phrasewordproc

gazim · Post by **gazim** » Thu Sep 23, 2004 5:03 pm

phrasewordproc in "all" mode doesn't seem to be working for me. My query looks something like "select author from mytable where author like '"bear arm"'". Its returning 0 hit.

Data in mytable.author field :

author
------------+
bearing potatoe arms
bearing potatoe arm
bear arms
bear arm
bearing arms
beared arms

Apicp settings:
---------------

<apicp allinear "on">
<apicp alpostproc "on">
<apicp minwordlen 3>
<apicp exactphrase "on">
<apicp suffixproc "on">
<$suffixlist="'" "ed" "'s" "" "er" "ual" "ing" "s">
<apicp suffix $suffixlist>
<apicp phrasewordproc "all">

index Expressions:
------------------
set delexp=0;
set addexp='[\alnum\punct]{1,30}';
set addexp='\punct{1,5}';
set keepnoise='on';
set addexp='\alnum{1,99}';
set addexp='>>\alpha{1,50},=\alpha{1,50}';
create metamorph inverted index idxmsuff_auth on mytable(author);"

texis version info: Commercial Version 5.00.1086121238 20040601 (i686-unknown-linux2.4.9-64-32)

Thanks in advance.

Post by **mark** » Fri Sep 24, 2004 11:22 am

Are you getting any errors from that query (check the html source of the results page)? The combination of quotes you have won't work. Try
<$q='"bear arm"'>
<sql "select ... like $q">

Post by **John** » Fri Sep 24, 2004 11:55 am

The phrasewordproc all mode was documented before being implemented. It is not currently supported by the software.

gazim · Post by **gazim** » Fri Sep 24, 2004 12:29 pm

I tried the correct combination of quotes. There is no error message. I just get 0 hit. If i modify my query to <$q='"bear* arm"'>, it returns,
[bearing potatoe arms,bearing potatoe arm,bear arms,bear arm,bearing arms,beared arms.]
With my initial query, it should've returned at least 1 hit since I have exactphrase set to "on". With my second example I got hit becuase of the wildcard. Shouldn't <$q='"bear arm"'> return the same number of hits as the wildcard query since I have phrasewordproc set to "all" with minwordlen=3?

gazim · Post by **gazim** » Fri Sep 24, 2004 12:36 pm

Thank you for the clarification John. Is there a plan to implement this feature in future releases?

Post by **John** » Mon Sep 27, 2004 11:47 am

Yes, which is how it ended up in the documentation.

gazim · Post by **gazim** » Mon Sep 27, 2004 12:53 pm

Thanks John. We were kind of looking for this enhancement. Wildcard after one of the words in a phrase search always spans upto 80 characters. For example, search term "claim* superintendent manual" wil find 'claim department superintendent manual' or 'claims departmental office superintendent manual'. However, we want this search to locate 'claims superintendent manual' or 'claiming superintendent manual'. Phrasewordproc in 'all' mode, if implemented, will give us a reasonable workaround. Is there another way to produce these hits with the current texis version? Thanks in advance.