OSRT image
OSRT people image
OSRT projects image
OSRT papers image
OSRT links image
OSRT contact image

Miles Efron

Miles Efron is a recent graduate of the doctoral program in Information Science at the University of North Carolina-Chapel Hill. His research focuses on information retrieval, specifically empirical methods for concept extraction in automatic text indexing. With the OSRT he has helped develop the Open Source Metadata Framework, a metadata standard that will improve classification and resource discovery within the archives of the Linux Documentation Project and in other open source projects.

Projects:

Project Title Project Abstract

Papers:

Paper Title Paper Abstract
Amended Parallel Analysis for Optimal Dimensionality Estimation in Latent Semantic Indexing.
PDF

SILS Technical Report TR-2002-03.

Dimensionality Reduction and Statistical Model Building for Information Retrieval
Postscript File
Eigensystems and Attractors: Stable Points in Self-Organizing Information Spaces
Postscript File
Eigenvalue-based Estimators for Optimal Dimensionality Reduction in Information Retrieval
Complete Postscript version
Abstract: Postscript | PDF

Unpublished doctoral dissertation, University of North Carolina, Chapel Hill.

Latent Semantic Indexing (LSI) extends Salton's vector space model (VSM) of information retrieval, using dimensionality reduction to construct a statistical model of the relationships among the terms in a document collection. This dissertation pusues the viability of five statistical methods for estimating the optimal dimensionality of LSI systems.
The Emergence of Hypertextual Ecology from Individual Decisions.
PDF
A paper co-written with Steven Goodreau and Vishal Sanwalani at the 2002 Complex Systems Summer School at the Santa Fe Institute in Santa Fe, New Mexico.
Is it All About Connections? Factors Affecting the Performance of a Link-Based Recommender System
PDF File
A paper done with Gary Geisler, presented at the 2001 ACM SIGIR Workshop on Recommender Systems.
Using Dimensionality Reduction to Improve Similarity Judgments for Recommendation
PDF File
A paper done with Gary Geisler, presented at the joint DELOS/NSF Workshop on Personalisation and Recommender Systems in Digital Libraries.

Contact

mefron@metalab.unc.edu
CB #3456, Manning Hall
UNC-Chapel Hill
Chapel Hill, North Carolina 27599-3456

OSRT top logo