Implement agglomerative clustering for value categories

[lisv]

Following the meeting on the 8th of March

[lisv]

Update

• ResponsesFitnessClustering uses agglomerative clustering to calculate the fitness for individuals

• A hashmap is kept that contains the keys of the responseObject and the corresponding values of AgglomerativeClustering objects.

• Clusters are thus created per response object structure (the JSONObject keys in the response). The values corresponding to the keys are put in a feature vector and the feature vectors are then clustered based on similarity metrics (defined for strings and numbers, also for booleans but this is not used/redundant).

• To take into account the depth of values (e.g. if a JSONObject A contains some values, of which one is another JSONObject B, the values of JSONObject A should be more important for the dissimilarity than the values of JSONObject B), I implemented a weightVector that contains weights corresponding to the values in the featureVector.