Suggestions
Hi Yan,
I have read your work, it is every organized, clear and easy to understand. But I found that there are some extra information that related to this topic that I suggest you to expand in this page:
1. How to make the decision networks, or how to solve the sequential decision problems, I think most readers are more interested in how to make it or how to use it to solve problem other than just to know what it is. (e.x. add decision variables first or utility first)
2. By drawing decision networks, how it helps to make decisions?
Thanks for the crisp introduction to decision network. I like the way you gave the precise definitions and examples. However the content is little too short to the extent decision network should be discussed. Here are few suggestions that you might like to add:
- You have talked about the nodes in your definition section. However no insight is provided corresponding to the edges. How edges connect nodes, and what are other properties in terms of the entire network must be met, can be discussed (like presence of cycle etc).
- Clear illustration how decision network is learnt. As you must know identifying nodes and their roles is not a trivial task. Hence how this is systemically addressed, the corresponding method can be illustrated
- In the application section you just named a few fields. How those fields actually use decision nets can be elaborated.
- It will be good to highlight the limitations of traditional decision net. And then motivating how the models described in the extension section helped certain aspects of those limitations.
- The bibliography you added is not correctly referred in the text.
best, Prithu