sbstnerhrdt / labeling Goto Github PK

https://mpg-labeling.erhardt.net/app/organizations/0b913295-e846-4b4b-b343-c77c01ede816/59bf8694-67ee-11ec-90d6-0242ac120003/87ffd7de-67ee-11ec-90d6-0242ac120003/label

Raw data shows:
...specimens. Results: The mean ...

Label tool shows:
`...specimens ns. Resu The mean...

if a user has labeled a text wrong, the user should be able to quickly change the mistake by right.-clicking on the label an choose the correct one instead.

Add multi labeling for in combination with aspect based sentiment analysis

Problem: If multiple users create labels for the same instance, the system might have problems detecting them. Its not feasible to separate them based on the author, since updates are possible.

Solution: Create a data structure for a so called label group. Where multiple labels are combined.

• Data Structure
• Backend GraphQL Resolver
• Adapt frontend mutations and queries

Must be implemented before #13, since the grouping functionality must be done first.

The selection of text does not work on Firefox. It works on Safari on the same laptop.

Steps

• OpenHome>Organizations>MPI-INNO>Logic-Mill>Label Abstracts and then Start Labeling
• Select text to remove with the mouse

Result

• Nothing happens
• Error in the console:

TypeError: t.baseNode is undefined
    NuxtJS 4
        selectText
        ee
        n
        _wrapper

Expected behavior

• text turns red and is add to list of text to remove

System info:

• OS: MacOS 11.6
• Firefox: 95.0.2 (64-bit)

When including an existing selection in a new selection, a very large arbitrary area is being selected.

Now select specimens. Results: the mean

Result:

After that, no normal selection can be made

Work around:

• Reset all

When labeling, sometimes you are in doubt. Is this a reference or not? I don't know how to label something.

• An option would be to either continue/cancel this item and to continue with the review of the next one. The item may show up for someone else, who knows how to handle this.
• Another option would be to mark the item so it gets reviewed. For example: I am unsure if this is a reference that should be marked. You may need to add a note for the reviewer.

Once a text is selected and another text is selected within the selection, the app break.

Export the data to xml std

If the user labels a long text, the user must scroll down and up again to change the labeling class.

Make the label bar sticky

extend the labeling tool for aspect based sentiment analysis

• Create Overview of already labeled instances
• Filter by
  • Flag
  • User