records in the database containing the word apples. This interpretation is literal - if you spell a word differently from the way it is spelt in the database, your search will fail.

Simple search terms - words and phrases

Search terms can be single words - made up of characters or numbers - or short phrases. For example:

     apple       
     banana
     fruit salad

'Wildcards' can be used to deal with plurals, and multiple word endings:

     ?       can replace a single character
     *       can represent one or more characters
     [a-z]   can represent a range of characters

For example:

     disab*    will get all the records containing the words
               'disabled', 'disability', 'disabling'
     labo*r    will get all the records containing 'labour' and
               'labor'
     199[3-5]  will get all the records containing '1993', '1994'
               and '1995'

Note: '*' in particularly needs to be used carefully - a search for architecturally oriented information, using arch* as a search term , will also find records on archdukes, archery and archangels as well.

Searching with numbers

Whole numbers can be searched using 'relational operators' in order to do such tasks as select out the latest or the oldest information, or to select a particular range of business classification codes. For example:

     >1994     gets all records containing a date greater than
               1994
     <1995     gets all records containing a date less than 1995 
     <=1992    gets all records containing a date up to and
               including 1992
     >=1003    gets all records containing a date after 1002

Numeric ranges can be searched as follows:

     [1990 1995]   gets all records for the years 1990, 1991,
                   1992, 1993, 1994, 1995

Constructing search queries by combining search terms

Search words, numbers and phrases can be combined to make more complex, and hopefully more specific search queries, using the concepts of and, or and not:

     &     AND
     |     OR
     #     EXCLUSIVE OR
     !     NOT

Examples are:

     apples & pears
     grapes | mango
     pineapple # sardines
     housing ! London

Brackets should be used to avoid confusion, where search terms contain several of these combined queries:

     (apples & pears) | (oranges & grapes)

Search order can be forced, using either ':' or ';' at the end of a search term, for example

     (cats & dogs);     gets the words in whichever order they
                        are written 
                        (this normally happens, as default)    
     (rats & toads):    gets them only if they appear in the
                        record in the order as written in the  
                        search term

Proximity - how close the search words appear to each other in a record - can be enforced using the '%' character, as follows:

     (ministers & sleaze);%3     gets the records where the
                                 search words appear in any order 
                                 but within three words of each 
                                 other

Note: if you want to use any of these characters - '&' '|' '!' '#' '%' ';' ':' - as part of your search word, than you must precede them with '\' for example:

     housing \& manchester     finds the actual phrase 'housing
                               & manchester'

Field searching - searching in part of the database only - check the help text

To run a search in a specific field, add the fieldname preceded by '$' onto the end of each search word or phrase:

     stockport$title     will run the search on the term
                         'stockport' in the title field of each 
                         record only

For information and advice on how best to use field searching, check the support information for each specific database.