Citation Graph Based Ranking in Invenio

Abstract

Abstract. Invenio is the web-based integrated digital library system developed at CERN. Within this framework, we present four types of ranking models based on the citation graph that complement the simple approach based on citation counts: time-dependent citation counts, a relevancy ranking which extends the PageRank model, a time-dependent ranking which combines the freshness of citations with PageRank and a ranking that takes into consideration the external citations. We present our analysis and results obtained on two main data sets: Inspire and CERN Document Server. Our main contributions are: (i) a study of the currently available ranking methods based on the citation graph; (ii) the development of new ranking methods that correct some of the identified limitations of the current methods such as treating all citations of equal importance, not taking time into account or considering the citation graph complete; (iii) a detailed study of the key parameters for these ranking methods.

Personalized Search

“…The primary motivation for this work is based on feedback from real users of CDS and previous work performed at CERN on search and personalized experiences for CDS [3] [45] [30] [19].…”

“…The latest contribution, the d-Rank project [44] [45] is a significant contribution to CDS, attempting to aggregate multiple ranking methods such as latest first, citations, tf-idf, and download history, in order to improve the ranking of search results. Other notable contributions to the Invenio search engine are the implementation of Solr by Patrick Glauner [17], evaluation of the d-Rank project performed by Olivier Canévet [8] and Citation graph-based ranking [30].…”

“…However, research has shown that citation counts alone are not sufficient for this approach. This is due to its inability to account for importance and creation date of the citation targets [30]. The work of Marian [30] presents four types of ranking models based on the citation graph that complements this approach.…”

“…This is due to its inability to account for importance and creation date of the citation targets [30]. The work of Marian [30] presents four types of ranking models based on the citation graph that complements this approach. One of the models presents a time-decaying weighting function, where the citation weight of a item is calculated as weight i = j,j→i e −w(tpresent−t j ) , where t present is the present time, t j is the publication date for item j, and (w ∈ (0, 1]) is a parameter that decides the aggressiveness of the ranking model.…”

A Novel Approach for Document Ranking in Digital Libraries using Extractive Summarization

“…There may be the case, the user need recent relevant papers instead of most cited papers. So, Ludmila Marian [7,8] proposed an extended version of Citation Count Algorithm i.e. Time Dependent Citation Count Ranking Algorithm.…”