this query isn't efficient - ruby-on-rails

Question

i have a single search field that is querying three different columns from two different tables: "companies" and "industries" from a positions table and "schools" from an educations table. it is successfully returning all users that meet ALL fields entered into the search field (using select_tag). this is from my view:
<%= form_tag(search_path, :method => :get, :id => "people_search") do %>
<div class="row-fluid">
<div class="span4">
<table>
<tr>
<td>
<div class="search-table" style="padding-left:55px">
<%= select_tag "all", options_for_select((#visible_companies + #visible_industries + #visible_schools).uniq, params[:all]), { :placeholder => "Search by companies, industries and schools...", :multiple => true, :js => "if (evt.keyCode == 13) {form.submit();}" } %>
</div>
</td>
<td>
<%= submit_tag "Add", id: "send-button", style:"width:175px;" %>
</td>
</tr>
</table>
</div>
<% end %>
</div>
and controller:
#visible_positions = Position.where{ is_visible.eq('true') }
#visible_educations = Education.where{ is_visible.eq('true') }
#visible_companies = #visible_positions.order("LOWER(company)").map(&:company).uniq
#visible_industries = #visible_positions.order("LOWER(industry)").map(&:industry).uniq
#visible_schools = #visible_educations.order("LOWER(school)").map(&:school).uniq
#c = #visible_positions.where{company.in(my{params[:all]})}.map(&:user_id)
#i = #visible_positions.where{industry.in(my{params[:all]})}.map(&:user_id)
#s = #visible_educations.where{school.in(my{params[:all]})}.map(&:user_id)
#C = #visible_positions.where{company.in(my{params[:all]})}.map(&:company)
#I = #visible_positions.where{industry.in(my{params[:all]})}.map(&:industry)
#S = #visible_educations.where{school.in(my{params[:all]})}.map(&:school)
#blacklist = []
#cis = #c + #i + #s
#experiences = ([#C,#I,#S].reject(&:empty?).reduce(:&))
#cis.uniq.each do |user_id|
unless #C.empty?
#C.uniq.each do |company|
unless Position.find_all_by_company(company).map(&:user_id).include?(user_id) || Position.find_all_by_industry(company).map(&:user_id).include?(user_id) || Education.find_all_by_school(company).map(&:user_id).include?(user_id)
#blacklist << user_id
end
end
end
unless #I.empty?
#I.uniq.each do |industry|
unless Position.find_all_by_industry(industry).map(&:user_id).include?(user_id) || Position.find_all_by_company(industry).map(&:user_id).include?(user_id) || Education.find_all_by_school(industry).map(&:user_id).include?(user_id)
#blacklist << user_id
end
end
end
unless #S.empty?
#S.each do |school|
unless Education.find_all_by_school(school).map(&:user_id).include?(user_id) || Position.find_all_by_company(school).map(&:user_id).include?(user_id) || Position.find_all_by_industry(school).map(&:user_id).include?(user_id)
#blacklist << user_id
end
end
end
end
unless #c.empty? && #i.empty? && #s.empty?
#users = User.find(#cis - #blacklist)
end
the search looks like this (notice the single field), with a sample query included (notice the AND filter...i'm the only user in the database that fits all search terms ['dartmouth college' for school, 'world health organization' for company, 'internet' for industry]):
i realize this is not an efficient query and am thinking of ways to speed it up, but could use some ideas at this point.
happy turkey day :)

Answer 1

Based on your description rather then on understanding your code I figured out something like this
User.joins(:positions, :educations).where("lower(positions.company) like lower(?) and lower(positions.industry) like lower(?) and lower(educations.school) like lower(?) and positions.is_visible and educations.is_visible", "%#{company}%", "%#{industry}%", "%#{school}%")
or if there is only one company or industry in column
User.joins(:positions, :educations).where("(lower(positions.company) = lower(?) or lower(positions.industry) = lower(?)) and lower(educations.school) = lower(?) and positions.is_visible and educations.is_visible", company,industry, school)
But to put many industries, companies, schools as params will be more complicated
and create indexes
create index positions_lower_company on positions (lower(company));
create index positions_lower_industry on positions (lower(industry));
create index educations_lower_school on educations (lower(school));
I hope it will help somehow.