Statistics of what variants of a meta word exist should be generatable
administrator2013-01-04 14:05
Right now if one has a meta words like os:linux, os:windows, etc one can
find all the pages where the machine announced the operating system was
linux or windows. From this one can get a count of pages for that os as well.
It is much harder in yioop to do queries like the top 10 operating systems
in terms of pages without knowing the operating system in advance.
As of version 0.92 Yioop has all variants of most meta words. For example,
os:all gives all pages where the operating system was given by the host.
Iterating over os:all, one can see what operating system is returned by
the first result. Say it is windows. Then one can
query
"os:all -os:windows"
to look for other kinds of os tags. Say we find linux. We could then query.
"os:all -os:windows -os:linux" and so on.
This issue proposes that for every meta word we accumulate such statistics when we click on the statistics links and that we do it in this fashion rather than the ad hoc fashion that the statistics are now being calculated.
Right now if one has a meta words like os:linux, os:windows, etc one can find all the pages where the machine announced the operating system was linux or windows. From this one can get a count of pages for that os as well. It is much harder in yioop to do queries like the top 10 operating systems in terms of pages without knowing the operating system in advance.
As of version 0.92 Yioop has all variants of most meta words. For example, os:all gives all pages where the operating system was given by the host.
Iterating over os:all, one can see what operating system is returned by the first result. Say it is windows. Then one can query "os:all -os:windows" to look for other kinds of os tags. Say we find linux. We could then query. "os:all -os:windows -os:linux" and so on.
This issue proposes that for every meta word we accumulate such statistics when we click on the statistics links and that we do it in this fashion rather than the ad hoc fashion that the statistics are now being calculated.