Ontology-driven normalization and duplicate elimination of postal addresses in multi-lingual environment (ONDE)

Starting date
August 1, 2008
Duration (months)
12
Departments
Computer Science
Managers or local contacts
Cristani Matteo
Keyword
Statistical natural language processing, Formal and descriptive ontologies, Knowledge representation, Artificial intelligence, database

  1. Comparative study of the postal addresses placename classes, of their classification patterns and of the class matching penalties given to the patterns. The study is based upon the Poste Italiane’s official address book and it is related to the placename classification to which the current literature refers, in particular to the classification used by urban information systems and by satellite navigators;
  2. experimental analysis of the classification’s effectiveness by taking the precision value and by analysing the rejected addresses;
  3. reconstruction of the classification by integration, clustering and revision;
  4. experimental study of a pre-existent prototype behaviour. The prototype was developed by the Department of Computer Science in order to solve the problem of the duplicate elimination and revision;
  5. development of a prototype for the addresses normalisation based upon the classification built as in Point 1;
  6. integration of the duplicate elimination prototype and the normalisation one.

Sponsors:

ADDRESS SOFTWARE S.R.L.
Funds: assigned and managed by the department
Syllabus: PROGATENEO - Progetti d'Ateneo
Ateneo
Funds: assigned and managed by the department
Syllabus: PROGATENEO - Progetti d'Ateneo

Project participants

Elisa Burato
Matteo Cristani
Assistant Professor

Activities

Research facilities