Editing Repository configuration (section)

==ArchiveTextIndexingConfig.pm==
This module you probably won't need to change unless you want to modify how eprints does searches for words in strings.

When a record is added to the system eprints uses this module to turn a string into a list of values which are indexed. By default these are words with 3 letters or more except some predefined stop words. It also turns latin characters with acutes into the their plain ascii (no acute/grave) versions.

It then does the same with the search string and looks for these keys.

Example:


<code>
 The rain in spain falls mainly on the plains.
</code>
Is turned (by default) into the keys:


<code>
 rain spain fall mainly plain
</code>
Thus searching for "rain" or "plain" or "plains" or "MaiNlY" will all match this string.

You may wish to add your own "stop words". eg. If you are running an archive about badgers, a search for the word "badger" will return almost all the records.

At a more complex level you may wish to add handling for non-european character sets (I have no idea how well the default setting will work on these), or do "stemming" - removing "ed", "ing", "ies", "s" etc. from the end of words so that "land" will match "land", "landed", "landing" and "lands". (It current removes 's').

Another suggestion is using ''soundex'' or similar techniques to match words which sound similar.

Changing the indexing on a live system will require you to regenerate the indexes using the reindex script. (If you don't then some of the search results will be wrong).