This is a sample template for SageMaker named entity recognition ground truth solution. This has 3 components

1. The HTML template that the workers will use to work on the task
2. Lambda functions for pre and post processing rules.

Note This template currently only supports one type of entity

1. Does not support multi-word entities on the user interface.

1. If you want try this sample but dont have the data to test this, you cam use this a sample file tests/sample_input_data_pubtator.txt as input data to evaluate the workflow.

2. Create pre and post processing lambda functions

   • Note Using the naming convention SageMaker-* for your lambda functions automatically gives access to Sagemaker using the standard template. Otherwise you would have to use create an IAM policy and provide access to Sagemaker to execute the lambda function

   • Create a lambda function SageMaker-EntityAnnotationPreProcessing with runtime python 3.6 using the code source/lambda_preprocess/preprocess_handler.py.

   • Create a lambda function SageMaker-EntityAnnotationPostProcessing with runtime python 3.6 using the code source/lambda_postprocess/postprocess_handler.py. Make sure this has access to read the s3 bucket containing the results from Sagemaker groundtruth job you are about to create

3. Configure SageMaker Ground Truth as follows:

   • Choose custom template in Sagemaker Ground Truth

   • In the custom template section, copy paste the html from source/template/entityrecognition.html

   • In the Pre-labelling task lambda function, select Sagemaker-EntityAnnotationPreProcessing

   • In the Post-labelling task lambda function, select Sagemaker-EntityAnnotationPostProcessing

export PYTHONPATH=./source
pytests