Janssens, F., Zhang, L., De Moor, B., & Glänzel, W. (2009). Hybrid clustering for validation and improvement of subject-classification schemes. Information Processing & Management, 45(6), 683-702.

Janssens, F., Zhang, L., De Moor, B., & Glänzel, W. (2009). Hybrid clustering for validation and improvement of subject-classification schemes. Information Processing & Management, 45(6), 683-702.



科學認知映射(cognitive mapping of science)能將科學的結構(the structure of science)加以視覺化，起初應用於資訊服務(information services)、後來發現也可將其應用在科學政策(science policy)與研究評鑑(research evaluation)，現在則有越來越多將其應用在發現新興與正在融合中的領域以及主題劃分(subject delineation)的改善。

科學認知映射主要可分為依據引用資訊、依據文本以及混合上述兩種資訊等三種方法。本研究利用文本與引用資訊混合的方法，將2002-2006年Web of Science資料庫內的期刊進行集群，以集群結果產生的認知映射，檢驗目前的期刊主題分類架構，如果可行的話，也將提出改善方式。

本研究首先評估ESI (Essential Science Indicators) 的22個領域主題分類架構，並且將其視覺化。圖1的左右分別是以交互引用與文本方式測量22個ESI領域的Silhouette值，從圖上發現生物學及生物化學(#2)、臨床醫學(#4)、工程學(#7)、植物及動物科學(#19)以及社會科學(#21)等領域上的期刊並沒有足夠好的一致性(coherent)。



從詞語的TF-IDF可以找出每個領域的描述詞語，而這裡也可發現不少領域的描述語互有重疊，例如工程學(#7)與電腦科學(#5)、化學(#3)與材料科學(#11)、植物及動物科學(#19)與環境/生態學(#8)，以及生物學及生物化學(#2)、分子生物學及遺傳學(#14)與臨床醫學(#4)等等，此外，從描述社會科學(#21)的詞語也可以了解這個領域高度的異質性(heterogeneity)。


圖2則是以Pajek畫出22個ESI領域的結構圖，圖上也可發現生物學及生物化學(#2)與臨床醫學(#4)、化學(#3)與材料科學(#11)、電腦科學(#5)與工程學(#7)、環境/生態學(#8)與植物及動物科學(#19)等領域之間有很強的關連。

然後將約8300種期刊利用餘弦相似法及Wade的凝聚式階層集群演算法(Wade's  agglomerative hierarchical cluster algorithm)進行集群，再比較集群結果與分類架構。值得說明的是文字部分的資訊可以提供集群結果的標示(labelling)，而引用部分則可產生交互引用圖(cross-citation graph)提供視覺化，並且輸入PageRank演算法以決定代表性期刊。

決定集群的數目可以根據集群結果的品質，而集群品質有內在或外在驗證測量(internal or external validation measures)等兩種評估方式。內在驗證只考慮資料與集群的統計特性，例如dendrogram、Silhouette值與模組性(modularity)等；外在驗證需要將集群結果與一個已知的劃分標準進行比較，例如計算兩者間的Jaccard相似性 (Jaccard similarity)。本研究以dendrogram的視覺化方式將期刊首先分為三大群，再分為七群，最後分為22群，三大群約等於自然與應用科學(生物學、農學及環境科學；物理、化學及工程學；數學及電腦科學)、醫學與社會科學以及人文學。以TF-IDF描述語來看，七個群組中有三個屬於自然與應用科學，兩個是生命科學(生物科學及生物醫學與臨床、實驗醫學及神經科學)與兩個是社會科學以及人文學(經濟學、商學及政治學與心理學、社會學及教育學)

圖6是22個群組的結構圖，圖上可看到屬於社會科學以及人文學的群組(#1、#6、#14及#22與#9、#11及#21)、地理學、環境科學、生物學及農學(#2、#15及#19)、物理、化學及工程學(#4、#20及 #5)、數學及電腦科學(#8及#18)、生物科學及生物醫學(#3、#13及#16)與臨床、實驗醫學及神經科學(#7、#10、#12及#17)。



表3比較22個ESI領域與本研究利用引用、文本以及混合等方法產生的22個群組的集群品質，可以發現混合引用及文本資料在各種指標上幾乎都有最好的表現。


圖8是利用Jaccard指標比較集群結果與ESI架構的一致性(concordance)。



最後，本研究並分析期刊轉移(migration)的情形，也就是期刊不屬於原本ESI架構的領域所對應的集群，而被分配到另一個不同集群的現象，好的轉移(Good migration)能使分類的一致性增加，也就是Silhouette值或是模組性增加。本研究希望利用這個現象，從集群與領域的一致性的基礎上提出改善目前期刊主題分類架構的方法。

The main bibliometric techniques are characterised by three major approaches, particularly the analysis of citation links (cross-citations, bibliographic coupling, co-citations), the lexical approach (text mining), and their combination.

A hybrid text/citation-based method is used to cluster journals covered by the Web of Science database in the period 2002–2006. The objective is to use this clustering to validate and, if possible, to improve existing journal-based subject-classification schemes.

In a first step, the 22-field subject-classification scheme of the Essential Science Indicators (ESI) is evaluated and visualised. In a second step, the hybrid clustering method is applied to classify the about 8300 journals meeting the selection criteria concerning continuity, size and impact.

The hybrid method proves superior to its two components when applied separately. The choice of 22 clusters also allows a direct field-to-cluster comparison, and we substantiate that the science areas resulting from cluster analysis form a more coherent structure than the ‘‘intellectual” reference scheme, the ESI subject scheme.

Moreover, the textual component of the hybrid method allows labelling the clusters using cognitive characteristics, while the citation component allows visualising the cross-citation graph and determining representative journals suggested by the PageRank algorithm.

Finally, the analysis of journal ‘migration’ allows the improvement of existing classification schemes on the basis of the concordance between fields and clusters.

The history of cognitive mapping of science is as long as the history of computerised scientometrics itself. While the first visualisations of the structure of science were considered part of information services, i.e., an extension of scientific review literature (Garfield, 1975, 1988), bibliometricians soon recognised the potential value of structural science studies for science policy and research evaluation as well. At present, the identification of emerging and converging fields and the improvement of subject delineation are in the foreground.

