Developing a novel recommender network-based ranking mechanism for library book acquisition
| Pages | 50-68 |
| DOI | https://doi.org/10.1108/EL-06-2015-0094 |
| Date | 06 February 2017 |
| Published date | 06 February 2017 |
| Author | Fan Wu,Ya-Han Hu,Ping-Rong Wang |
| Subject Matter | Information & knowledge management,Information & communications technology,Internet |
Developing a novel recommender
network-based ranking
mechanism for library
book acquisition
Fan Wu, Ya-Han Hu and Ping-Rong Wang
Department of Information Management, National Chung Cheng University,
Chiayi, Taiwan
Abstract
Purpose –Most academic libraries provide book recommendation services to enable readers to recommend
books to the libraries. To facilitate decision-making in book acquisition, this study aimed to develop a method
to determine the ranking of the recommended books based on the recommender network.
Design/methodology/approach –The recommender network was conducted to establish relationships
among book recommenders and their similar readers by using circulation records. Furthermore, social
computing techniques were used to evaluate the degree of representativeness of the recommenders and
subsequently applied as a criterion to rank the recommended books. Empirical studies were performed to
demonstrate the effectiveness of the proposed ranking system. The Spearman’s correlation coefcients
between the proposed ranking system and the ranking obtained using reader circulation statistics were used
as performance measure.
Findings –The ranking calculated using the proposed ranking mechanism was highly and moderately
correlated to the ranking obtained using reader circulation statistics. The ranking of recommended books by
the librarians was moderately and poorly correlated to the ranking calculated using reader circulation
statistics.
Practical implications –The book recommender can be used to improve the accuracy of book
recommendations.
Originality/value –This study is the rst that considers the recommender network on library book
acquisition. The results also show that the proposed ranking mechanism can facilitate effective
book-acquisition decisions in libraries.
Keywords Academic libraries, Decision support systems, Recommendation, Book acquisition,
Recommender network
Paper type Research paper
Introduction
The purpose of a library is to provide the information its users require (Fagan, 2014;
Siguenza-Guzman et al., 2015). Given the current explosive growth in knowledge, buying
new books that suit the requirements of readers has become a critical task for libraries (Kao
et al., 2003;Kiilu and Kiilu, 2014;Wu et al., 2011). To enhance the appropriateness of books
acquired under the constraints of limited budgets, librarians must choose books carefully.
However, readers may not always nd the chosen books suitable because librarians cannot
fully gauge the requirements of readers (Ameen and Haider, 2007;Shieh and Wei, 2003;
Sitanggang et al., 2010). Consequently, the buying of suitable books is a key topic of study in
This research was supported in part by the Ministry of Science and Technology of the Republic of China
(grant number MOST 104-2410-H-194-070-MY3).
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0264-0473.htm
EL
35,1
50
Received 15 June 2015
Revised 31 December 2015
16 March 2016
Accepted 3 June 2016
TheElectronic Library
Vol.35 No. 1, 2017
pp.50-68
©Emerald Publishing Limited
0264-0473
DOI 10.1108/EL-06-2015-0094
library science (Ho et al., 2008;Kao et al., 2003;Wu, 2003). Libraries typically buy books
following a collection-development policy. Rashid (1990) indicated that a library should
apply decision-making techniques in book acquisition in an attempt to avert reader
dissatisfaction. Recently, considerable research has explored using various technologies in
book acquisition, such as statistical methods (Shieh and Wei, 2003;Wu et al., 2004),
mathematical modelling (Li et al., 2007), linear goal programming (Wise and Perushek, 1996),
goal programming (Wise and Perushek, 2000) and data mining (Kao et al., 2003;Sitanggang
et al., 2010;Wu, 2003;Wu et al., 2004).
The circulation records of a library contain reader names and the dates when books were
borrowed, reserved, renewed and returned. Thus, the circulation of books can be used as a
measure of appropriateness when acquiring books (Hu et al., 2012;Wise and Perushek, 2000).
However, circulation records represent only the requirements of readers in terms of the books
currently owned by a library and cannot reect reader requirements in terms of new books
to be acquired by the library. To capture what readers require regarding new books, Ameen
and Haider (2007) suggested that librarians should encourage readers to recommend the
books that should be purchased. Currently, numerous libraries provide a book
recommendation service to readers. However, because readers are likely to recommend
books by considering only their own requirements or interests, librarians cannot ensure that
these recommended books suit the requirements of other readers. Therefore, evaluating and
ranking the books in a recommended book list is a crucial decision-making step in book
acquisition.
The aim of this study is to develop a novel recommender network-based ranking scheme
for library book acquisition. Specically, the egocentric recommender network is used to
establish relationships between a book recommender and readers who borrowed the same
books as that recommender (who are referred to as related readers hereafter). In this network,
the strength of a link represents the similarity between recommenders and their related
readers and the number of links connecting a recommender to readers denotes the authority
(or representativeness) of the recommender. Based on the recommender network, the
importance of books can be calculated, as well as the degree of representativeness of
recommenders in each category to rank the books recommended for acquisition. In this
manner, this study addressed the following research questions:
RQ1. How can the recommender network-based ranking scheme accurately rank the
book list based on the recommender network?
RQ2. Can the proposed ranking scheme perform better than two existing ranking
approaches?
In experimental evaluation, recommender networks were built for several categories of
books in the case library. Based on these networks, a system was implemented to rank the
books recommended for acquisition. Two existing book acquisition policies (i.e. two ranking
approaches) used in the case library were considered as the baseline approaches: the rst one
ranked books by considering circulation statistics, while the second one ranked books by
librarian expertise and acquisition policies.
Literature review
Budget allocation in an academic library is one of the most important tasks to maintain the
quality of a library. Therefore, determining how to allocate appropriate budget on
purchasing new materials for an academic library became a research topic of interest in the
library science domain (Siguenza-Guzman et al., 2015). Previous studies have developed
various approaches to enhance the accuracy and efciency of material acquisition
51
Library book
acquisition
To continue reading
Request your trial