Document–document similarity approaches and science mapping: Experimental comparison of five approaches

Ahlgren, Per; Colliander, Cristian

doi:10.1016/j.joi.2008.11.003

Cited by 75 publications

(73 citation statements)

References 23 publications

(31 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…These approaches neglect one of the most important parts of a recommender system, which makes the approaches very similar to classic search, or related document search [338][339][340], where users provide search terms or an input paper, and receive a list of search results or similar papers. Of course, neither classic search nor related-document search are trivial tasks in themselves, but they neglect the user modeling process and we see little reason to label such systems as recommender systems.…”

Section: Neglect Of User Modelingmentioning

confidence: 99%

Research-paper recommender systems: a literature survey

et al. 2015

View full text Add to dashboard Cite

In the last 16 years, more than 200 research articles were published about research-paper recommender systems. We reviewed these articles and present some descriptive statistics in this paper, as well as a discussion about the major advancements and shortcomings and an overview of the most common recommendation concepts and approaches. We found that more than half of the recommendation approaches applied content-based filtering (55 %). Collaborative filtering was applied by only 18 % of the reviewed approaches, and graph-based recommendations by 16 %. Other recommendation concepts included stereotyping, item-centric recommendations, and hybrid recommendations. The content-based filtering approaches mainly utilized papers that the users had authored, tagged, browsed, or downloaded. TF-IDF was the most frequently applied weighting scheme. In addition to simple terms, n-grams, topics, and citations were utilized to model users' information needs. Our review revealed some shortcomings of the current research. First, it remains unclear which recommendation concepts and approaches are the most promising. For instance, researchers reported different results on the performance of contentbased and collaborative filtering. Sometimes content-based filtering performed better than collaborative filtering and Linnaeus University, Kalmar, Sweden sometimes it performed worse. We identified three potential reasons for the ambiguity of the results. (A) Several evaluations had limitations. They were based on strongly pruned datasets, few participants in user studies, or did not use appropriate baselines. (B) Some authors provided little information about their algorithms, which makes it difficult to re-implement the approaches. Consequently, researchers use different implementations of the same recommendations approaches, which might lead to variations in the results. (C) We speculated that minor variations in datasets, algorithms, or user populations inevitably lead to strong variations in the performance of the approaches. Hence, finding the most promising approaches is a challenge. As a second limitation, we noted that many authors neglected to take into account factors other than accuracy, for example overall user satisfaction. In addition, most approaches (81 %) neglected the user-modeling process and did not infer information automatically but let users provide keywords, text snippets, or a single paper as input. Information on runtime was provided for 10 % of the approaches. Finally, few research papers had an impact on research-paper recommender systems in practice. We also identified a lack of authority and long-term research interest in the field: 73 % of the authors published no more than one paper on research-paper recommender systems, and there was little cooperation among different co-author groups. We concluded that several actions could improve the research landscape: developing a common evaluation framework, agreement on the information to include in research papers, a stronger focus on non-accuracy aspect...

show abstract

Section: Neglect Of User Modelingmentioning

confidence: 99%

Research-paper recommender systems: a literature survey

et al. 2015

View full text Add to dashboard Cite

show abstract

“…At the document level, Small (1999a;1999b) developed a hierarchical map of science through a method that combined fractional counting of cited documents, single-and complete-linkage clustering and two-dimensional ordination based on a geometric triangulation process. Ahlgren and Colliander (2009) tested the performance of the complete-linkage clustering method for visualizing and classifying a set of 43 documents of the journal 'Information Retrieval' according to several similarity measures based on document text, coupling and a hybrid approach. A combination of graphic presentations and clustering was also adopted by Boyack et al (2011), yet they applied average-link clustering on several similarity matrices based on significant words extracted from the title, abstract and keywords of the Medical Subject Headings (MeSH) of over 2 million scientific articles gathered from the Medline database.…”

Section: Clustering and Information Visualizationmentioning

confidence: 99%

Visualization and analysis of SCImago Journal & Country Rank structure via journal clustering

Gómez-Núñez¹,

Vargas-Quesada²,

Chinchilla‐Rodríguez

et al. 2016

AJIM

View full text Add to dashboard Cite

PurposeThe objective was to visualize the structure of SCImago Journal & Country Rank (SJR) coverage of the extensive citation network of Scopus journals, examining this bibliometric portal through an alternative approach, applying clustering and visualization techniques to a combination of citation-based links. MethodologyThree SJR journal-journal networks containing direct citation, co-citation and bibliographic coupling links are built. The three networks were then combined into a new one by summing up their values, which were later normalized through geo-normalization measure. Finally, the VOS clustering algorithm was executed and the journal clusters obtained were labeled using original SJR category tags and significant words from journal titles. FindingsThe resultant scientogram displays the SJR structure through a set of communities equivalent to SJR categories that represent the subject contents of the journals they cover. A higher level of aggregation by areas provides a broad view of the SJR structure, facilitating its analysis and visualization at the same time. ValueThis is the first study using Persson's combination of most popular citation-based links (direct citation, co-citation and bibliographic coupling) in order to develop a scientogram based on Scopus journals from SJR. The integration of the three measures along with performance of the VOS community detection algorithm gave a balanced set of clusters. The resulting scientogram is useful for assessing and validating previous classifications as well as for information retrieval and domain analysis.

show abstract

“…Bibliographic coupling was firstly proposed by professor Kessler (1963) in MIT (Massachusetts Institute of Technology in American).Since then, the strength of coupling has been used in measuring the correlation among journals in a journal-based cross-citation network (Ni et al2013;Qiu and Liu 2014), in article cross-citation network (Ahlgren and Colliander 2009), and among authors in an author-based cross-citation network gradually (Zhao and Strotmann 2008;Rousseau and Zuccala 2004;Qiu and Dong 2013).…”

Section: Correlation Measurementioning

confidence: 99%

“…Ahlgren and Colliander (2009) studied the subject-classification from article level, and the similarity measures between different articles contain text similarity, bibliographical coupling and a combination of the two, then hierarchical clustering method was applied to construct a final classification. Chen et al (2010) made some quantitative analysis of scientific structure using the author cross-citation behavior and article cross-citation behavior respectively.…”

Section: Introductionmentioning

confidence: 99%

The study of subject-classification based on journal coupling and expert subject-classification system

Zhang

Liu

2016

Scientometrics

View full text Add to dashboard Cite

As the framework of scientific research, subject-classification plays an important role in the development of science. In order to combine the development of science with the current expert subject-classification system and further give a more appropriate description of scientific output analysis from subject level, We study the relationship between the natural science related sub-categories of Chinese library classification using objective computerized scientometrics, and give some modification to the first two level subjects of the existing Chinese library classification system. Taking Chinese Science Citation Database as our data source, this article studies the similarity of subjects based on journal coupling strength. Then we try to set up an improved subject-classification system whose top categories are relied on Chinese library classification system and sub-categories are the ensemble clustering result based on journal coupling measure. Further, in order to help identifying and interpreting the rationality of this improved classification system, we make use of some text mining methods, such as key words recognition and topic detection, to explain the cause of similarity between some subjects from the perspective of semantic. Our study shows that the improved subject-classification system constructed in this article not only conforms to previous experience and cognitive but also combines subject development knowledge.

show abstract

Document–document similarity approaches and science mapping: Experimental comparison of five approaches

Cited by 75 publications

References 23 publications

Research-paper recommender systems: a literature survey

Research-paper recommender systems: a literature survey

Visualization and analysis of SCImago Journal & Country Rank structure via journal clustering

The study of subject-classification based on journal coupling and expert subject-classification system

Contact Info

Product

Resources

About