The main bibliometric techniques are characterised by three major approaches, particularly the analysis of citation links (cross-citations, bibliographic coupling, co-citations), the lexical approach (text mining), and their combination.

For instance, clustering based on co-citation and bibliographic coupling has to cope with several severe methodological problems. This has been reported, among others by Hicks (1987) in the context of cocitation analysis and by Janssens, Glänzel, and De Moor (2008) with regard to bibliographic coupling. One promising solution is to combine these techniques with other methods such as text mining (e.g., combined co-citation and word analysis: Braam, Moed, & Van Raan, 1991a; combination of coupling and co-word analysis: Small (1998); hybrid coupling-lexical approach: Janssens, Glänzel, & De Moor, 2007; Janssens et al., 2008).

Jarneving (2005) proposed a combination of bibliometric structure–analytical techniques with statistical methods to generate and visualise subject coherent and meaningful clusters. His conclusions drawn from the comparison with ‘intellectual’ classification were rather sceptical.

Despite several limitations, which will be discussed further in the course of the present study, cognitive maps proved useful tools in visualising the structure of science and can be used to adjust existing subject-classification schemes even on the large scale as we will demonstrate in the following.

The main objective of this study is to compare (hybrid) cluster techniques for cognitive mapping with traditional ‘intellectual’ subject-classifications schemes.

In a first study by authors related to the current work, the pilot study of Glenisson, Glänzel, & Persson (2005), further extended and confirmed by Glenisson, Glänzel, Janssens et al. (2005), full-text analysis and traditional bibliometric methods were serially combined to improve the efficiency of the individual methods. It was clear that clusters found through application of text mining provided additional information that could be used to extend and explain structures found by bibliometric methods, and vice versa. However, the integration was still limited to serial combination.

All textual content was indexed with the Jakarta Lucene platform (Hatcher & Gospodnetic, 2004) and encoded in the Vector Space Model using the TF-IDF weighting scheme reviewed by Baeza-Yates & Ribeiro-Neto (1999). Stop words were neglected during indexing and the Porter stemmer was applied to all remaining terms from titles, abstracts, and keyword fields. The resulting term-by-document matrix contained nine and a half million term dimensions (9,473,061), but by ignoring all tokens that occurred in one sole document, only 669,860 term dimensions were retained. Those ignored terms with document frequency equal to one are useless for clustering purposes.

The dimensionality was further reduced from 669,860 term dimensions to 200 factors by Latent Semantic Indexing (LSI) (Berry, Dumais, & O’brien, 1995; Deerwester, Dumais, Furnas, Landauer, & Harshman, 1990), which is based on the Singular Value Decomposition (SVD).

Text-based similarities were calculated as the cosine of the angle between the vector representations of two papers (Salton & Mcgill, 1986).

For simplicity and efficiency, the method used to summarise the subject of a field or cluster is based on selecting the terms with the highest mean TF-IDF weights over all journal papers in the field or cluster, where the IDF factor is calculated on the complete term-by-paper matrix (more than six million papers).

For example, Treeratpituk and Callan (2006) automatically select and assign a few concise labels to hierarchical clusters by combining statistical features from the cluster, parent cluster, and a corpus of general English into a descriptive score.

Geraci, Maggini, Pellegrini, and Sebastiani (2008) label clusters by combining intra-cluster and inter-cluster term extraction based on a variant of the information gain measure, and by looking within the titles of Web pages for the substring that best matches the selected top-scoring words.

The similarities Sij used for clustering were found by calculating the cosine of the angle between the pair of vectors containing all symmetric journal cross-citation values between the two respective journals (i and j) and all other journals (i.e., row or column of the matrix C):

The journal cross-citation graph is also analysed to identify important high-impact journals. We use the PageRank algorithm (Brin & Page, 1998) to determine representative journals in each cluster. Besides, the graph can also be used to evaluate the quality of a clustering outcome.

In order to subdivide the journal set into clusters we used the agglomerative hierarchical cluster algorithm with Ward’s method (Jain & Dubes, 1988).

In general, the number of clusters is determined by comparing the quality of different clustering solutions based on various numbers of clusters. Cluster quality can be assessed by internal or external validation measures. Internal validation solely considers the statistical properties of the data and clusters, whereas external validation compares the clustering result to a known gold standard partition

This compound strategy encompasses observation of a dendrogram, text- and citation-based mean Silhouette curves, and modularity curves. Besides, the Jaccard similarity coefficient is used to compare the obtained results with an intellectual classification scheme.

Up to a multiplicative constant, modularity measures the number of intra-cluster citations minus the expected number in an equivalent network with the same clusters but with citations given at random. Intuitively, in a good clustering there are more citations within (and fewer citations between) clusters than could be expected from random citing.

In Fig. 8, we use the Jaccard index to compare each cluster with every field from the intellectual ESI classification, in order to detect the best-matching fields for each cluster.

Nowadays two ISI systems are widely used, in particular, the ISI Subject Categories, which are available in the JCR and through journal assignment in the Web of Science as well, and the Essential Science Indicators (ESI).

While the first system assigns multiple categories to each journal and is too fine grained (254 categories) for comparison with cluster analysis, the ESI scheme is forming a partition (with practically unique journal assignment) and the 22 fields are large enough. ... This subject-classification scheme is in principle based on unique assignment; only about 0.6% of all journals were assigned to more than one field over a 5-year period.

Fig. 1 presents the evaluation of the 22 ESI fields based on the cross-citation- (left) and text-based (right) Silhouette values (see Section 3.3.3). Since the ESI fields form a partition, this approach allows to evaluate their consistency as if the fields were results of a clustering procedure. Multi-, interand cross-disciplinarity of journals can certainly affect the results.

Several fields seem not to be coherent enough from both perspectives (i.e., the cross-citation and textual approach). Above all, the Silhouette values of field #2 (Biology and Biochemistry), #4 (Clinical Medicine), #7 (Engineering), #19 (Plant and Animal Science) and #21 (Social Sciences) substantiate that at least five of the 22 fields are not sufficiently coherent.

