Overview of SeDDAT and Instructions for using it


SeDDAT- Semantic Driven Diversity Analytics Tool, consists of two main classes: "DiversityProperties_Variety_Coverage_Balance_WithinDisparity.java" and AcrossDisparity.java". Use these to generate diversity profiles of your datasets. There are 6 other classes that consist of functions for retrieval from input files and ontologies to support the diversity profiling.

Refer to Chpater 3 in thesis for description of diversity indices (algorithms in SeDDAT), profiles, and SEDDAT if you need more information.The thesis associated with SeDDAT is titled: Semantic Approach to Model Diversity in a Social Cloud by Entisar Abolkasim.

To use SeDDAT please:

- Downlod the ontologies (e.g. PreSOn) to your computer or locate its location and use this location as a URI for the input"Source" in the code in the main class "DiversityProperties_Variety_Coverage_Balance_WithinDisparity.java". If your entry point is "Thing", use the same URI as the Source. If you are using a different entry point, use the URI associated with the selected class (e.g. this can be found when you open ontology in protege"). The domain ontologies, PreSOn and Body Language OWL files are within SeDDAT project and can be downloaded from the link provided in thesis as well. The User Diversity Ontology is available from the same link. 

- Currently the input is read within the code for ease of use, input variables (e.g. Ontology URI, entry point, input files and ontologies) can be read from console if preferable.

Please note that:

- Variety, balance and coverage are calculated toghether in class: "DiversityProperties_Variety_Coverage_Balance_WithinDisparity.java", whereas across disparity is calauclated separetly in class "AcrossDisparity.java". Each class require seperate input file. The former takes XML file as  an input, while the later takes Microsoft Excel file. Sample files are provided to illustarte the process. The sample files are also in SeDDAT- One for domain diversity perspective of study A (illustrated in chapter 5 in thesis), labelled "StudyA-EntitiesPerVideo-DomainDiversityZoomin-OnlineCopy-v1.xml". The other file is for across disparity for the individual level, labelled "StudyA-FrequencyVector-IndividualLevel-OnlineCopy.xlsx".Using the sample files that are already within SeDDAT, you can generate diversity profiles. These results are reported in thesis if you are interested to check the analysis.

- Please follow similar file format for your input data and adjust code content based on your ontology (if you are using different ontologies). For example, AcrossDisparity.java class is designed to take up to 6 sub-categories of an entry point. This can be adjusted easily to be more or less based on your used ontology. This is illustrated in the code as the ontologies used in the thesis are different.

- SeDDAT was implemented in Eclipse.


- Please email me Entisar Abolkasim at: entesar_nsr@yahoo.com or my supervisor Dr Vania Dimitrova at: V.G.Dimitrova@leeds.ac.uk for more information.