Vim editor entities tagging/annotation tool

Question

I'm looking for a word/sentence annotation tool/plugin to be used INSIDE vim editor. Specifically I'm looking for a cli tool NER (named entity) tagging. Any idea/suggestion/tip?

By example, I would like to select a word or a list of contiguous words (using all vim conveniences) associating them to a a tag/label, say from a menu of tags (read form a config file, TBD).

The tagged text would be coloured with a nice background color and the file would be saved in a desired (configurable format). By example I would tag each NER span with a format like

[entity text span literal](entity_name)

By example, suppose I edit in vim a .txt or .yaml format (input) like this:

Original file (without tagging)

  - intent: body_temperature_data
    examples: |
      - il termometro misura 36.3 gradi
      - 36.9
      - 35.9
      - 35,9
      - la temperatura è di 36 e 7 gradi
      - 36 , 8

Tagged file: desired output file

I want to tag some numbers are entity body_temperature, saving the file (in output) like this:

  - intent: body_temperature_data
    examples: |
      - il termometro misura [36.3](body_temperature) gradi
      - [36.9](body_temperature)
      - [35.9](body_temperature) gradi
      - temperatura: [35,9](body_temperature)
      - la temperatura è di [36 e 7](body_temperature) gradi
      - [36 , 8](body_temperature)

Tagged file: visualization in NORMAL mode

Visually, during editing inside vim, I would like to see tagged words (numbers in the example) visualized with a background color (changing for each (preselected from a menu) label).

Example of visualization with web-based interface:

See visualization examples:

• https://github.com/doccano/doccano
• https://the-examples-book.com/prodigy/introduction

I guess one could realize al this coding a plugin in vimscript.

I have in mind a similar example, the plugin https://github.com/elzr/vim-json to escape double quotes in editing JSON. In that case when you are editing the current line, double quotes are visible, otherwise aren't. In analogy:

• when in INSERT mode I would like to edit full text LINE
• whereas, when in NORMAL mode I would like to tag with the above mentioned experience (selecting& tag). By example: the cursor is INSIDE a word or I selected a span pf text in VISUAL mode, afterwad, maybe with a special key, I display a menu with all possible enitiy labels and I chose a label from menu, tagging the text.

BTW, related question: How to replace selected text T with some function(T), using a keyboard shortcut?

Any suggestion is welcome. Thanks

Answer 1

I answer myself, just to share a solution I'm working on:

• Entities visualization

  I developed a vim plugin to highlight words or sentences with a nice colorized background:

  https://github.com/solyarisoftware/Highlight.vim

• Entities Annotation

  I solved implementing a plugin that associates function keys to macros that allow to "decorate" a span of text with the []() entity annotation syntax:

  https://github.com/solyarisoftware/nera.vim

• what is missing is the conceal behavior mentioned in comments.

Any comment/improvement here or on the plugin repo is absolutely welcome!