Simultaneously to the above validation, the textual approach also provides the best TF-IDF terms – out of a vocabulary of 669,860 terms – describing the individual fields. These terms are presented in Table 2. Although these terms already provide an acceptable characterisation of the topics covered by the 22 fields, considerable overlaps are apparent between pairs of fields, respectively: Engineering (#7) and Computer Science (#5), Chemistry (#3) and Materials Science (#11), Plant and Animal Science (#19) and Environment/Ecology (#8), as well as Biology and Biochemistry (#2), Molecular Biology and Genetics (#14) and Clinical Medicine (#4). In addition, the terms characterising the social sciences (#21) reflect a pronounced heterogeneity of the field.

The structural map of the 22 ESI fields based on cross-citation links is presented in Fig. 2. For the visualisation we used Pajek (Batagelj & Mrvar, 2003). The network map confirms the strong links we have found based on the best terms between fields #2 and #14, #3 and #11, #5 and #7, and #8 and #19, respectively.

In Table 3 we compare the quality of the partition of 22 ESI fields with the quality of the 22 clusters resulting from citation-based, text-based and hybrid clustering.

The cluster dendrogram shows the structure in a hierarchical order (see Fig. 4). We visually find a first clear cut-off point at three clusters, a second one around seven, and 22 clusters also seemed to be an acceptable/appropriate number.

The number of three clusters results in an almost trivial classification. Intuitively, these three high-level clusters should comprise natural and applied sciences, medical sciences, and social sciences and humanities.

The solution comprising of seven clusters results in a non-trivial classification. The best TF-IDF terms (see Table 5) show that three of these clusters represent the natural/applied sciences, whereas two classes each stand for the life sciences and the social sciences and humanities. This situation is also reflected by the cluster dendrogram in Fig. 4. A closer look at the best TF-IDF terms reveals that the social-sciences cluster (#1 of the 3-cluster solution) is split into the cluster #1 (economics, business and political science) and #6 (psychology, sociology, education), the life-science cluster (#3 in the 3-cluster scheme) is split into clusters #3 (biosciences and biomedical research) and #7 (clinical, experimental medicine and neurosciences) and, finally, the sciences cluster #2 of the 3-cluster scheme is distributed over three clusters in the 7-cluster solution, particularly, the cluster comprising biology, agriculture and environmental sciences (#2), physics, chemistry and engineering (#4) as well as mathematics and computer science (#5).

The social-sciences and humanities clusters form two groups that are each strongly interlinked; one consists of clusters #1, #6, #14 and #22 with focus on humanities, economics, business, political and library science, the other one comprises #9, #11 and #21 with sociology, education and psychology. This is in line with the hierarchical structure shown in Fig. 4. These two groups correspond to the two social-sciences clusters in the 7-cluster solution (cf. Section 4.4).

On the basis of the most important TF-IDF terms (see Table 6) we can assign clusters #2, #15 and #19 to geosciences, environmental science, biology and agriculture, which, in turn, form a larger group corresponding to the first of the three ‘‘megaclusters” in the 7-cluster solution.

These science clusters form two groups, #4, #20 and #5 form one group of chemistry, physics and engineering, while #8 and #18 form the third group comprising mathematics and computer science.

Here we have a biomedical and a clinical group. These two groups are in line with the hierarchical structure of the dendrogram in Fig. 4 but less clearly distinguished in the graphical network presentation (Fig. 6). Nonetheless, the terms provide an excellent description for at least some of the medical clusters: cluster #7 stands for the neuro- and behavioral sciences, #3 for bioscience, #10 for the clinical and social medicine, #13 microbiology and veterinary science, #12 non-internal medicine, #16 hematology and oncology and #17 cardiovascular and respiratory medicine. According to the dendrogram clusters 3, 13, 16 and clusters 7, 10, 12, 17 form one larger cluster each. On the basis of the best terms, we can characterise these groups as the bioscience–biomedical and the clinical and neuroscience group, respectively.

In this subsection we compare the structure resulting from the hybrid clustering with the ESI subject classification. This comparison is based on the centroids of the clusters and fields. The centroid of a cluster or field is defined as the linear combination of all documents in it and is thus a vector in the same vector space. For each cluster and for each field, the centroid was calculated and the MDS of pairwise distances between all centroids is shown in Fig. 7.

In Fig. 8, we use the Jaccard index to determine the concordance between our clustering solution and the ESI Scheme by comparing each cluster with every field, in order to detect the best-matching fields for each cluster. The darker a cell in the matrix, the higher the Jaccard index, and hence the more pronounced the overlap between the corresponding cluster and ESI field.

If clustering algorithms are adjusted or changed, one can observe the following phenomenon. Some units of analysis are leaving clusters they formerly belonged to and end up in different clusters. This phenomenon is called ‘migration’. We can distinguish between ‘good migration’ and ‘bad migration’.

‘Good migration’ is observed if the goodness of the unit’s classi- fication improves, otherwise we speak about ‘bad migration’. We can also apply this notion of migration to the comparison of clustering results with any reference classification. In the following we will use the ESI scheme as reference classification.

Out of 8305 journals under study, there were more than one third, namely, 3204 journals that were not assigned to the cluster which best matches their ESI field. As already mentioned above, we call these journals ‘migrated journals’.

‘Good migrations’ are observed if journals improved their Silhouette values after migration. Based on their titles and scopes (not shown), apparently they should indeed be assigned to the cluster to which they have moved.

Although the Silhouette and modularity values substantiate a more coherent structure of the hybrid clustering as compared with the ESI subject scheme, not all clusters are of high quality. Problems have been found, for instance, in clusters #1 and #12 where interdisciplinarity and strong links with other clusters distort the intra-cluster coherence.

Thijs, B., Zhang, L., & Glänzel, W. (2015). Bibliographic coupling and hierarchical clustering for the validation and improvement of subject-classification schemes. Scientometrics, 105(3), 1453-1467.

Thijs, B., Zhang, L., & Glänzel, W. (2015). Bibliographic coupling and hierarchical clustering for the validation and improvement of subject-classification schemes. Scientometrics, 105(3), 1453-1467.

Thijs, B., Zhang, L., & Glänzel, W. (2013, January). Bibliographic coupling and hierarchical clustering for the validation and improvement of subject-classification schemes. In Proceedings of ISSI (pp. 237-249).

