Recommended pipeline analysis tagged data

閲覧: 10 回
最初の未読メッセージにスキップ

Yanir Marmor

未読、
2021/09/06 3:45:232021/09/06
To: inception-users
Good day, everyone!

I'm planning to begin a hermeneutic and literary labeling project for a text, and I'm considering using Inception and adding some dedicated tags.
However, I'm attempting to determine the best method for continuing to analyze the tagged data. Is it necessary to export the tagged data or can it be analyzed internally? What are the recommended tools for performing basic statistical analysis, visualizations, and queries on the tagged data if it is impossible to analyze inside?
Unfortunately, as far as I know, there was no reference to the subject of analysis in the documentation.

Thanks,
Yanir

Zesch, Torsten

未読、
2021/09/07 15:56:382021/09/07
To: incepti...@googlegroups.com

My understanding is that for most analyzes (beyond inter-annotator agreement etc.) you need to export the data.

 

As we had similar needs, we have started developing some more analysis capabilities and recently made it open source here (warning: still in early development):

https://github.com/zesch/inception-analytics

 

-Torsten

--
You received this message because you are subscribed to the Google Groups "inception-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inception-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/inception-users/a7e9fa00-a934-41d6-9d3b-70be5c0aee5an%40googlegroups.com.

Richard Eckart de Castilho

未読、
2021/09/07 16:00:432021/09/07
To: incepti...@googlegroups.com
Indeed, exporting and external analysis is currently the only option.

That said, we have started looking into using the advanced functionalities of
the annotation indexing library that we are using (MTAS) and aim to provide
at least a few basic statistical functionalities based on that. But the full
potential range of analysis one might care to perform cannot be built in,
so there will always be the case for external analysis such as the project
Torsten pointed to.

Cheers,

-- Richard
全員に返信
投稿者に返信
転送
新着メール 0 件