Publication venue recommendation in heterogeneous information networks

When a new paper is completed, choosing a good conference or journal in which to publish this new paper is of critical importance to all researchers. Authors often make their decision based on the topics suitability between the paper content and target venues, the likelihood of getting accepted into the venues, the publication history of the authors and other reasonable considerations. A good number of works do content-based analysis to match the topics of the paper and target venues. Such approaches often use full texts, abstracts along with other meta data. The main challenge, for this line of works, is to resolve topic ambiguity because many venues share similar topics and topics evolve over time. Another line of works are network-based approaches, which make recommendations using co-author networks and author-venue links in the bibliographic information networks. However, we have not yet seen a general framework that incorporates a broad range of both content-based features and network-based features, which are potentially capable of delivering more information to help solve the problem. In this thesis, we propose a general framework to automatically find appropriate venues for a new paper using a heterogeneous information network approach. First, meta path-based topological features are systematically extracted from the underlying bibliographic network. Then, a supervised model is used to learn the weights associated with different topological features in deciding the most suitable venues. Experiments on Microsoft Academic Graph(MAG) datasets show that our new approach consistently outperforms existing works by venue prediction accuracy. Results also show that not only topics information but authors' networks and publication history are important factors in the the problem of which venue to submit a new paper, we further tell from our experiments results that different authors have different influence over the final choice of venues.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2016

Author dc:creator

• Cai, Haoyan

Contributors dc:contributor

• Han, Jiawei

• venue recommendation
• heterogeneous information networks
• meta paths

Statement dc:rights

• Copyright 2016 Haoyan Cai

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/90825

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/90825