本研究利用書目耦合(bibliographic coupling)資訊將收錄於Web of Science資料庫內的期刊分群，利用二次相似性(second order similarities)改善書目耦合資訊產生的相似性矩陣過於疏鬆的問題，再以dendrogram及silhouette測量等資訊決定由Ward的凝聚法(Ward’s agglomeration method)產生集群的數目，最後比較集群結果與Glänzel & Schubert (2003)提出的以期刊為基礎的主題分類架構(the journal-based subject-classification scheme)，了解兩者間的對應情形，並決定集群的命名。

使用書目耦合的優點是因為需要的資料都已呈現在論文或資料庫上，計算論文(即本文所謂的publications)和期刊間的連結不會有延遲，並且建立連結後將會持續保持一致，不隨時間改變。然而，如同其他使用引用資訊的方法，相關的論文或期刊並無法共同具有所有的參考文獻，在相似性矩陣上無可避免地會有大量的0出現(Janssens, 2007; Janssens et al., 2008)，產生極為大量的單一個體(singletons)，並影響後續集群分析的品質。過去解決這個問題的一種做法是將引用文獻與詞語相似性的結果混合，例如Janssens et al. (2008)；另一種的做法是以二次相似性產生相似性矩陣，例如Janssens (2007)、Ahlgren & Colliander (2009)與 Thijs et al. (2013)等研究。

本研究即是利用二次相似性對Web of Science資料庫內的期刊分群，分析的期刊為2006到2009年間出版100筆論文或以上的期刊，共8282種。作法如下：

1. 以Salton提出的餘弦測量法(cosine measure)產出一次相似(first order similarity)，然後以一次相似矩陣再次進行餘弦測量法，產生二次相似性。經過二次相似性的計算後，有10種期刊沒有與其他期刊相連，因此加以移除，因此剩餘的期刊網路上共有8272種。

2. 以Ward的凝聚法產生階層集群，並利用dendrogram及silhouette測量推測可能的集群數目，在本研究由上到下分別有6、14及24種集群數目。silhouette測量方法是對每一種期刊計算一個介於-1和1之間的silhouette數值，正值代表該期刊被分配到適當的集群中。然後將期刊依照其群集分組並以silhouette數值大小排序，產生的圖形可以表示各集群的分群品質，如果在正值的部分有較大的面積，換言之，這個集群具有較多的期刊具有適當的分配，則代表有較好的集群劃分結果。

為了檢驗分群的效果，產生的集群結果與Glänzel & Schubert (2003)提出的以期刊為基礎的主題分類架構進行比較，並且在每一個集群上找出具有代表性的核心期刊(core journals)來分析結果的每個期刊集群。圖五是以網絡來表示集群之間的關係，在圖上可以發現藝術與人文(Arts and Humanities)遠離其他集群，神經科學和行為科學(Neurosciences & Behaviour)介於社會科學和生命科學之間，化學則處於生物科學(Biosciences)、醫學和物理學的中間，較特別的是 一般、區域與社區議題(General,Regional and Community Issues)與生命科學之間有很強的連結。



此外，並且計算14個集群與Glänzel & Schubert (2003)的15個學科(排除Multidiscipline下的期刊)之間的Jaccard指標，呈現為表三。

除了與Glänzel & Schubert (2003)的期刊架構比較以外，本研究也將分群的結果與 ESI (essential science indicators)的類別比較，然而卻發現ESI的劃分與本研究分類結果的結構並不一致。

An attempt is made to apply bibliographic coupling to journal clustering of the complete Web of Science database. Since the sparseness of the underlying similarity matrix proved inappropriate for this exercise, second-order similarities have been used.

Cluster labelling was made on the basis of the about 70 subfields of the Leuven-Budapest subject-classification scheme that also allowed the comparison with the existing two-level journal classification system developed in Leuven. The further comparison with the 22 field classification system of the Essential Science Indicators does, however, reveal larger deviations.

The issue of subject classification and the creation of coherent journal sets has been a major topic in our field since the seventies (see e.g., Narin et al., 1972; Narin, 1976).

The development of computerised methods and the availability of large datasets have shifted the attention from mapping small or single disciplines to the generation of global science maps (Garfield, 1998).

Jarneving (2005) applied bibliographic coupling to map and to analyse the structure of an annual volume of the Science Citation Index.

Janssens et al. (2008; 2009) used a combination of cross-citations and a lexical approach to map journals. Zhang et al. (2010) validated this approach.

The advantage of bibliographic coupling is that there is no delay for the calculation of the link between publications or journals as all data needed are present upon publication or indexing in the database. This also means that link between documents, once established will remain constant over time.

This disadvantage is a result of the very sparse nature of the link matrix (Janssens, 2007; Janssens et al., 2008). The overwhelming number of document pairs does not share any reference at all and thus a large number of zeros occur in the similarity matrix. This deteriorates the quality of the subsequent clustering and may result in an unrealistic large number of singletons (cf. Jarneving, 2005).

As cross-citation data suffers from the same problem, Janssens et al. (2008) introduced a hybrid approach, where they combined citation-based with lexical similarities.

Another solution to overcome the sparseness problem is the use of second order similarities (Janssens, 2007; Ahlgren & Colliander, 2009; Thijs et al., 2013).

A set of journals was compiled from the Web of Science database (SCI-Expanded, SSCI and AHCI). All journals covered in this database between 2006 and 2009 with at least 100 publications in this period are taken into account. This resulted in a set of 8282 journals.

To express the strength of a link between two journals we calculated a first order similarity based on Salton’s cosine measure. The mathematical derivation and interpretation of this similarity measure in the framework of a Boolean vector space model can be found in (Sen & Gan, 1983; Glänzel & Czerwon, 1996).

As bibliographic coupling tends to produce very sparse similarity matrices we applied a second order similarity to reduce this effect. While the first-order similarity is based on the angle between two reference vectors, the second-order similarity is calculated as the cosine of the angle of two vectors holding the first order similarity between two journals.

After the calculation of the second-order similarities, ten journals were removed from the set as they appeared to be singletons without any link to the other journals in the set. The network thus included 8272 journals in total.

