EMKA System and User Guide
About the tool
Documentation Links:

Parkinson's Disease is the second most common neurodegenerative affection worldwide. The exact cause of the disease is yet unknown, and as a direct implication, there is no complete treatment on the market. For a better understanding of the currently developing medical knowledge in relation with the existing background knowledge, there is a need for powerful natural language processing tools.

EMKA is using textual entailment processing of an expert medical data set, which was built using an ontology. The tool used methods of translation of the information stated in descriptive language in natural language. The hypothesis which resulted from this process, were used in the textual entailment level. Fuzzy logics were integrated in order to map the human thinking behaviour, trying to quantify the certainty of the decisions. Finally, the argumentation protocols which were developed for the mediation of opinions between two agents, helped to increase the quality of the enacted textual entailment process through joint learning. The architecture of the system can be observed in Figure 1.

Figure 1
User's manual
Login
  • Insert correct user name
  • Insert correct password
Figure 2
Verify Training Corpus | Testing Corpus | Onology Data
    For verifying the Traning Corpus, Testing Corpus or Ontology Data, without the permision of editing, press the corresponding labeled button.
Figure 3
Start Textual Entailment Process
    The Textual Entailment process can be started by accessing the Textual Entailment Process Start button, and the visualisation of the ongoing process can be done via the Eclipse Console.
Figure 4
Verify Results on TestSet
  • Verify the results of Textual Entailment for the first agent (see Figure 5).
  • In the section of the second agent, verify the results of Textual Entailment for the second agent.
Figure 5
Fuzzy Aggregation of singular decisions
  • Access Fuzzy Aggregation button.
  • Insert Pair Id.
  • Start the fuzzy aggregation process and view the results.
Figure 6
Joint Learning Steps
  • Access Joint Learning Enter button.
  • Follow all the 4 Steps by accessing the corresponding frames. (see Figure 7)
  • Step 1 - Identify the relation between agents regarding TE results and the correct protocol. (see Figure 8)
  • Step 2 - Fuzzify the relation.
  • Step 3 - Start the argumentation process.
  • Step 4 - View results.
Figure 7




Figure 8






Contact
mandy_nagy [at] yahoo.com
Adrian.Groza [at] cs.utcluj.ro