manivannanmurugavel / spacy-ner-annotator

Train Spacy ner with custom dataset

spacy-ner-annotator

Installation

pip3 install spacy

Steps for usage

1. Open index.html file and open data on it.
2. Post annotations download the data and convert to spacy format using convert_spacy_train_data.py
3. Split data into train and test if you wish and add it to train.py
4. finally run the train.py after setting the hyper-parameters. Iterations are losses are logged in output_log.txt. And precision, recall and f1 scores are logged in train_output.txt and test_output.txt
5. Check progress by running losses_plotter.py.
6. If you wish to train over a model download the model and add its name in train.py

Details & Credits

Visit this url:

https://manivannanmurugavel.github.io/annotating-tool/spacy-ner-annotator/

spacy-ner-annotator's Issues

The json training file isn't saved

Hi, thank you for your work, your interface and code would be very useful to my project !
I tried using it and after annotating a few sentences, i saved the json file but it wasn't saved on my computer, is there an issue ?
Edit : It works on google chrome ;)

Cannot capture multiple words at once

Hi,

When I try to highlight a sentence. It would state an error saying: 'Please select entity inside content'

What I want to do is input the sentence as a whole.

Convert to Spacy format

Hi !
Does your tool convert files generated by webanno (uima binary cas, uima cas json, uima cas xmi, conll, tsv3) to Spacy NER training data format?
Thank you in advance !!!

Graphic problem when scrolling down the text in the Content Section

Hi, firstly thank you for your work.
When I put in the Content window a text which is longer than the window and press the save button, when scrolling down the text, there is a graphic bug that overlap the text.
I attached a photo to show the problem.

# thanks giving

Great work . I find this helpful . please ignore this issue if you all want. Just give a great thanks to you guys .

GoldParse is no longer available

Since the GoldParse is no longer available in spacy. How would you go about rewriting the evaluate function?

Error : object of type 'NoneType' has no len()

i have used the code exactly as written, changed the TRAIN_DATA, used your ner annotator to get the data but still the error thrown as soon as train_spacy function is called.
it's returning null I guess hence the error. Please help.

Annotator is taking the same indices if the two entities have same value

For example, if the text is : Start Date/Time: 4/12/2019 2:00:00 AM EASTERN End Date/Time: 4/12/2019 6:00:00 AM EASTERN Description: Section Roll Expected Impact: 4 Hours Contact Information: Call the SCCC, and reference RFC-0051317 Thank you, Cincinnati Bell Change. It is giving the following output:
"entities": [[627,636,"StartDate"],[637,647,"StartTime"],[627,636,"EndDate"],[681,691,"EndTime"],[648,655,"TimeZone"].
Here the value of start and end date is same (4/12/2019). So, it is taking the same indices for both the indices. Please check it. Thank You in advance

How do we start the server

I have a basic question, how do you start the server, so that we can open in the webpage and train the data.

The youtube video is in silent mode, so I couldnt hear what you were telling.

Selected text is double when tagged

I'm selecting text to associate with an entity label but sometimes when I do so, the text itself (and sometimes other unselected texts) is repeated in an overlay that looks like a bug.

I have seen this in Chrome and Microsoft Edge.

Save into an existing model

I have a question, when
modelfile = input("Enter your Model Name: ") prdnlp.to_disk(modelfile) is executed and. it asks us for the model name, would entering an already existing model dictionary name save it there or would. it conflict?

Thanks

Cannot download json

Browser fails to downlaod the train.json once I click on the complete button

This is a good tool as I annotated everything in one hour. But as the title says, my browser (Internet Explorer) does not download my train.json when I'm done. Any idea to solve this?

Need atleast one array to conatenate error

Getting this error

add keyboard shortcuts

keyboard shortcuts to choose class, skip, next content and complete.

TypeError: 'tuple' object is not callable

Thank you for your annotator.

But when I executed train.py after copying text into TRAIN_DATA variable, I got this Error

File ".\train.py", TypeError: 'tuple' object is not callable.

I am working with python 3.7.

Could you please help me ?

Labels disappears if clicking the "Edit" button

Hi,
suppose I have labeled some text.
At some point I want to edit it for some reason.
If click on the Edit button and then Save, the labels disappear.
Is this a problem or an expected behaviour?
In addition, when i delete some text, it doesn't desappear until I don't click the "Save" button
Many thanks in advance