1. It ll have n+1 sheets ..
2. Like a 3d periodic table for methods
3. every step begins with a new row every block of which in a diff column will contain a symbol/operator/number etc.
4. first n sheets will contain n methods to solve problem x,with n*m(i) error functions .[m(i) number of steps in ith)
5. n sheets will be connected semantically and contextually optimized
6. combinations of steps.
7. on n+1 th sheet user feeds his problem ,
8. If the context is matched,he is asked to feed his method steps to
solve the problem on the RHS of the n+1 th sheet he gets validity/proof
of his steps and recommended steps for optimized error in terms of
contextual tolerance
9. #example spreadsheet
https://spreadsheets.google.com/...
assume that recommender is added to this sheet's back-end
Look at this sheet
now if i want to make a new sheet to strategize say my academics
and start feeding parameters to this new sheet
[switch to listview of spreadsheet]
when i click a drop down in list view i ll see the recommended data formula to feed with it's row*column location.
10. I find it related to structured prediction
11. #use case
12. think of toy design or drug formulation companies.
##They have vast unorganized history{across past aand organization} of methododologies and results ##
which can be converted to methodbase ,
just as documents s are converted to knowledge base using machine learning semantics .
this method-base is comparable to the old sheets on the example spreadsheet i messaged you earlier......
The
charles sheet on the spreadhseet was comparable to GUI of the product
where the drug formulator /toy designer feeds his objective and blocks
of procedure( steps of solution ,methods, images etc }
this gets compared with what is recorded on old sheets[method-base]
and
on the RHS of Charles sheet recommended steps /methods appear such that
noise/signal ,learning curve,resource stress ,risk and error count is
mitgiated .
http://www.quora.com/Recommendation-Engines/I-am-integrating...
--
[email protected]