ETD system

Electronic theses and dissertations repository

 

Tesi etd-05262009-114423


Thesis type
Tesi di laurea specialistica
Author
BENOTTO, GIULIA
URN
etd-05262009-114423
Title
Semantic relation extraction and classification. Experiments on Wikipedia.it
Struttura
INTERFACOLTA'
Corso di studi
INFORMATICA UMANISTICA
Supervisors
Relatore Prof. Lenci, Alessandro
Parole chiave
  • relation classification
  • wikipedia
  • semantic web
  • relation extraction
Data inizio appello
11/06/2009;
Consultabilità
Completa
Riassunto analitico
Semantic relations between concepts or entities exist in textual documents, keywords or key
phrases, and tags generated in social tagging systems. Relation extraction refers to the
identification and assignment of relations between concepts or entities. Basically, it can explore relations that are implicit to underlying data and then add new knowledge to the different domains.
The purpose of our work was to develop a semi-unsupervised system that was able to automatically extract semantical relations between nominals in a dump extracted from the ialian Wikipedia in November 2008. In addition, we wanted it to correctly classify semantical relations between nominals.
We used a seed-based, pattern-based, semi-unsupervised approach for Relation extraction, while we implemented a variation of Vector Space Model for relation classification. we used manually selected seeds for both purposes. in addition, we implemented a script for the automatic extraction of seed pair to be used with our algorithm.
File