navigating-stories / orange-story-navigator

Add-on to the Orange3 data mining toolkit with text processing widgets from the project Navigating Stories
https://research-software-directory.org/projects/navigating-stories
Other
2 stars 2 forks source link

Custom tags dictionary words are not correctly highlighted in Actors and Actions widgets #77

Closed kodymoodley closed 2 months ago

kodymoodley commented 3 months ago
Story Navigator version

0.0.19

Orange version

3.37.0

Expected behavior

Correct words are highlighted in green when "Custom tokens" checkbox is checked in the Actors and Actions widgets. That is, only the words contained in the custom dictionary I supply should be highlighted and not any other words.

Actual behavior

Many other words in the story texts are highlighted in green that are not in the Custom tags dictionary file. See screenshot below:

image

Steps to reproduce the behavior

Create the workflow as depicted in next section. Import the stories attached and import the custom dictionary file attached. Open the Actors / Actions widgets and ensure the checkbox for "Custom tokens" is selected.

Additional info (worksheets, data, screenshots, ...)

Workflow I used:

image

Stories I loaded: corona_testdata.zip

Custom dictionary file I used: dict_single_column.csv

kodymoodley commented 3 months ago

Similar to #46 and #48 but the solution proposed in #46 does not apply here (I have tried it).

kodymoodley commented 3 months ago

Theory for what could be going wrong: the column data type of the custom tags dictionaries could be being converted back in the code from text to categorical (both are internal Orange column data types).

ThijsVroegh commented 2 months ago

Hi @kodymoodley , I tried to reproduce the error above, but in my case it does seem to work correctly (i.e., I only get the words from the custom tag list). Do you think this error might be dependent on the system one has (which I doubt btw)?

ThijsVroegh commented 2 months ago

Hi @eriktks, could you perhaps also briefly check if you get the same results as Kody? It works fine on my computer, so I wanted to double check.

ThijsVroegh commented 2 months ago

@eriktks ; nevermind, I reproduced the error now..

eriktks commented 2 months ago

solved by PR #79