The searchtotal table
Take a look at what the search_total table contains after indexing example nodes 2 and 3:
raysql> select * from search_total order by count desc;
|
| word 1 count j____________i___________ | |
|
are |
0.30103 |
|
the |
0.30103 |
|
new |
0.30103 |
|
and |
0.162727 |
|
talk |
0.162727 |
|
ponies |
0.118099 |
|
badgers |
0.0649408 |
|
testing |
0.0163904 |
|
links |
0.0163904 |
|
taxonomy |
0.0163904 |
Figure 20: Global normalized totals for the words in the index.
After each time that indexing occurs, all of the words that have been marked by search_dirty will be updated in the search_total table. The count value is a normalization according to Zipf's law that says a word's value to the search index is inversely proportionate to its overall frequency therein. Here is the count calculation expressed in code:
Post a comment