Good thought. With a semantic tree, you work with distinctions between words more, and you can have deeply nested subcategories. The subcategories can also contain information on the usage of words; for example, you can make a distinction that some words can't be used with people or can only be used with people. At the very lowest subcategory, you could also add frequent collocations or, for Chinese, the coverbs, or add example sentences for a single word. So you are much freer to add relevant information and represent it sensibly. There isn't enough room for such information in a thesaurus, where you just have about 10-20,000 categories of words, with 10-50 words each.
It is of course much less complex to compose a thesaurus than a semantic tree. But it would surely be great to have a good semantic tree.
Good afternoon Sobriaebritas, (it's still 10:45 AM here, sometimes I wrote at night)
this is truly excellent material, even in Chinese, thank you! It forms an excellent basis for a Chinese semantic tree. You must be quite an expert. I can basically discard what I've done so far and start with your semantic categories. If I consider it a hobby activity and stay with it, maybe in 3 years I will have something presentable.