Amazon SageMaker

Requirements for consuming the service: a. Subscription access to Amazon Sage maker and the model product. b. S3 bucket for specifying input and output locations. c. Necessary permissions to S3 locations and model packages. d. Code scripts for encoding/decoding files to/from base64 format for use in consumer scripts.

Input schema: To use the model create a batch transform job and place your input data in the following schema in a json file in your input s3 bucket: { "geo" : "AUS", "org" : "ABC Limited", "threshold" : "0.8", "file" : "sbfuf()&%#=dhbsdb++fb*++" } geo: The geographical location. Currently Australia (AUS) & Canada (CAN) are supported geographies. org: Name of the organization. threshold: Threshold of the confidence score file: Base64 encoded pdf file

Output schema: Batch transform job will create an output file of extension ‘.out’ in the output s3 bucket specified by user in the request. The schema of output is described below. { "status":"success", "output":[ { "text":{ "File":"ABC Limited", "Sentence":"Yours sincerely Mr Xyz Chairman." }, "classes":[ "BOARD_MEMBER_DESG" ], "confidence":[ "0.96" ], "entity1":{ "entity_name":"Mr Xyz", "entity_nerclass":"PERSON" }, "entity2":{ "entity_name":"Chairman", "entity_nerclass":"DESG" } }, { "text":{ "File":"ABC Limited", "Sentence":"In April 2014, Mr ABC was appointed Chief Executive Officer and has continued to play an instrumental role in driving and delivering key innovation and successes for the Group." }, "classes":[ "BOARD_MEMBER_DESG" ], "confidence":[ "0.99" ], "entity1":{ "entity_name":"Mr ABC", "entity_nerclass":"PERSON" }, "entity2":{ "entity_name":"Chief Executive Officer", "entity_nerclass":"DESG" } } ] }

status: Whether the request completed successfully. text: The details of inference i.e. which text in the file was used for designation extraction. a. File: Name of the organization b. Sentence: The sentence identified for information extraction.

class: Class of extracted relation. This is internal to IEF and may not be useful for the user. confidence: The confidence score of the extracted controller of organization. entity1 & entity2: a. entity_name: Name of the entity extracted from text. b. entity_nerclass: NER tag of the entity.

Examples for Base64 encoding
About Wipro HOLMES™ EKYC Solution that this model is part of