-
Notifications
You must be signed in to change notification settings - Fork 1
Whitakers: handle alternate spellings of principal parts #17
Comments
Test cases for this: aberis |
Looking more closely at the whitakers output, it seems that most, if not all of these are due to differing spellings of the principal parts. So we can apply the same meaning to all of them. I guess we need to allow for multiple variations on spellings of principal parts, aggregated in one entry. E.g. here is how we treated it in V1: |
started work on this. Still to be done: when aggregating lemmas for a lexeme, make sure the lemma that is assigned as the primary lemma is the most frequent one. |
needed for #17 - lemma variations often come from different sources
@monzug you can test this with the build in https://github.com/alpheios-project/webextension/tree/issues-whitakers-engine |
@monzug this can also now be tested in https://github.com/alpheios-project/webextension/tree/qa-2.0.3-3 |
tested in Chrome in build 2.0.3-5. |
Bridget, giving back to you. |
yeah, this fix only fixes some of the scenarios. I wasn't sure if all of the words listed above fell into this category. Most do, as you have noted some do not. There are issues on the morphsvc which describe some of the other scenarios I found: Some of these may be problems with the original Whitaker's source code, and some are problems with our wordsxml wrapper on top of it. This fix addresses the scenario where our wordsxml wrapper puts more than one There is only so much normalize I can (and really should) do on the client side here. We will have to decide if we are going to open up the old Ada code or find a new parser to fix all of them. |
let me know if you want the list of which word has been fixed, which one doesn't look like it could be fixed, and the one that might be merged. |
yes that would be great. thanks! |
here we are. I added a number next to each word
aberis 1 |
Thank you! Have split the 2s off into a new issue at alpheios-project/morphsvc#12 |
See alpheios-project/morphsvc#3
If a parser returns multiple
dict
elements and a singlemean
element should themean
be applied to both? Can we recover from this parser error?The text was updated successfully, but these errors were encountered: