An Upperbound to the Performance of Ranked‐output Searching: Optimal Weighting of Query Terms Using a Genetic Algorithm

Abstract

This paper describes the development of a genetic algorithm (GA) for the assignment of weights to query terms in a ranked‐output document retrieval system. The GA involves a fitness function that is based on full relevance information, and the rankings resulting from the use of these weights are compared with the Robertson‐Sparck Jones F4 retrospective relevance weight. Extended experiments with seven document test collections show that the ga can often find weights that are slightly superior to those produced by the deterministic weighting scheme. That said, there are many cases where the two approaches give the same results, and a few cases where the F4 weights are superior to the ga weights. Since the ga has been designed to identify weights yielding the best possible level of retrospective performance, these results indicate that the F4 weights provide an excellent and practicable alternative. Evidence is presented to suggest that negative weights may play an important role in retrospective relevance weighting.

Genetic algorithms in relevance feedback: a second test and new contributions

“…They then obtain the mean value of these similarities (our functions fitness 3, fitness 4, and fitness 5, with the last two being practically the same as the first but using the inner product and the cosine, respectively). Robertson and Willett (1996) implement a GA that uses relevance feedback in learning the weights of the terms of a query. To calculate an individualÕs fitness they first determine the inner product of each query of the collection with each document of the database.…”

“…In recent years, there have appeared numerous applications of ''genetic algorithms'' (GAs) to information retrieval (Belew, 1989;Chen, 1995;Chen, Chung, & Ramsey, 1998;Chen & Iyer, 1998;Cordo on, Moya, & Zarco, 2000Gordon, 1988aGordon, ,b, 1991Horng & Yeh, 2000;Kraft, Petry, Buckles, & Sadasivan, 1994Lo opez-Pujalte, 2000;Lo opez-Pujalte, Guerrero Bote, & Moya Anego on, 2002a,b;Martı ın-Bautista, 2000;Martı ın-Bautista, Vila, & Larsen, 1999;Raghavan & Agarwal, 1987;Raghavan & Birchard, 1979;Robertson & Willett, 1994, 1996Sanchez, 1994;Sanchez, Miyano, & Brachet, 1995;Sanchez & Pierre, 1994;Smith & Smith, 1997;Vrajitoru, 1997Vrajitoru, , 1998Yang & Korfhage, 1992, 1993, 1994. Most of these applications refer to ''relevance feedback'', as is indicated in a review of the topic (Cordo on, Moya, & Zarco, 1999).…”

A test of genetic algorithms in relevance feedback

“…These are the functions (sometimes with certain modifications to adapt them to our experiment) that in the present work we shall implement within the previously determined optimal feedback GA. Robertson and Willett (1996) implement a GA which learns the weights of the query terms by means of relevance feedback. To calculate the individuals' fitness, they first find the inner product of each query of the collection with each document of the database.…”

“…• Mutations: random mutations (Radcliffe, 1991;Michalewicz, 1995), and random mutations within the interval ½À1; 1, since the GA of the work of Robertson and Willett (1996) showed improved behaviour when negative weights were introduced.…”

Structure of morphologically expanded queries: A genetic algorithm approach

“…IQBE [6] is a process in which the user does not provide a query, but document examples and the algorithms induce the key concepts in order to find other relevant documents. Mainly two information retrieval problems have been tackled with GAs: assigning weights to the query terms [21,28,23,14,13], and selecting query terms. Let us consider a number of proposals in the latter case, the one on which we focus our work.…”