uimaFIT S4 SDK

Skip to end of metadata
Go to start of metadata
Table of Contents
Source code
The source code for the uimaFIT SDK is available from the S4 GitHub repository.

Introduction

This SDK provides access to the S4 text analytics services based on the uimaFIT library for configuring, running and testing UIMA text analytics pipelines. The S4 uimaFIT component does not have any requirements or assumptions about the type of pre-processing or post-processing of the textual data being annotated. It can be used as a standalone component of as part of uimaFIT pipelines. The following sections describe how to integrate and execute S4 text analytics services as part of uimaFIT pipelines.

Prerequisites

  • Maven dependency of the uimaFIT-s4-sdk
  • valid S4 API keys. More details on acquiring S4 API keys are available in the S4 Management Console documentation.

SDK Usage

The uimaFIT SDK for S4 transforms the results of the S4 text analytics services into XMI format (XML Metadata Interchange), so that the annotations can viewed with the various UIMA GUI tools (Document Viewer or CAS Visual Debugger) or further processed by other UIMA tools.

S4 Console Tool

We have included a sample pipeline in the SDK which also could be launched through a special console tool called s4pipeline (available for download here)

After downloading the executable jar you can see all its available options:

Note that the options prefixed with "*" are required

Example usage of the tool:

Java Code

To create a simple uimaFIT pipeline you have to follow the three steps:

  • Provide S4 Annotator parameters
  • Instantiate the component descriptions
  • Run the pipeline to process the input documents

View the Output Annotations

  • Step 1: Open Annotation Viewer GUI

A GUI like this will pop up:

  1. Input directory -> the directory with your processed documents
  2. TypeSystem or AE descriptor file -> path to a type system XML file describing the annotations or an Analysis Engine descriptor file containing a type system description
  • Step 2: Click View and select a document by double-clicking on it:

  • Step 3: Explore the results

Labels:
None
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.