how to handle aspect categories which may not be in the sentence?

[harrywang]

Thanks a lot for developing such a nice framework and related tools and datasets.

I have a question about annotation - the aspect terms are from the reviews and some aspect categories might be pre-defined and not in the text.

For example, "the waiter is nice and we spent lots of money on this meal': the waiter and money can be annotated using your tool as aspect terms but what is there are two pre-defined aspect categories such as service and price? Can PyABSA handle this? If yes, how?

Thanks again!

[yangheng95]

Not available for this now. This case belongs to asepct category detection.

[yangheng95]

If it is possible, you can contribute the initial implemention of asecpt category detection and datasets in the future.

[harrywang]

If it is possible, you can contribute the initial implemention of asecpt category detection and datasets in the future.

We will try to see what we can do. Thanks!

[harrywang]

Do you by any chance know any Chinese aspect category detection and polarity classification open system/code that we can use? Thanks a lot!

[yangheng95]

You need to search by yourself on github or elsewhere, there may be no code as easy as this repo.