Archivematica

Yeah, mentioning to users that they can do wildcard searches might be good rather than changing the tokenizer.

If a user enters "*master*" or "Members_Master2009.xls" now they'd end up seeing the entry for Members_Master2009.xls. If we change the tokenizer so it splits words up by underscores, etc., we might lose the ability to search for "Members_Master2009.xls" precisely.

I've posted a question online to see if anyone has any tokenizer advice on how we could tokenize "Members_Master_2009.xls" into "members", "master", and "Members_Master_2009.xml".

Updating analysis then testing...

curl -XPOST 'http://192.168.1.70:9200/aips/_close'
curl -XPUT 'http://192.168.1.70:9200/aips/_settings' -d '{
"index": {
"analysis" : {
"analyzer": {
"default": {
"tokenizer" : "standard",
"filter" : ["preserve_hyphens_filter", "lowercase", "stop"]
},
"filter" : {
"preserve_hyphens_filter" : {
"type" : "word_delimiter",
"generate_word_parts": false,
"catenate_words": true
}
}
}
}
}
}'
curl -XPOST 'http://192.168.1.70:9200/aips/_open'
curl -XGET '192.168.1.70:9200/aips/_analyze?field=msg&pretty=1' -d "Run to the hills and rock-around-town."