Hierarchical clustering with Ward’s agglomeration method was used to create a hard clustering of all the journals.

This method does not provide any automated optimum number of clusters so that the decision was made on the basis of the dendrogram and the silhouette statistics (Rousseeuw, 1987).

Three different levels were chosen. The dendrogram holds strong arguments for a six cluster partitioning while the silhouette plot shows a first peak at 7 clusters. For the highest hierarchical level in the following analysis we use the six cluster solution. At a lower level, the silhouette plot suggests the solutions with 14 and 24 clusters, respectively.

For the evaluation of the specific cluster solution we can rely on the silhouette graphs presented in Figure 4. Each graph presents the silhouette values of the journals in the respective cluster. For each journal a silhouette value is calculated. These values range between 1 and -1 where positive values indicate an appropriate clustering of the journals. Journals are grouped by cluster and ordered from highest silhouette value to lowest. As a consequence the graph gives a good profile of the quality of each cluster. A larger area at the positive side of the vertical axis thus represents a better partitioning.

In order to find an acceptable solution, we decided to use the journal-based subject-classification scheme developed in Leuven (Glänzel & Schubert, 2003). This solution proved most advantageous since both clustering and classification scheme are based on journal assignment. Table 1 presents the hierarchical structure of the three level partitioning. For each cluster the number of journals is mentioned. The labels for the higher levels can be deduced from the lowest level. These labels are taken from the Leuven classification system . The label from the most prominent subject category has been assigned to the corresponding cluster.

Another way to describe the cluster is by using core journals. This notion can be analogously defined as core documents introduced by Glänzel & Czerwon (1996) and extended by Glänzel & Thijs (2011).

In this particular application, a core journal can be identified as journal with at least n links with other journals of at least a given strength r on the second order similarity measure. For the identification of core journals in each cluster we set the number of strong links to at least half the set of journals in the cluster.

As we are using second order similarities this choice is not unreasonable. The value of the strength is chosen such that 12 journals within each cluster comply with both criteria. This means that for more dense clusters the choice of appropriate r-value is higher than in clusters where the journals are not as strongly linked.

Above all, chemistry is at each level a separate cluster. One might expect that at the highest level, chemistry is merged with Physics but we found different patterns.

The second noteworthy observation concerns cluster 17 (Public Health & Nursing). This is a cluster within the ‘Psychology – Neuroscience’ cluster at the highest, six-cluster level. In other partitions or subject classification systems this is attributed to Non-Internal Medicine.

To visualise relations between the 24 clusters we created an additional map. Figure 5 shows these relations.

Despite these multiple assignments we used the Jaccard Index to measure the concordance between the two journal The results are presented in Table 3.

Arts and Humanities is an outlier, Neurosciences & Behaviour acts as a bridge between Social Sciences and Life Sciences, Chemistry takes a central position between Biosciences, Medical Sciences and Physics.Most striking observation in their map is the position of General,Regional and Community Issues which is strongly linked with the Life Science fields.

A 24 cluster solution can be compared with the 22 categories from the classification of Thomson Reuters’ Essential Science Indicators (ESI).

Janssens et al. (2009) showed very low mean silhouette values for the ESI category system in a space with respectively textual distances, cosine similarities of cross-citation vectors and combined distances.

Also in the present study, not all clusters have a unique counterpart in the ESI classification system and vice versa (cf. Janssens et al., 2009). Notably, the ESI fields clinical medicine and engineering, mathematics and social sciences, general are almost uniformly spread over numerous clusters.

Based on this analysis we have to conclude that the segmentation of journals in the ESI categories is not supported by the structure found with bibliographic coupling between journals.

Given this rather weak association between the clustering based on cross-citations and bibliographic coupling, it is a legitimate question to ask which of both methodologies is performing best. A comparison of the mean silhouette values and the silhouette values within each cluster reveals that the methodology presented in this paper results in a more consistent solution.

The 15 cluster solution of cross-citation has a value of 0.04 while the bibliographic coupling results in a value of 0.13.

The main advantage of this method is that clustering can be made as soon as a new database volume is available. The only issue is the lacking cluster labelling that cannot directly be obtained from the method. As a substitute, intellectual classification schemes can be used as reference system. Cluster labelling was made on the basis of the Leuven-Budapest subject-classification scheme that also allowed the comparison with the existing two-level journal classification system developed in Leuven.

The further comparison with the 22 field classification system of the Essential Science Indicators does, however, revealed some striking deviations. These concerned, above all, the fields of clinical medicine, engineering, mathematics and the social sciences. New developments in computer science, neuroscience and psychology as well as in public health (cf. Glänzel & Thijs, 2011) do certainly contribute to such growing deviation.

The main objective of this study was to analyse whether the proposed methodology is appropriate for multi-level journal clustering and to what extent the solutions fit in the framework of traditional subject classification. Further comparison with other solutions such as cross-citation and hybrid methods will be part of future research.

Tseng, Y.-H. and Tsay, M.-Y. (2013) Journal clustering of library and information science for subfield delineation using the bibliometric analysis toolkit: CATAR. Scientometrics, 95, 503-528. doi: 10.1007/s11192-013-0964-1.

Tseng,  Y.-H. and Tsay, M.-Y. (2013) Journal clustering of library and information science for subfield delineation using the bibliometric analysis toolkit: CATAR. Scientometrics, 95, 503-528. doi: 10.1007/s11192-013-0964-1.

