Entity recognition at random string location

[riyaz7us]

I need to extract the numeric entities from the utterances like the following:

find schools with 3 gpa
find schools with gpa 3
find schools with gpa score 3
find schools with gpa 3 and rank 5

Without training anything, it labels these entities as "number", to which I would like to give the appropriate name like "gpa" or "rank"
Although we could use addAfterCondition or addBeforeCondition, It could work with either of them, not both.

Is there any implementation that identifies these entities at random location? Please help, Thanks!

[MarketingPip]

@riyaz7us - I am not a maintainer - that said this library is pretty well dead. Tho I would suggest using possibly a combination of something like this library and a library like compromise or another NLP package / library that can be used to extract entities etc...

Hope this helps you.

[Apollon77]

I would porpose to create PRs ... then we can see how it goes forward :-)

There are communications with the AXA Open source team to solve the situation.

[MarketingPip]

I would porpose to create PRs ... then we can see how it goes forward :-)

There are communications with the AXA Open source team to solve the situation.

@Apollon77 tho I would love if that were true. I think it's a pipe dream to think this will get carried on by Axa Group.

I think the most practical solution would be for you etc - to carry on the task of a fork of this project and go from there.

Or - scrap this project (use some ideas for project structure / layout) and build off the giants before you.

Possibly a NLP system with intents - and if intent is detected and entity needs to be extracted etc (runs a HMM model to extract entities 😉).

ps; let's keep in mind too - soon or later we will be able to utilize AI models everywhere (almost already can). Which again might make project like these obsolete besides basic intent recognition etc for hard set rules in AI model etc....