SULJE VALIKKO

Lakko vaikuttaa verkkokaupan toimituksiin ... LUE LISÄÄ

avaa valikko

"Reductive and Generative Approaches to Morphological Variation of Keywords in Monolingual Information Retrieval Acta Universita
37,80 €
Tampere University Press. TUP
Sivumäärä: 16090 sivua
Julkaisuvuosi: 2007 (lisätietoa)
Kieli: Englanti

This thesis concerns use of reductive and generative methods in management of keyword variation in information retrieval with best-match retrieval systems. The main results of the thesis are related to Finnish language IR, but we present also results of Swedish, German and Russian IR.

The main contributions of this study can be summed up as follows.

Our main contribution was to show that generative methods are also appropriate for IR in morphologically complex languages in a best-match retrieval environment. For Finnish we evaluated inflectional stem generation and its enhancements. We also created a new method, Frequent Case Generation, FCG, for inflectionally at least moderately complex languages and evaluated the method with four languages. The main idea of the method is to use only the most frequent nominal word forms of keywords as search terms. For three of the languages (Finnish, Swedish and German) the method was shown to yield good retrieval results when lemmatization was used as comparison. For Russian the results were inconclusive and the method should be re-evaluated with a better Russian collection. The method is based on skewness of word form distributions, and thus it is also expected to be applicable to other morphologically complex languages.

For Finnish best-match IR we have shown that besides lemmatization, also stemming, inflectional stem generation and its enhancements and most frequent case form generation of keywords yield good retrieval results when compared to the state-of-the-art, lemmatization. This broadens the spectrum of possible morphological tools for the handling of morphological variation of Finnish, which has been considered challenging in IR. As Finnish can be seen as a “worst case” language with respect to morphological variation, our results should also show the way to other languages having a fair degree of morphological variation.

Most of the methods evaluated in the study are shown to work for both long laboratory type queries and more realistic very short queries, which resemble user queries at least in the number of the keywords, although the research setting was a typical laboratory IR environment.



Tuotetta lisätty
ostoskoriin kpl
Siirry koriin
LISÄÄ OSTOSKORIIN
Tilaustuote | Arvioimme, että tuote lähetetään meiltä noin 5-8 arkipäivässä
"Reductive and Generative Approaches to Morphological Variation of Keywords in Monolingual Information Retrieval Acta Universita
Näytä kaikki tuotetiedot
ISBN:
9789514470875


Toimitusehdot


Asiakaspalvelu


YHTEYSTIEDOT


SEURAA MEITÄ

Booky.fi | Kotimainen kirjakauppasi netissä

Löydä seuraava lukuelämyksesi meiltä. Valikoimassamme ovat kaikki kotimaiset kirjat sekä noin 25 miljoonaa ulkomaista teosta.
Toimitamme tilaukset maailmanlaajuisesti!



Tietosuojaseloste