Adaptive calculation of scores for fresh information retrieval

Minoru Uehara, Nobuyoshi Sato, Yoshifumi Sakai

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

In business, we need fresh information. In order to realize fresh information retrieval, we need not only to collect documents in a short time, but also to rank the results in the suitable order. However, conventional ranking methods are not suited for fresh information retrieval because they ignore temporal value of information. So, we have proposed the novel ranking method FTFIDF for fresh information retrieval. FTF·IDF extends TF·IDF by means of using FTF(fresh term frequency) instead of TF(term frequency). FTP differs from TF because FTP decreases as time goes. The speed of decreasing FTP is determined by the dumping factor. The dumping factor is sensitive against small changes of documents. So, we use a threshold to ignore such small changes. In some papers we published, we detect the optimal threshold manually. In this paper, we proposed an adaptive calculating method in order to detect threshold automatically. In this method, the optimal value is determined by iterating to test generated thresholds. In this paper, we describe our method and its evaluation.

Original languageEnglish
Title of host publicationProceedings - 11th International Conference on Parallel and Distributed Systems Workshops, ICPADS 2005
EditorsL. Barolli
Pages750-755
Number of pages6
DOIs
Publication statusPublished - 2005 Sep 1
Event11th International Conference on Parallel and Distributed Systems Workshops, ICPADS 2005 - Fukuoka, Japan
Duration: 2005 Jul 202005 Jul 22

Publication series

NameProceedings of the International Conference on Parallel and Distributed Systems - ICPADS
Volume1
ISSN (Print)1521-9097

Other

Other11th International Conference on Parallel and Distributed Systems Workshops, ICPADS 2005
CountryJapan
CityFukuoka
Period05/7/2005/7/22

Keywords

  • Distributed search engine
  • Fresh information retrieval
  • Ranking

ASJC Scopus subject areas

  • Hardware and Architecture

Fingerprint Dive into the research topics of 'Adaptive calculation of scores for fresh information retrieval'. Together they form a unique fingerprint.

Cite this