I have been reading a paper: Wang X, Golbandi N, Bendersky M, Metzler D, Najork M. Position bias est

Alisa Durham

Alisa Durham

Answered question

2022-05-12

I have been reading a paper: Wang X, Golbandi N, Bendersky M, Metzler D, Najork M. Position bias estimation for unbiased learning to rank in personal search.
I am wondering how they derive the M-step in an EM algorithm that has been applied to the problem in their paper.
This is the data log-likelihood function:
log P ( L ) = ( c , q , d , k ) L c log θ k γ q , d + ( 1 c ) log ( 1 θ k γ q , d ) .
where observed data are (c:click, q:query, d:document, k:result position) rows, and θ k is the probability a search result at rank k is examined by the user (assumed to be independent of query-document pair), and γ q , d is the probability a search result, i.e., a query-document pair is actually relevant (assumed to be independent of result position).
Here are some hidden variable probability expressions (E means examination and R means relevant, they assume that a click event implies a search result is both examined and relevant):
P ( E = 1 , R = 1 C = 1 , q , d , k ) = 1 P ( E = 1 , R = 0 C = 0 , q , d , k ) = θ k ( t ) ( 1 γ q , d ( t ) ) 1 θ k ( t ) γ q , d ( t ) P ( E = 0 , R = 1 C = 0 , q , d , k ) = ( 1 θ k ( t ) ) γ q , d ( t ) 1 θ k ( t ) γ q , d ( t ) P ( E = 0 , R = 0 C = 0 , q , d , k ) = ( 1 θ k ( t ) ) ( 1 γ q , d ( t ) ) 1 θ k ( t ) γ q , d ( t )
The paper has derived the M-step update formulas:
θ k ( t + 1 ) = c , q , d , k I k = k ( c + ( 1 c ) P ( E = 1 c , q , d , k ) ) c , q , d , k I k = k γ q , d ( t + 1 ) = c , q , d , k I q = q , d = d ( c + ( 1 c ) P ( R = 1 c , q , d , k ) ) c , q , d , k , I q = q , d = d
however, how to get these two formulas?
Note: M-step is usually derived from this formula:
Q ( θ θ ( t ) ) = E Z X , θ ( t ) [ log L ( θ ; X , Z ) ]
θ ( t + 1 ) = a r g m a x θ   Q ( θ θ ( t ) )
where θ are parameters (i.e., θ k and γ q , d in this case), X are data (c,q,d,k in this case), and Z are hidden variables (E and R?).

Answer & Explanation

Emmy Sparks

Emmy Sparks

Beginner2022-05-13Added 17 answers

Actually, I have found how to reach those updated formulas, although the final form looks a bit different, I am happy to share here in case anyone can find out why.
First, write Q function in more separated cases:
Q ( θ k , γ q , k θ k ( t ) , γ q , k ( t ) ) = X = c , q , d , k [ P ( E = 1 , R = 1 C = 1 , q , d , k ) c log ( θ k γ q , d ) + P ( E = 1 , R = 0 C = 0 , q , d , k ) ( 1 c ) log ( θ k ( 1 γ q , d ) ) + P ( E = 0 , R = 1 C = 0 , q , d , k ) ( 1 c ) log ( ( 1 θ k ) γ q , d ) + P ( E = 0 , R = 0 C = 0 , q , d , k ) ( 1 c ) log ( ( 1 θ k ) ( 1 γ q , d ) ) ]
Then simply take the derivatives:
0 = θ k Q ( θ k , γ q , k θ k , γ q , k )
The final theta update rule looks like
θ k ( t + 1 ) = c + P ( E = 1 , R = 0 C = 0 , q , d , k ) ( 1 c )

Do you have a similar question?

Recalculate according to your conditions!

Ask your question.
Get an expert answer.

Let our experts help you. Answer in as fast as 15 minutes.

Didn't find what you were looking for?