Go-Freeling: Natural Language Processing in Go (opens in new tab)

(github.com)

80 pointscnbuff41011y ago16 comments

16 comments

14 comments · 7 top-level

panamafrank11y ago· 3 in thread

There are no unit tests, and the only comments are odd bits of code. How do you know any of it works?

Dewie311y ago

Maybe he reasoned about his code in his head/on paper?

panamafrank11y ago

where's the proof that his reasoning is correct?

Dewie311y ago

You know as well as I that most programming reasoning is informal and doesn't have anything to do with any kind of proof.

1 more reply

andrewstuart211y ago· 2 in thread

I'd suggest making sure it's go-gettable and will build following the instructions in the README when placed in the canonical go path ($GOPATH/github.com/advancedlogic/go-freeling). After fixing the paths, I'm still getting a cgo linking error with the mitie.h file.

Also, probably avoid using . imports, as it's considered a Go antipattern.

Seems really promising, but if you want it to become a project that the Go community can get behind and maintain, you'll probably want to make sure it's familiar and easy to get running. :-)

mrkrwtsn11y ago

I'm unable to get it to run either. Turning text into tokens is something I've wanted to be able to do in golang before, so it would be super nice if it was easy to install with `go get`. I still need to work through some of the other libraries here: http://biosphere.cc/software-engineering/go-machine-learning...

I agree though, seems promising.

mschoch11y ago

If you just want to segment larger blocks of text into tokens you can try the segment library (it implements the word boundary portion of unicode annex 29):

https://github.com/blevesearch/segment

If you need more manipulation of tokens after segmentation/tokenization, you could look at the analysis sub-package of bleve. Its intended to be able to be used indepenently of the rest of the library.

https://github.com/blevesearch/bleve

roscopcoltrane11y ago· 2 in thread

Why is it underhanded to let a user know that his government is censoring/surveiling his posts? This should be the freedom-c contest.

roscopcoltrane11y ago

oops. commented on wrong post. sorry

Dewie311y ago

You can delete comments if you want.

frik11y ago

Very nice, something I have been looking for.

Please create a wiki with some more information and examples. Is there a training part involved? Does it support languages other than English, if so which data format or how to train the language model? Is there a WordNet integration? Or can I use the FreeLing website as reference http://nlp.lsi.upc.edu/freeling/index.php?option=com_content... ?

hendler11y ago

The newer versions of Freeling have a socket server you can use from any socket client. I wrote one as my first go project a while back

https://github.com/Hendler/goling

nkozyra11y ago

Excited to try this, although I don't immediately see what model format is used. I've been a bit surprised to see ML and NLP lag a bit on Go compared to rather swift adoption with other languages.

hoare11y ago

that's awesome! im a little worried about test coverage though. Did you run some kinds of performance tests?

j / k navigate · click thread line to collapse

16 comments

14 comments · 7 top-level

panamafrank11y ago· 3 in thread

There are no unit tests, and the only comments are odd bits of code. How do you know any of it works?

Dewie311y ago

Maybe he reasoned about his code in his head/on paper?

panamafrank11y ago

where's the proof that his reasoning is correct?

Dewie311y ago

You know as well as I that most programming reasoning is informal and doesn't have anything to do with any kind of proof.

1 more reply

andrewstuart211y ago· 2 in thread

Also, probably avoid using . imports, as it's considered a Go antipattern.

Seems really promising, but if you want it to become a project that the Go community can get behind and maintain, you'll probably want to make sure it's familiar and easy to get running. :-)

mrkrwtsn11y ago

I agree though, seems promising.

mschoch11y ago

If you just want to segment larger blocks of text into tokens you can try the segment library (it implements the word boundary portion of unicode annex 29):

https://github.com/blevesearch/segment

https://github.com/blevesearch/bleve

roscopcoltrane11y ago· 2 in thread

Why is it underhanded to let a user know that his government is censoring/surveiling his posts? This should be the freedom-c contest.

roscopcoltrane11y ago

oops. commented on wrong post. sorry

Dewie311y ago

You can delete comments if you want.

frik11y ago

Very nice, something I have been looking for.

hendler11y ago

The newer versions of Freeling have a socket server you can use from any socket client. I wrote one as my first go project a while back

https://github.com/Hendler/goling

nkozyra11y ago

Excited to try this, although I don't immediately see what model format is used. I've been a bit surprised to see ML and NLP lag a bit on Go compared to rather swift adoption with other languages.

hoare11y ago

that's awesome! im a little worried about test coverage though. Did you run some kinds of performance tests?

j / k navigate · click thread line to collapse