This chapter presents a model for knowledge extraction from documents written in natural language. The model relies on a clear distinction between a conceptual level, which models the domain knowledge, and a lexical level, which represents the domain vocabulary. An advanced stochastic model (which mixes, in a novel way, two well-known approaches) stores the mapping between such levels, taking in account the linguistic context of words. Such a stochastic model is then used to disambiguate documents’ words, during the indexing phase. The engine supports simple keyword-based queries, as well as natural language-based queries. The system is able to extend the domain knowledge, by means of a production-rules engine. The validation tests indicate that the system is able to extract concepts with good accuracy, even if the train set is small.

Knowledge Extraction from Natural Language Processing

SBATTELLA, LICIA;TEDESCO, ROBERTO
2012-01-01

Abstract

This chapter presents a model for knowledge extraction from documents written in natural language. The model relies on a clear distinction between a conceptual level, which models the domain knowledge, and a lexical level, which represents the domain vocabulary. An advanced stochastic model (which mixes, in a novel way, two well-known approaches) stores the mapping between such levels, taking in account the linguistic context of words. Such a stochastic model is then used to disambiguate documents’ words, during the indexing phase. The engine supports simple keyword-based queries, as well as natural language-based queries. The system is able to extend the domain knowledge, by means of a production-rules engine. The validation tests indicate that the system is able to extract concepts with good accuracy, even if the train set is small.
2012
Methodologies and Technologies for Networked Enterprises
9783642317385
INF
File in questo prodotto:
File Dimensione Formato  
KnowledgeExtraction.pdf

Accesso riservato

: Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione 1.26 MB
Formato Adobe PDF
1.26 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/663459
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? ND
social impact