I tried to add a sentence for 改变主意 that I found on tatoeba.com:

"即使 太阳 从 西边 出来 ， 我 也 不会 改变 主意。“

I got an error message: "That sentence already exists."

Does this mean the sentence is in the database and assigned for other words? Why can't it also be used for this phrase?

Because it's being segmented into 改变 and 注意, so it's showing up for those words, but it doesn't think that 改变主意 is one word in that sentence, and so it's not indexed like that.

I think this problem might happen often for phrases which are compounds themselves: the sentences which contain them won't be indexed that way. Hmm. It's possible that we could tweak things so that you it would be easier for us to combine words in the segmentation, but that also comes with its own problems (the sentence wouldn't show up for 改变 or 注意--not that this matters for those words, which have tons of sentences).

Nick, I'll leave it to your better judgment. I've only found the problem with 改变主意 so far, which I can live without a sample sentence. Once the sentences are integrated into the app, I might change my mind, but so far it's not a big deal.

Can't wait to have this in the app by the way.

Thats going to be a problem. There are thousands of four syllable "words" that are similarly structured to the example above.

In terms of solutions, why can't they be indexed to both?

Suffice it to say that it's technically infeasible to index the database this way. Let's see how well the segmentation works in practice first, though.

JinXiao,

I thought the same, and started looking at some sample 4 character words, and inputting sample sentences ... and it's been working so far. The only one that's been rejected so far is 改变主意. You should try the system out first.