YourSoft írta:
Hi,
I have a suggestion to improve nutch search results.
The "big" search engines (like google) measure the distance between
the query words.
E.g.:
query string: lucene in action
When you search for it in google, google will boost up that documents
where the "lucene in action" is in the same sequence.
I think it is possible in nutch/lucene (e.g. if your search string is:
"lucene in action"), but nutch don't make it.
Any ideas how to make it?
Regards,
Ferenc
I'm sorry something is missing from previous mail:
When search the keywords, there something also improve the boost:
- How many times found the full query ('Lucene in action") in document.
(The length of total document / full query count - if it is bigger than
10-20% it is BAD)
- How many times found the query words in document ("lucene" "in" "action")