近幾十年來，發展出許多科學計量分析技術，包括為了群集(clustering)書目資料所需的各種相似度(similarity)計算技術，如共被引(co-citation)、書目耦合(bibliographic coupling)與詞語共現分析(co-word analysis)，這些技術的比較分析可參見Yan and Ding (2012)。並且有很多可以在網路上自由下載使用的軟體工具製作並包裝這些技術，提供科學計量分析應用，知名的軟體工具如CiteSpace (Chen 2006, Chen et al. 2010)、Sci2 Tool (Sci2 Team 2009)、VOSviewer (Van Eck and Waltman 2010)、BibExcel (Persson 2009)及Sitkis (Schildt and Mattsson 2006)，這部分的分析則可參見Cobo et al. (2011)。本研究包含兩個部分：提出包含一系列利用書目計量資訊進行群集與映射(mapping)技術的科學計量分析軟體工具集 CATAR，並且將此工具集應用於圖書資訊學(library and information science, LIS)領域後，希望能夠利用期刊群集的結果，確認與分析次領域，以及建議適合研究評估(research evaluation)用途的LIS期刊集合。

Åström (2002)從領域概念的視覺化研究獲得一個結論：期刊的選擇確實影響研究領域如何被知覺與定義，也就是研究領域的界定(delineation)與期刊的選擇有密切關係。已經有許多的研究對圖書資訊學進行次領域界定，而這些研究大多參考ISI的JCR主題分類中與圖書資訊學最相關的類別IS&LS(Information Science and Library Science)。IS&LS類別下並不只包含圖書資訊學的相關期刊，這個類別涵蓋兩個密切相關的領域資訊科學(Information Science)和圖書館學(Library Science)，此一範圍與圖書資訊學有些微不同。根據Leydesdorff (2008)，JCR主題分類以期刊的題名、引用模式(citation patterns)等等做為標準進行分類，但是這個分類結果與從資料庫本身的引用資料所產生的網路上的主要成分(principal components)得到的分類結果並不十分相符。因此次領域界定研究大多經過人為的挑選做為分析資料的期刊，並沒有完整收錄IS&LS主題下的所有期刊。

進行次領域界定時常使用的技術包括：利用共被引分析比較一對項目，利用凝聚式階層群集(agglomerative hierarchical clustering, AHC)將項目分群產生樹狀圖(dendrogram)，利用多維尺度(multi-dimensional scaling, MDS)產生視覺化的二維或三維映射圖。若干重要的研究如：Åström (2002)從圖書資訊學重要期刊中選取1135篇出版在1998到2000年的文章，利用BibExcel軟體工具進行作者共被引(author co-citation)以及關鍵詞共現分析，並產生MDS映射圖，52位高被引作者的共被引產生三個群集："硬"資訊檢索(hard information retrieval)、"軟"資訊檢索(soft information retrieval)以及書目計量學(bibliometrics)，47個較常出現的關鍵詞則分為圖書館學(library science，LS)、資訊檢索(information retrieval，IR)及書目計量學。Åström (2002)認為作者共被引分析沒有出現圖書館學的原因可能與圖書館學研究的出版管道有關，如果引用的資料像是書籍或地區期刊沒有出現在JCR，圖書館學作者便無法出現在引用為基礎的排名上。Åström (2007)對55種在JCR 2003主題類別下的期刊，選擇21種圖書資訊學相關期刊的13605篇文章進行文件共被引分析，在從1990到2004年的三個時段發現圖書資訊學可分為資訊計量學(informetrics)和資訊搜尋與檢索(information seeking and retrieval)兩個穩定的次領域，而隨著全球資訊網的普及，網路計量學(webometrics)在兩個次領域上都成為主要的研究議題。Jassen et al. (2006) 對2002到2004年五種圖書資訊學相關期刊的938篇文章，應用一系列的全文分析技術以及MDS和AHC，將938篇文章分為六個群集：兩個群集與書目計量學有關、一個群集為IR、一個包含一般議題、另兩個較小但愈來愈重要的群集分別是網路計量學和專利分析(patent analysis)。Moya-Anegon et al. (2006)從24種較有影響力的期刊中選擇17種期刊，排除將資訊科學(information science, IS)應用到特定技術或知識領域(例如：醫學、地理學、電訊傳播等)，從17種期刊引用的參考文獻，對77位最常被引用的作者和73篇最常被引用的期刊進行共被引分析，映射使用的技術包括MDS和AHC以及自組織映射圖(self-organizing map)。作者共被引分析的結果產生六個次領域：科學計量學、引用分析、書目計量學、"軟"(認知導向)資訊檢索、"硬"(演算法導向)資訊檢索以及傳播理論(communication theory)。而期刊共被引分析的結果則有四個群集：IS、LS、科學研究(science studies)以及管理學(management)。在期刊共被引分析的科學研究大致上可以對應為作者共被引分析的科學計量學、引用分析、書目計量學，IS為"軟"資訊檢索和"硬"資訊檢索。如Åström (2002)同樣的原因，LS沒在作者共被引分析的結果當中。Waltman et al. (2011)以JASIST為種子，選擇與該期刊共被引較多的期刊，連JASIST共48種，進行期刊的書目耦合(bibliographic coupling)分析，並且利用VOSviewer呈現視覺化結果，共分為LS、IS以及科學計量學等3個次領域。Milojevic et al. (2011)使用詞語共現分析探討1998到2007年出版的16種期刊上的10344篇文章，16種期刊根據Nisonger and Davis (2005) 的研究所挑選，分析100個文章題名上最常出現的詞語，進行共現分析，並以AHC歸類，結果三個主要群集為LS、IS以及書目計量學/科學計量學。

Åström (2002)以關鍵詞的共現分析所得到的結果包括LS次領域，但作者共被引分析所得到的映射圖上並沒有產生這個次領域。Moya-Anegon et al. (2006)的期刊共被引分析與作者共被引分析也略有不同，期刊共被引分析的結果上有作者共被引分析沒有的LS和管理學兩個次領域，反之，作者共被引分析的結果上則可以發現期刊共被引分析沒有的傳播學理論(communication theory)。一般認為這和作者引用的行為有關，LS作者的引用次數大多沒有達到分析的門檻，因此無法在上述兩個研究的作者共被引分析結果上呈現。

Ni et al. (2012)從JCR的IS&LS類別下的61種期刊，排除3種非英語的期刊，將選取的58種期刊進行場域-作者耦合(venue-author coupling)、期刊共被引分析、詞語共現分析、期刊連結(journal interlocking)等四種分析。分析的結果再進行MDS與AHC分析，四種方式所得到一致的次領域包括：管理資訊系統(managment information systems, MIS)、IS、LS和特殊化群集(specialized clusters)，並且在四種方法所得到MDS映射的圖形上都可以發現MIS與其他群集分離，Ni and Ding (2010)與Ni and Sugimoto (2011)建議JCR上的圖書資訊相關期刊應進行適當的重組。

