Course talk:CPSC522/WeightedModelCounting

Looks good! Some comments on the page:

Define $\lambda _{x_{i}}$ notation before you use it. And you define $\lambda _{\theta }$ but do not use it.

Tell us what an indicator variable is.

How does ENC1 and ENC2 differ from the encoding already given? They are presented independently, and the reader doesn't know how they fit together.

Explain what "local structure" and "determinism" mean (when they are first introduced). Don't assume we know.

There are some problems with missing spaces that can be fixed with a careful proofread. And a few other cases where the grammar isn't quite right. (I didn't read the "To Add" section as this was meant for your rough notes.)

-$\lambda _{\theta }$ was a mistake (it is $\theta _{x|u}$)

-I considered this as a definition: "Where the $\lambda$ is an indicator variable representing assignment of a value to a variable"

-I added a small section for similarities and differences though there is more differences in the end of ENC2.

-Added local structure definition to the abstract. For determinism, I mentioned that it is what follows.

Some comments here:

> Main article: Bayesian Networks

Dead link

> Followed by the chain rule and using independences we have,

Probability page doesn't introduce the chain rule.

> The following is an example encoding of this probabilistic model as a CNF.

The latex expressions are hard to read. Put some space between expressions?

> Weight of the positive literal λ θ x | u {\displaystyle \lambda _{\theta _{x|u}}} {\displaystyle \lambda _{\theta _{x|u}}} is P r θ x | u {\displaystyle Pr_{\theta _{x|u}}} {\displaystyle Pr_{\theta _{x|u}}}

I don't see this literal appear anywhere. I think you meant \theta _{x|u}

> The weights for the literals define a weight for each model w {\displaystyle w} {\displaystyle w} as follows.

What is a model?

> For each parameter P r ( x i | u 1 , u 2 , . . . , u n ) {\displaystyle Pr(x_{i}|u_{1},u_{2},...,u_{n})} {\displaystyle Pr(x_{i}|u_{1},u_{2},...,u_{n})}we generate the following clauses,

What are the variables u_1, ..., u_n? (for ENC1 and ENC2)

> Main article: Knowledge Compilation

Dead link.

> Model counting and arithmetic circuit

"Smooth d-DNNF" and "Model counting and arithmetc circuit" need some plain explanations.

> The irrelevant variables are defined as the variables that their value does not affect the probability (that is uniform over literals).

the variables that their value => the variable whose values. What probability?

> Decomposability

I think the scope gets too large all of a sudden here. ENC1 and ENC2 were pretty coherent.

"Dead link."

The template or macro

link format was not compatible with the pages,

It seems that it is fixed, I added "Course:CPSC522/" and it fixed the problem but it still doesn't look good.

"Probability page doesn't introduce the chain rule."

I think I mistook with the Wikipedia page, Bayesian_Networks also uses the same link for chain-rule, I linked it to Bayesian_Networks for now.

"The latex expressions are hard to read. Put some space between expressions?"

I centralized most of the main formulas for more readability and added \qquad before some of the clauses.

"I don't see this literal appear anywhere. I think you meant \theta _{x|u}"

That was a mistake changed to \theta_x|u and Pr(x|u). (This instance corrected)

"What is a model?"

Unfortunately, there are two different concepts. A probabilistic model(e.g. similar to graphical models) and an assignment to the variables both are called a model. Both are mentioned in the initial paragraph. I added another sentence for further clarification.

"What are the variables u_1, ..., u_n? (for ENC1 and ENC2)"

I think this was Ok, but I added a sentence in ENC1.

"'Smooth d-DNNF' and 'Model counting and arithmetic circuit' need some plain explanations."

There is a main page which is referenced. I just briefly explained d-DNNF and I don't know much more about arithmetic circuits.

"the variables that their value => the variable whose values. What probability?"

whose value sounds right,(that is (added: their weight is) uniform over literals)

"I think the scope gets too large all of a sudden here. ENC1 and ENC2 were pretty coherent."

I think taking advantage of these structures were important. Though it was beyond the available time. added:(that is the factor value in the CPT is uniform over literals of that variable)

This is a complex topic, and I think that overall it is well presented.

In connection to the other reviewer's remark, does "determinism" in this context have something to do with the fact that logic is used?

Although it is not part of the entry itself, I am little confused by the references to apparently incomplete section (4.1) and tables (16 and 17) in the "To Add" part. Maybe it is just wording, but the conclusion in that section sounds as if you have been arguing for a certain position, rather than just describing basics. Anyways, these all concern the "To Add" part, so that it is not very relevant. I am sure that if you had had sufficient time, you could have incorporated the material nicely.