Fulltext search on client portals

Full-text query write rules

The way of searching in the AToM editorial environment may differ from the way of searching on end-client portals using AToM data.

Full-text search on a general client portal distinguishes between two basic methods of constructing the search text.

1. Word group search - words are not delimited by quotation marks
2. Phrase search - words are delimited by quotation marks

Both methods can be combined in one query. The text of the search field may contain a part of the text that is delimited by quotation marks, in which case this part (or parts) is processed as a phrase, and the rest of the text is processed as a group of words. There may be more than one quoted part of the text.

The words of the search text may be separated by spaces or other characters (period, comma, question mark, exclamation mark, colon, semicolon and slashes).

Only those attributes that are marked for inclusion in the full-text index are searched. These are primarily name attributes (basename*, alias), text attributes (text), and possibly other attributes.

The search is performed by the CONTAINS method of MS SQL Server with the determination of the weight of the found result, which indicates how much the found result meets the selection criteria. For name attributes, the resulting weight is multiplied and thus strengthens the weight of the result for records where a fragment of the search text is directly in the name attribute.

Word group search

If the text of the search field is not delimited by quotation marks, results are retrieved that contain all words in the text group (AND operator); words that have at least three letters, excluding numbers, may have been previously edited, and the search compares only the beginnings of the individual words corresponding to the length of the search word after editing (an asterisk * is appended to the end of the word).

Word editing consists of removing some punctuation marks at the end of words and some word endings involving, for example, certain endings or consonants.

Word editing, i.e. the specific selection of endings and consonants, is dependent on the specific implementation on the client portal.

Single-letter words, except for numbers, are excluded from the search text.

Searching for a phrase

A phrase is a group of words delimited by quotation marks for which the order of the words matters. If the text of the search box is delimited by quotation marks, a phrase search is applied and results are found that contain all the words of the phrase (AND operator) in the text side by side in the order specified. No modification or elimination of words is performed.

If there is a "*" (asterisk) at the end of a phrase, the words in the phrase are matched against the beginning of the search words ("beginnings of words in phrase*"). Placing an asterisk anywhere other than at the end of a phrase has no effect for matching against the beginnings of words in a phrase.

Examples:

For [aion], the same result is obtained as for ["aion"], plus other records that contain, for example, the abbreviation "AIONAV" in the alias attribute.

For [aion maps] the same search result is obtained as for [maps aion].

For [top a map], the same search result is obtained as for [topic maps], plus additional records that contain e.g. "map topic", or "topology in charts or maps", or even "mapping top ten".

For [topic maps], the same search result is obtained as for ["topic maps"], plus additional records that contain e.g. "topic item" and "mind maps", or even "topik tests" and "maps of smokestacks"

For ["topic maps"] a different search result is obtained than for ["maps topic"].

For ["top map*"] the same result is obtained as for ["topic maps"], plus other records that contain e.g. "topik maps".

For ["aion topic maps"], different records are obtained than for ["aion maps"], if anything is found for ["aion maps"].

For ["aion topic maps"] the same records are obtained as for [aion maps], but for [aion maps] the result may be much more extensive in number of records.