Experimental evaluation of parameter settings in calculation of hybrid similarities: effects of first- and second-order similarity, edge cutting, and weighting factors

Meyer-Brötz, Fabian; Schiebel, Edgar; Brecht, Leo

doi:10.1007/s11192-017-2366-2

Cited by 7 publications

(12 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Glänzel and Thijs [ 10 , 19 ] found that choosing the weights of citation-based similarity as 0.875 and 0.833 can obtain a balanced combination of two types of similarities. Meyer-Brötz et al [ 30 ] found that decreasing the textual weight can obtain a more coherent clustering result and set the weights as 0.5 or 0.6 to obtain the best result. In this study, Fig 4 shows the F1 measure and RI values with different values of parameter α .…”

Section: Dataset and Experimental Resultsmentioning

confidence: 99%

“…Based on the dataset used, first, the optimal parameters related to the weights of citation- and text-based similarities were analyzed according to the F1 measure and RI metrics. Different parameter sets have significant influences on the clustering results, which were also analyzed by Meng et al [ 20 ] and Meyer-Brötz et al [ 30 ]. Based on the optimal weight of the citation-based similarity, i.e., setting 0.55 as the weight of citation-based similarity in the hybrid similarity, four comparative groups with different parameter sets were discussed.…”

Section: Discussionmentioning

confidence: 97%

“…Second, the Louvain method can take each paper as a vertex so that the clustering analysis can be implemented directly [ 16 ]. Third, Colliander and Ahlgren [ 43 ] and Meyer-Brötz et al [ 30 ] found that it is easy to obtain considerably more similar distributions of clustering sizes when the Louvain method is applied to cluster documents with different parameter setting of similarities. Moreover, some existing studies have shown the efficient performance of the Louvain method, such as those by Liu et al [ 16 ] and Meng et al [ 20 ].…”

Section: Model and Methodologymentioning

confidence: 99%

“…Third, to solve the third issue, Glänzel and Thijs [ 10 ] proposed a linear combination of the angles of similarities. Following the proposal of this hybrid method, some studies conducted further analysis and proposed applications [ 19 , 28 – 30 ]. In this study, we also use the cosine angles underlying the linear combination of the citation- and text-based similarities, which are determined to represent the hybrid similarity between papers.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Hybrid self-optimized clustering model based on citation links and textual features to detect research topics

Wang

Zhang

et al. 2017

PLoS ONE

View full text Add to dashboard Cite

The challenge of detecting research topics in a specific research field has attracted attention from researchers in the bibliometrics community. In this study, to solve two problems of clustering papers, i.e., the influence of different distributions of citation links and involved textual features on similarity computation, the authors propose a hybrid self-optimized clustering model to detect research topics by extending the hybrid clustering model to identify “core documents”. First, the Amsler network, consisting of bibliographic coupling and co-citation links, is created to calculate the citation-based similarity based on the cosine angle of papers. Second, the cosine similarity is also used to compute the text-based similarity, which consists of the textual statistical and topological features. Then, the cosine angle of the linear combination of citation- and text-based similarity is considered as the hybrid similarity. Finally, the Louvain method is applied to cluster papers, and the terms based on term frequency are used to label clusters. To test the performance of the proposed model, a dataset related to the data envelopment analysis field is used for comparison and analysis of clustering results. Based on the benchmark built, different clustering methods with different citation links or textual features are compared according to evaluation measures. The results show that the proposed model can obtain reasonable and effective clustering results, and the research topics of data envelopment analysis field are also analyzed based on the proposed model. As different features are considered in the proposed model compared with previous hybrid clustering models, the proposed clustering model can provide inspiration for further studies on topic identification by other researchers.

show abstract

Section: Dataset and Experimental Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 97%

Section: Model and Methodologymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Hybrid self-optimized clustering model based on citation links and textual features to detect research topics

Wang

Zhang

et al. 2017

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…TF-IDF in Appendix A.2). Next, NetCulator calculates hybrid second-order similarities (See Appendix A.2) with severe edge cutting [24]. The latter improves the calculation speed by ignoring all weak edges that result from weak coupling between nodes.…”

Section: Bibliometric Analysismentioning

confidence: 99%

Techniques and Emerging Trends for State of the Art Equipment Maintenance Systems—A Bibliometric Analysis

et al. 2018

Self Cite

View full text Add to dashboard Cite

The increasing interconnection of machines in industrial production on one hand, and the improved capabilities to store, retrieve, and analyze large amounts of data on the other, offer promising perspectives for maintaining production machines. Recently, predictive maintenance has gained increasing attention in the context of equipment maintenance systems. As opposed to other approaches, predictive maintenance relies on machine behavior models, which offer several advantages. In this highly interdisciplinary field, there is a lack of a literature review of relevant research fields and realization techniques. To obtain a comprehensive overview on the state of the art, large data sets of relevant literature need to be considered and, best case, be automatically partitioned into relevant research fields. A proper methodology to obtain such an overview is the bibliometric analysis method. In the presented work, we apply a bibliometric analysis to the field of equipment maintenance systems. To be more precise, we analyzed clusters of identified literature with the goal to obtain deeper insight into the related research fields. Moreover, cluster metrics reveal the importance of a single paper and an investigation of the temporal cluster development indicates the evolution of research topics. In this context, we introduce a new measure to compare results from different time periods in an appropriate way. In turn, among others, this simplifies the analysis of topics, with a vast amount of subtopics. Altogether, the obtained results particularly provide a comprehensive overview of established techniques and emerging trends for equipment maintenance systems.

show abstract

Developmental tendencies in the academic field of intellectual property through the identification of invisible colleges

2018

View full text Add to dashboard Cite

Intellectual property became a relevant academic cross-disciplinary field in an international context with the demand for the global governance of knowledge. However, the degree of consolidation of cross-disciplinary academic communities is not clear. To determine how closely related are these communities, this paper proposes a mixed methodology to find invisible colleges in the production of intellectual property. Scientific articles from 1994 to 2016 were extracted from Web of Science, taking into account the signature of the agreement on Trade-Related Aspects of Intellectual Property Rights in the early 90's. A total of 1,580 papers were processed through bibliographic coupling network analysis. A special technique was applied, which combines algorithms of community detection and defines a population of articles through thresholds of shared references. To contrast the invisible colleges that emerged with the existence of formal institutional relations, a qualitative tracking of the authors was made with respect to their institutional affiliation, lines of research, and meeting places. Both methods show that the subjects of interest can be grouped into 13 thematic modules related to the intellectual property field. Even though most are related to Law and Economics, there are weak linkages between disciplines which could indicate the construction of a cross-disciplinary field.

show abstract

Experimental evaluation of parameter settings in calculation of hybrid similarities: effects of first- and second-order similarity, edge cutting, and weighting factors

Cited by 7 publications

References 27 publications

Hybrid self-optimized clustering model based on citation links and textual features to detect research topics

Hybrid self-optimized clustering model based on citation links and textual features to detect research topics

Techniques and Emerging Trends for State of the Art Equipment Maintenance Systems—A Bibliometric Analysis

Developmental tendencies in the academic field of intellectual property through the identification of invisible colleges

Contact Info

Product

Resources

About