Upgrade to UDPipe 1.1.0

[reckart]

https://github.com/ufal/udpipe/releases/tag/v1.1.0

• probably don't need the hack for installing the library locally anymore (Hardcoded static library loading code conflicts with custom library resolving ufal/udpipe#10)
• upload the ud-pipe.jar to UKP repo or Maven Central (@reckart)

Additional tasks deferred to #1141.

[reckart]

@kouylekov-usit is this something you would like to do?

[kouylekov-usit]

Yes we are. Should I prepare the new build script ? I have created a version with the segmenter wrapper to handle the offsets (will make a pull request for this one). But will be interesting to see what they mean by: * Improve|input_format|API to allow inter-block state (for correct tracking of inter-sentence spaces and document-level offsets). Cheers Milen

…

On 03/31/2017 03:41 PM, Richard Eckart de Castilho wrote: @kouylekov-usit <https://github.com/kouylekov-usit> is this something you would like to do? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1039 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AQGHYI1YMryJwLurgyRNYUqmXqHrxocSks5rrQJ8gaJpZM4Mvrka>.

[reckart]

@kouylekov-usit - yes, the build.xml needs updating - not sure if the models changed, but the binary definitely. No idea what the format change means.

[kouylekov-usit]

I am on it On Apr 3, 2017 11:31 AM, Richard Eckart de Castilho <notifications@github.com> wrote: @kouylekov-usit<https://github.com/kouylekov-usit> - yes, the build.xml needs updating - not sure if the models changed, but the binary definitely. No idea what the format change means. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#1039 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AQGHYAL6rMBlg5I70Vm_4LRM9sAB3S0zks5rsLxPgaJpZM4Mvrka>.

[oepen]

i have been in touch with milan stranka (the UDPipe developer) on a different project; in that thread, he wrote

UDPipe 1.1 is able to store token offsets in the CoNLL-U [...]

so possibly the ‘magic’ to recover character offsets from comparing tokens to the underlying document may no longer be required when moving to the new version?

[kouylekov-usit]

I will verify. We have a working/tested version of the wrapper if it does not. Cheers Milen On Apr 3, 2017 11:39 AM, Stephan Oepen <notifications@github.com> wrote: i have been in touch with milan stranka (the UDPipe developer) on a different project; in that thread, he wrote UDPipe 1.1 is able to store token offsets in the CoNLL-U [...] so possibly the ‘magic’ to recover character offsets from comparing tokens to the underlying document may no longer be required when moving to the new version? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#1039 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AQGHYEIdCRSvBwjJ-kq1yXFWjv11uVIoks5rsL5EgaJpZM4Mvrka>.

[reckart]

@kouylekov-usit I have uploaded a new udpipe 1.1.0 JAR file to the UKP repo and also deployed the new binary JAR there. The POM has been updated accordingly.

[reckart]

@kouylekov-usit You should check again in the new 1.1.0 Java API that is now available to the module if you find the methods for accessing the offset information.

[reckart]

@kouylekov-usit any news regarding the offsets or any further plans atm?