Retrieve values from UIMA FSArray

Question

I have an annotation which has a feature of the type FSArray. This feature should contain a list of strings.

FSArray fsArray = (FSArray)annotation.getFeatureValue(fe);

How do I get the list of strings from the FSArray?

Looping through fsArray.toStringArray() only returns the string "FSArray" and not the actual value.

Answer 1

There are some important concepts which are important to understand when retrieving values from a FSArray within UIMA:

• org.apache.uima.cas.Type - The Type describes the data model. It's similar to the concept of classes in java. A Type has a name space and it defines attributes (features).
• org.apache.uima.cas.Feature - Is an attribute described by a Type.
• org.apache.uima.jcas.cas.TOP - Is the most general Type and could be compared with java.lang.Object.
• org.apache.uima.cas.FeatureStructure - A FeatureStructure can best be described as an instance of a Type. The FeatureStructure is what you use to access data.

Let say that we have the following two Types:

• com.a.b.c.ColoredCar
• com.a.b.c.Car

And we have the following sentence:

Car A and car B are both blue.

Lets assume that a previous UIMA stage has annotated the entire sentence using the Type com.a.b.c.ColoredCar as following:

begin: 0
end: 24
color: "blue"
cars: FSArray

Lets also assume that we known from the type definition that the feature cars is an FSArray of com.a.b.c.Car and that Car contains the following values:

begin: 4
end: 5
manufacturer: "Volvo"

begin: 14
end: 15
manufacturer: "Toyota"

The following code will then demonstrate how to retrieve the manufacturer attributes / features of the cars FSArray.

public void process(JCas aJCas) throws AnalysisEngineProcessException {
    List<TOP> tops = new ArrayList<TOP>(JCasUtil.selectAll(aJCas));
    List<String> manufacturers = new ArrayList<>();
    for (TOP t : tops) {
        if (t.getType().getName().endsWith("ColoredCar")) {
            Feature carsFeature = t.getType().getFeatureByBaseName("cars");
            FSArray fsArray = (FSArray) t.getFeatureValue(carsFeature);
            FeatureStructure[] arrayStructures = fsArray.toArray();
            for (int i = 0; i < arrayStructures.length; i++) {
                FeatureStructure fs = arrayStructures[i];
                Feature manufacturerFeature = fs.getType().getFeatureByBaseName("cars");
                manufacturers.add(fs.getStringValue(manufacturerFeature) );
            }
        }
    }
}

To dig deeper into this, it's a good idea to read how Type systems, Heap and Index Repository works within CAS.