本研究(Tseng and Tsai 2013)應用的資料範圍為2000到2004與2005到2009在Web of Science 的Journal Citation Report中 Information Science & Library Science (IS&LS)主題分類下的所有期刊，在前期(2000~2004年)共50種，後期(2005~2009年)共66種。本研究的分析程序採用Borner et al. (2003)整理的一般工作流程，步驟包括：1) 資料蒐集(data collection)；2)文本分段(text segmentation)；3)相似性計算(similarity computation)；4)多階段群集(multi-stage clustering)；5)群集標名(clustering labeling)；6)視覺化(visualization)；7)面向分析(facet analysis)。這些步驟中所需的技術都已經整合到軟體工具CATAR(Content Analysis Toolkit for Academic Research, http://web.ntnu.edu.tw/~samtseng/CATAR/)上。在計算文件間的相關性時，本研究以一種期刊做為一個文件，所有論文引用的期刊做為文件的特徵，然後利用Dice係數(Salton 1989)計算期刊相似性，例如兩種期刊X與Y，R(X)與R(Y)分別是它們引用的期刊，它們之間的相似性計算為Sim(X, Y) = 2 ∙ |R(X)∩R(Y)|/(|R(X)|+R(Y)|)。也就是利用書目耦合計算期刊之間的相似性。期刊的群集則是利用完全連接階層群集法(complete-linkage hierarchical clustering)。首先將每個文件視為一個群集，然後將一對最相似的群集合併起來，產生一個較大的群集，然後重複進行上面的步驟，而兩個群集的相似性定義為兩個群集間最小的文件相似性，如果相似性超過某個預先設定的閾值，便將兩個群集合併，一直到無法再產生合併為止。此外，本研究採用Silhouette指標(Ahlgren and Jarneving 2008; Rousseuw 1987; Jassen et al. 2006)。

此一研究的資料包含JCR的IS&LS主題下的期刊，分為2000-2004年與2005-2009年兩個時期，前一個時期包含50種期刊，9546筆論文資料；後一時期則有66種期刊，11471筆論文資料。從群集結果的樹狀圖(dendrogram)和MDS映射的結果顯示，IS&LS主題下的期刊在兩個時期都有IR、MIS、科學計量學、學術圖書館(academic library)、醫學圖書館(medical library)、館藏發展(collection development)，以及開放取用(open access)和地區圖書館(regional library)兩個後期出現並且較小的群集。並且MIS群集的期刊在知識基礎(intellectual base)上與IS&LS主題的其他期刊分離，表示這群集下的期刊具有較特殊的引用模式。本研究以期刊的書目耦合進行分析，從期刊知識基礎(intellectual base)得到MIS群集與其他分離的研究結果，與Ni et al. (2012)利用期刊共被引分析、期刊連結、術語使用(terminology usage)和合著(co-authorship)研究等不同方法的研究結果相同，這也為許多探討圖書資訊學認知結構的研究認為不應將MIS相關期刊與其他期刊包含在ISI的同一個主題IS&LS下，在進行分析時需要排除MIS相關期刊提供了佐證(Larivière et al. 2012)。此外，並且以多樣性指標(diversity index)分析群集特性，揭露出某些次領域具有地區(regional)特性。

Chen, C., Ibekwe-SanJuan, F. and Hou, J. (2010), The structure and dynamics of cocitation clusters: A multiple-perspective cocitation analysis. Journal of the American Society for Information Science and Technology, 61 (7), 1386–1409. doi: 10.1002/asi.21309

Chen, C., Ibekwe-SanJuan, F. and Hou, J. (2010), The structure and dynamics of cocitation clusters: A multiple-perspective cocitation analysis. Journal of the American Society for Information Science and Technology, 61 (7), 1386–1409. doi: 10.1002/asi.21309

確認科學領域的專業(specialties)本質是資訊科學的一項基本挑戰 (Morris & Van der Veer Martens, 2008; Tabah, 1999) 。由於1)可取用的書目資料來源愈來愈普及；2)網路上愈來愈多可提供分析與視覺化的電腦軟體工具；3)從多元來源而大量的資料吸收的要求愈來愈劇烈等原因，因此有愈來愈多的相關研究。共被引分析是對科學進行量化分析最常用的方法之一，特別是作者共被引分析 (author cocitation analysis, ACA; Chen, 1999; Leydesdorff, 2005; White & McCain, 1998; Zhao & Strotmann, 2008b)以及文件共被引分析 (document cocitation analysis, DCA; Chen, 2004; Chen, 2006; Chen, Song, Yuan, & Zhang, 2008; Small & Greenlee, 1986; Small & Sweeney, 1985; Small, Sweeney, & Greenlee, 1985)。作者共被引分析的目的在透過被相關文獻一起引用的作者群集，確認領域裡的專業。重要的作者共被引分析研究包括White & McCain (1998)，這個研究以1972到1995年間12種資訊科學相關期刊的120位高被引作者進行作者共被引分析，研究結果發現當時的資訊科學分為兩個基本上彼此獨立的陣營：資訊檢索(information retrieval)與文獻(literature)。Zhao and Strotmann (2008a, 2008b) 以1996-2005年的資訊科學相關期刊資料重新進行了相同的研究，他們的結果發現了5個主要的專業：使用者研究(user studies)、引用分析(citation analysis)、實驗型檢索(experimental retrieval)、網路計量學 (Webometrics)以及知識領域的視覺化(visualization of knowledge domains)，其中新興的兩個專業：網路計量學和知識領域的視覺化連繫了引用分析以及實驗型檢索，而使用者研究則是此時最大的專業。Aström (2007) 則是使用文件共被引分析的例子，他們分析了1990到2004年的21種圖書資訊學期刊，利用多維尺度法(multidimensional scaling, MDS)產生結果，他們的結果與White & McCain (1998)的研究類似，整個領域可分為兩個陣營，不過Aström (2007)的結果將稱為資訊尋求與檢索(information seeking and retrieval)，而不是資訊檢索。

