Meta-Path-Based Ranking with Pseudo Relevance Feedback on Heterogeneous Graph for Citation Recommendation

Abstract

The sheer volume of scholarly publications available online significantly challenges how scholars retrieve the new information available and locate the candidate reference papers. While classical text retrieval and pseudo relevance feedback (PRF) algorithms can assist scholars in accessing needed publications, in this study, we propose an innovative publication ranking method with PRF by leveraging a number of meta-paths on the heterogeneous bibliographic graph. Different meta-paths on the graph address different ranking hypotheses, whereas the pseudo-relevant papers (from the retrieval results) are used as the seed nodes on the graph. Meanwhile, unlike prior studies, we propose "restricted meta-path" facilitated by a new context-rich heterogeneous network extracted from full-text publication content along with citation context. By using learning-to-rank, we integrate 18 different meta-path-based ranking features to derive the final ranking scores for candidate cited papers. Experimental results with ACM full-text corpus show that meta-path-based ranking with PRF on the new graph significantly (p < 0.0001) outperforms text retrieval algorithms with text-based or PageRank-based PRF.

A survey of heterogeneous information network analysis

“…For example, Liu et al [125] develop a publication ranking method with pseudo relevance feedback by leveraging a number of meta paths on the heterogeneous bibliographic graph. Applying the tensor analysis, Li et al [21] propose HRank to simultaneously evaluate the importance of multiple types of objects and meta paths.…”

“…3(c) as an example, the constrained meta path AP A|P.L = Data M ining represents the co-author relation of authors in data mining field through constraining the label of papers with "Data Mining". Moreover, Liu et al [125] propose the concept "restricted metapath" which enables in-depth knowledge mining on the heterogeneous bibliographic networks by allowing restrictions on the node set. In addition, traditional HIN and meta path do not consider the attribute values on links, while weighted links are very common in real applications.…”

“…Along the X-axis, the network structure becomes more complex, and semantics information becomes richer along the Y-axis. For example, RankClus [17] NetClus [20] RankClus [17] CPIH [103] HeteRecom [34] MPBR [125] PIPL [37] GNetMine [91] UGES [59] PathSim [14] HeteSim [13] ComClus [38] HeProjI [80] PathSelClus [15] Evo−NetClus [180] HeteRec [133] DYNALINK [115] TRAIL [111] MNA [40] FSPG [181] Hete−CF [138] SemRec [39] HRank [21] Network structure Semantics exploration PCRW [50] classification similarity measure clustering link prediction recommendation ranking MCD [156] Fig. 12.…”

Feedback de relevância orientado a termos: um novo método para ordenação de resultados de motores de busca.

“…Alguns trabalhos afirmam obter bons resultados com a aplicação desta estratégia em contextos específicos (CAO et al, 2008;ZHAI, 2010;GROC;TANNIER, 2012;CARPINETO;ROMANO, 2012;BHATNAGAR;PAREEK, 2014;LIU et al, 2014). Esta estratégia geralmente melhora principalmente os valores de cobertura dos resultados, trazendo mais resultados mais parecidos com os primeiros resultados da busca.…”

“…As principais contribuições neste grupo estão relacionadas principalmente à seleção de quais documentos ou quais termos serão automaticamente selecionados e aplicados como feedback na geração de uma nova consulta (BHATNAGAR;PAREEK, 2014;CAO et al, 2008;LIU et al, 2014;PARAPAR;PRESEDO-QUINDIMIL;BARREIRO, 2014;YE;HUANG, 2014).…”