D3.3 Implement hierarchy reconstructing methods

Humans are social creatures. They try to find contacts with each other, form communities and act in groups. When a group grows larger, or needs to achieve some goal, the loosely connected “herd” becomes a structure. This structure, called hierarchy, helps to make the group to function in a more effective way. Advises, instructions or questions and requests will be exchanged within the group. Or, in general, information will flow inside the group or even further, between the group and the environment.

The 3rd  Task and Deliverable of the Social Network Analysis (SNA) work package of the RED-Alert solution, aims to provide software solutions for revealing the hidden, sometimes unconscious hierarchical structure from unstructured relations or simple co-occurrences.

The Task T3.3 “Implement hierarchy reconstructing methods” of SNA work package fits well into the RED-Alert solution. The task builds on results coming from the NLP and SMA work packages. Flat, unstructured relations are expected as input from T2.3 “Linguistic feature identification and extraction” and T2.4 “Audio, image and video feature extraction”. Note, that all inputs will be anonymized by the tools from T5.1 “Privacy-preserving data processing” in order to ensure protection of innocent individuals. The main output of this task will be utilized mostly by the interactive user/analyst through T5.2 “Data visualisation tools”. We expect that the CEP engine will benefit from longer sequences of time ordered output of the module in the T4.2 “Event driven probabilistic model” to cope with uncertainty in event content, event occurrence, and event rules”.

Read the full report here: D3 3 Hierarchical models FINAL