[Question type] Is pattern matching fully explained in documentation?

[trsh]

Document: https://winkjs.org/wink-nlp/custom-entities.html

Statements "Each option is separated by a vertical pipe character as in [NOUN|PROPN]" and "An option may be empty as in the case of the first two sets of options — [|DET] and [|ADJ]" suggest that the symbol "|" can be used as OR operator between 2 options, and if at beginning, the option can be absent.

Are there any more special symbols we can use to achieve certain pattern matching, besides "^" for escaping?

[sanjayaksaxena]

Hello @trsh

Idea to show absence right in the beginning is more of a best practice — basically to avoid any inadvertent miss.

There is only a single character — ^ available for escaping; if you wish to detect it in the pattern then please use ^^.

We can help further if you explain the use-case.

Best,
Sanjaya

[trsh]

@sanjayaksaxena thank you for your reply. 3 questions.

1. In my case the patterns can be tricky, something like a email, but different problem that would require a regex. But I see its not supported for custom entities? What would you suggest in such case?

2. Also a more simple case is Nouns that can be plural or singular, but should match in both cases. What would you recommend?

[trsh]

3. And how about matching entities that are very similar? Use case would be a misspelled word in text. Like someone wrote "bonana" instead of "banana".

[sanjayaksaxena]

1. Regex is not supported now. We are currently evaluating the possibility to support regex inside a pattern.

2. Current version only supports value or normal in the configuration. We will be adding lemma as an option in the upcoming version.

3. For misspelled words suggest that use options i.e. [banana|bonana].

[trsh]

1. Great.
2. How does value and normal connect to my question? I must be missing something here.
3. This is not an option, because I don't know input. It can be misspelled in lots of ways. That's why I ask about similarity feature.

[sanjayaksaxena]

1. Please refer to config section at https://winkjs.org/wink-nlp/learn-custom-entities.html.
2. Similarity inside pattern is not available. You can leverage wink-distance — it provides distances/similarities like levenshtein, jaccard and more.