Academic paper recommender system using multilevel simultaneous citation networks

Son, Jihoon; Kim, Seoung Bum

doi:10.1016/j.dss.2017.10.011

Cited by 79 publications

(32 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Haruna et al [32] also exploited single level paper-citation relation by mining hidden associations between paper of interest and its citations to find similar neighbors. Son and Kim [33] proposed a recommendation model that utilizes multilevel citation networks to find relevant papers. They considered both 'cites' and 'cited by' relationships between scientific papers beyond a single level to recommend high quality papers.…”

Section: B Non Priori User Profile Based Techniquementioning

confidence: 99%

“…Most of these works [3], [27]- [30], [33]- [39] either extract some paper contents or utilize graph model-based techniques for recommending papers. Apart from them, in this work we propose a neighbor-based collaborative approach for scientific paper recommendation that utilizes only easily obtained publicly available contextual citation relations information.…”

Section: B Non Priori User Profile Based Techniquementioning

confidence: 99%

See 1 more Smart Citation

A Collaborative Approach Toward Scientific Paper Recommendation Using Citation Context

2020

View full text Add to dashboard Cite

Researchers face difficulties in finding relevant papers to their research interest as the number of scientific publication is rapidly increasing on the web. Scientific paper recommenders have emerged as a leading solution to help researchers by automatically suggesting relevant and useful publications. Several approaches have been proposed on improving recommender systems. However, most existing approaches depend on priori user profiles, and thus they cannot recommend papers to new user. Furthermore, the existing approaches utilize non-public contextual information, and thus it cannot adequately find similarities between papers due to copyright restrictions. Also, the existing approaches consider only single level paper-citation relation to identify similarities between papers. Considering the above challenges, this paper presents a collaborative filtering based recommendation approach for scientific papers that does not depend on priori user profiles and which utilizes only public contextual information. Using citation context, we utilized 2-level paper-citation relations to find hidden associations between papers. The rational underlying this approach is that, two papers are co-occurred with same cited paper(s) and two papers are co-occurring with same citing paper(s) are significantly similar to some extent. To evaluate the performance of the proposed approach, publicly available datasets are used to conduct extensive experiments. The experimental results demonstrate that the proposed approach has significantly outperforms the baseline approaches in terms of precision, recall, F1, mean average precision, and mean reciprocal rank, which are commonly used information retrieval metrics. The novelty of this study is that, with the proposed approach, researchers are able to find relevant and useful publications over the internet regardless of their previous research experiences and research area.

show abstract

Section: B Non Priori User Profile Based Techniquementioning

confidence: 99%

Section: B Non Priori User Profile Based Techniquementioning

confidence: 99%

A Collaborative Approach Toward Scientific Paper Recommendation Using Citation Context

2020

View full text Add to dashboard Cite

show abstract

“…There are several research paper recommendation methods which focus on finding similarity between research articles [14]. These methods include: (1) collaborative filtering [15] (2) meta-data based [16], [17] (3) content-based [18], [19] (4) citation-based [9], [10], [20], [21] (5) multi-level citation network [13] (6) and 7user profile-based [22]- [24] approaches.…”

Section: Related Workmentioning

confidence: 99%

“…The resultant network is shown in Figure 3. The reason for using tenlevel network lies in the state-of-the-art literature, as it has been recommended as a reasonable size and using more than ten levels may include papers not related to the paper of interest [13]. The importance of each paper with the paper of interest is examined by applying four centrality parameters named betweenness centrality, eigenvector centrality, degree centrality and closeness centrality.…”

Section: Introductionmentioning

confidence: 99%

A Hybrid Approach Toward Research Paper Recommendation Using Centrality Measures and Author Ranking

et al. 2019

View full text Add to dashboard Cite

The volume of research articles in digital repositories is increasing. This spectacular growth of repositories makes it rather difficult for researchers to obtain related research papers in response to their queries. The problem becomes worse when a researcher with insufficient knowledge of searching research articles uses these repositories. In the traditional recommendation approaches, the results of the query miss many high-quality papers, in the related work section, which are either published recently or have low citation count. To overcome this problem, there needs to be a solution which considers not only structural relationships between the papers but also inspects the quality of authors publishing those articles. Many research paper recommendation approaches have been implemented which includes collaborative filtering-based, content-based, and citation analysis-based techniques. The collaborative filtering-based approaches primarily use paper-citation matrix for recommendations, whereas the content-based approaches only consider the content of the paper. The citation analysis considers the structure of the network and focuses on papers citing or cited by the paper of interest. It is therefore very difficult for a recommender system to recommend high-quality papers without a hybrid approach that incorporates multiple features, such as citation information and author information. The proposed method creates a multilevel citation and relationship network of authors in which the citation network uses the structural relationship between the papers to extract significant papers, and authors' collaboration network finds key authors from those papers. The papers selected by this hybrid approach are then recommended to the user. The results have shown that our proposed method performs exceedingly well as compared with the state-of-the-art existing systems, such as Google scholar and multilevel simultaneous citation network.

show abstract

“…The nodes in such academic networks are publications and the links are the citations between the articles that provide information of association. There is active research in this field [24] which notes the key motivation is that researchers can spend considerable amounts of time searching for the relevant research in order to not allocate time on topics already explored with similar approaches. Being able to find associative research is of importance since it is possible for research to be directed in areas already investigated and waste time as well as materials in research such as studies requiring expensive lab equipment.…”

mentioning

confidence: 99%

Regularized Simple Graph Convolution (SGC) for improved interpretability of large datasets

Pho

Mantzaris

2020

J Big Data

View full text Add to dashboard Cite

Classification of data points which correspond to complex entities such as people or journal articles is a ongoing research task. Notable applications are recommendation systems for customer behaviors based upon their features or past purchases and in academia labeling relevant research papers in order to reduce the reading time required. The features that can be extracted are many and result in large datasets which are a challenge to process with complex machine learning methodologies. There is also an issue on how this is presented and how to interpret the parameterizations beyond the classification accuracies. This work shows how the network information contained in an adjacency matrix allows improved classification of entities through their associations and how the framework of the SGC provide an expressive and fast approach. The proposed regularized SGC incorporates shrinkage upon three different aspects of the projection vectors to reduce the number of parameters, the size of the parameters and the directions between the vectors to produce more meaningful interpretations.

show abstract

Academic paper recommender system using multilevel simultaneous citation networks

Cited by 79 publications

References 23 publications

A Collaborative Approach Toward Scientific Paper Recommendation Using Citation Context

A Collaborative Approach Toward Scientific Paper Recommendation Using Citation Context

A Hybrid Approach Toward Research Paper Recommendation Using Centrality Measures and Author Ranking

Regularized Simple Graph Convolution (SGC) for improved interpretability of large datasets

Contact Info

Product

Resources

About