mongodb - Semantic category prediction of hierarchal data -

mongodb - Semantic category prediction of hierarchal data -

i'm building web app services businesses of few industries. want build search box asks

what type of work do?

|____search box_____|

the user can come in text, , based on keywords, want suggest categories. ideally, want homecoming 5 results associated confidence levels , allow them take via checkboxes.

for example, if

keyword: scrubbing

results:

0: [house cleaning, 209] 1: [end of lease cleaning, 589] 2: [office cleaning, 210] 3: [outdoor cleaning, 202] 4: [general gardening, 161]

or if keyword: auto wheel

results:

0: [car spray painting, 533] 1: [panel beating, 104] 2: [car cleaning & detailing, 96] 3: [car mechanics, 101] 4: [automatic doors & gates, 185]

where number on right confidence.

notice using synonyms keywords. given have categories in mongodb document structured so:

{ 'cleaning' : { 'outdoor cleaning' : [ 'outdoor', 'yard', 'cleaning', 'field' ], 'office cleaning' : ['indoor', 'office', 'business', 'corporate', cleaning', 'scrubbing'] }, 'auto': { 'mechanic' : ['car', '...'] } }

how can homecoming desired result? trained naive bayes classifier classifies 1 category. looking term frequency–inverse document frequency (tf-idf) set of ranked results little worried info model not fitting algorithm.

will have include every possible synonym train categories? should wait till it's live manually train categories based on real keyword searches?

i can still alter info model if knows of improve way construction information... if has experience problem such this, insight much appreciated.

thank you

mongodb search machine-learning classification