Research Projects

Adaptive Learning System for Training of First Responders – Master's Research (In progress)

Look for UX skills including Research design, User interviews, Understanding, Analysis

One of the most valuable resources of any organization is the implicit knowledge of its employees and the lessons they have learned by working there over time. As these employees retire or go to other organizations, that knowledge is lost. Our project is focused on emergency response domain and one way to capture that knowledge in this domain is to extract information from lessons learned documents or After Action Reports (AARs). However, those documents often sit on a file server unlooked at after they were created. My end goal is to build an adaptive learning system to train new personnel by harvesting the knowledge in these reports, as encapsulated in the AARs. To do this, we will have three parts:

  1. Information extraction from the existing documents using NLP techniques.
    • Manual extraction of strengths and weakness from After Action Reports by Subject Matter Experts (SMEs). These documents are called annotations.

      • Fig.1 Sample Annotation

    • Find similarities between participants' annotation of each document using Bag of Words Approach.

    Fig.2 R Code snippet
    Fig.3 Three documents were annotated by three SMEs. All the nine combinations are shown

  2. Storage of this information in an ontology with axioms (logical rules) to provide domain knowledge for common scenarios (for example, a fire may require certain responses, which would be inappropriate for another type of disaster, like a flood). (In progress)

    3. Fig.4 Ontology Model. Information in ontology was obtained from Subject Matter Experts

  3. From this information, we can create new training scenarios and evaluate how well new employees respond based on strengths and weaknesses extracted from previous scenarios as described in the AARs. (In progress)