• Name : Djem Toker
  • Age : 25
  • Country : Japan
  • Town : Kawasaki
  • Job : Software engineer
  • University Paris 13 - Sup-Galilée

Valid XHTML + RDFa Valid CSS !

[Valid RSS]

Web interface for precalculated ontologies

This project aimed to make an interface that display precaculated ontologies. The Ontologies are calculated by the LIM&BIO platform. The results of the platform are hardly human readable files of millions of lines reffering informations with IDs. This make hard the visualisation of the files for the human. The information calculated are done on medical subject and more particulary on the diabetes. Keep in mind that the information proceeded are written by human, so getting the sense of natural written text isn't an easy topic, especially when it's something complicated such as medicine.

There is multiple aim of this project :
  • Finding competences needed for the diabet disease for daily life by the people affected and their neightbour (family, friend ...) that the could help for the treatment. It could also be providing simple information to the people and providing them information on the disease.
  • Classifiying the topics into different competences
  • Finding relation between the symptoms and the medication gaven with the different information such as administration mode, dosage and so on
This tool provide as a result two main view :
  • One to view specific information on the topic such as named entities or semantic relations and so on
  • One to view the topic per category or skill required

This tool is used to show the possibility of the platform to expert working in this kind of subject and eventually provide them such kind of automatic treatment. It it also used to debug the eventual problem in the platform since it's more easy to check for bugs with the interface rather than checking into large XML files and checking the ID everywhere.

This program has been done during my internship in the end of my 2 year (4th university year) of engineering school and contain aproximatly 25 000 lines of code. The tool is written using the language PHP/HTML/Javascript/XML/XSD/DTD and also use bash script for index generation since the tool is made to parse and link information between thousand of files of thousand of lines. This tool has been optimized using php profilers and basic optimisation methods based for instance on caching methods.

Link to the project
The code source for this project is not avalaible.
Home hover About hover Portfolio hover CV hover Projects hover Contact hover