不管是作者共被引分析或是文件共被引分析其步驟大致如下：
1) 檢索引用資料。
2) 建構參考文件或作者共同被引用的矩陣。
3) 將共被引矩陣表示成節點與連結的圖(node-and-link graph)或是多維尺度法的組態(configuration)，並且可以利用尋路網路(Pathfinder network scaling)或最小生成樹(minimum spanning tree)裁減連結。
4) 利用群集、社群發現(community finding)、因素分析(factor analysis)、主成分分析(principle component analysis)或者隱含語意索引(latent semantic indexing)等各種演算法確認專業。例如Morris & Van der Veer Martens (2008)、 Persson (1994)、 Tabah (1999)、 White & Griffith (1982)以及Janssens, Leta, Glänzel, and De Moor (2006)。
5) 根據群集成員間共同的主題(themes)，解釋共被引群集的性質。通常需要豐富的領域知識，而且是一個花費大量時間與認知需求(cognitively demanding)的工作。

本研究對於作者共被引以及文件共被引形成的群集進行結構與動態的描述與解釋，分析的資料為1996到2008年間的12種資訊科學(information science)領域相關期刊，共計10853筆書目紀錄，引用的參考文獻為129060筆，引用次數為206180，而參考文獻的作者共有58711位。本研究以餘弦(cosine)測量作者或文件之間的關連大小，做為節點間的連結，建立網路；然後計算從原先網路導出的Laplacian矩陣(Laplacian matrices)的特徵向量(eigenvectors)找出群集。這種利用標準線性代數的頻譜群集(spectral cluster)演算法，較其他的群集演算法更有效率，而且因為不需要假設群集的形式，所以更有彈性與強健。標註群集方面則是利用引用文獻論文的詞語與摘要句，詞語包括題名與摘要中出現的名詞片語與索引詞(index terms)，利用 tf*idf (Salton, Yang, & Wong, 1975)、對數似然比(log-likelihood ratio, LLR)測試 (Dunning, 1993)以及相互資訊(mutual information, MI)等三種資訊做為判斷的參考。摘要句則是從題名與摘要尋找最有代表性的句子，例如以Enertex (Fernandez, SanJuan, & Torres-Moreno, 2007)對句子進行排序。

A multiple-perspective cocitation analysis method is introduced for characterizing and interpreting the structure and dynamics of cocitation clusters.

The generic method is applied to a three-part analysis of the field of information science as defined by 12 journals published between 1996 and 2008: (a) a comparative author cocitation analysis (ACA), (b) a progressive ACA of a time series of cocitation networks, and (c) a progressive document cocitation analysis (DCA).

Identifying the nature of specialties in a scientific field is a fundamental challenge for information science (Morris & Van der Veer Martens, 2008; Tabah, 1999).

The growing interest in mapping and visualizing the structure and dynamics of specialties is because of a number of reasons:
1. Widely accessible bibliographic data sources such as the Web of Science, Scopus, and Google Scholar (Bar-Ilan, 2008; Meho & Yang,2007) as well as domain-specific repositories such as ADS (http://www.adsabs.harvard.edu/) and arXiv (http://arxiv.org/).
2. Freely available computer programs and Web-based general-purpose visualization and analysis tools such as ManyEyes (http://manyeyes.alphaworks.ibm.com/) and Pajek (http://vlado.fmf.uni-lj.si/pub/networks/pajek/; Batagelj & Mrvar, 1998), special-purpose citation analysis tools such as CiteSpace (http://cluster.cis.drexel.edu/&u0007E;cchen/citespace/; Chen, 2004; Chen, 2006), and social network analysis such as UCINET (http://www.analytictech.com/ucinet6/ucinet.htm).
3. Intensified challenges for digesting the vast volume of data from multiple sources (e.g., e-Science, Digging into Data (http://www.diggingintodata.org/), cyber-enabled discovery, SciSIP; Lane, 2009).

Cocitation studies are among the most commonly used methods in quantitative studies of science, especially including author cocitation analysis (ACA; Chen, 1999; Leydesdorff, 2005; White & McCain, 1998; Zhao & Strotmann, 2008b) and document cocitation analysis (DCA; Chen, 2004; Chen, 2006; Chen, Song, Yuan, & Zhang, 2008; Small & Greenlee, 1986; Small & Sweeney, 1985; Small, Sweeney, & Greenlee, 1985).

For instance, once cocitation clusters are identified, assigning the most meaningful labels for these clusters is currently a challenging task because any representative labels of clusters must characterize not only what clusters appear to represent, but also the salient and unique reasons for their formation.

The new procedure reduces analysts' cognitive burden by automatically characterizing the nature of a cocitation cluster in terms of (a) salient noun phrases extracted from titles, abstracts, and index terms of citing articles and (b) representative sentences as summarizations of clusters.

ACA aims to identify underlying specialties in a field in terms of groups of authors who were cited together in relevant literature.

White & McCain (1998) presented a comprehensive view of information science based on 12 journals in library and information science across a 24-year span (1972–1995). It analyzed cocitation patterns of 120 most-cited authors with factor analysis and multidimensional scaling. The authors drew upon their extensive knowledge of the field and offered an insightful interpretation of 12 specialties identified in terms of 12 factors. The most well-known finding of the study is that information science at the time consisted of two essentially independent camps, namely, the information retrieval camp and the literature camp, including citation analysis, bibliometrics, and scientometrics.

Zhao and Strotmann (2008a, 2008b) followed up White and McCain's study using the same set of 12 journals and the same number of 120 cited authors in an updated time frame of 1996-2005. ... Zhao and Strotmann (2008b) found five major specialties and manually labeled them as user studies, citation analysis, experimental retrieval, Webometrics, and visualization of knowledge domains. In contrast to the findings of (White & McCain, 1998), experimental retrieval and citation analysis retained their fundamental roles in the field, and the user studies specialty became the largest specialty. Webometrics and visualization of knowledge domains appeared to make connections between the retrieval camp and the citation analysis camp.

A DCA by Aström (2007) studied papers published between 1990 and 2004 in 21 library and information science journals. Results were depicted in multidimensional scaling (MDS) maps. Aström's study also identified the two-camp structure found by (White & McCain, 1998). On the other hand, Aström found an information seeking and retrieval camp, instead of the information retrieval camp as in (White and McCain).

Although manually labeling a cocitation cluster can be a very rewarding process of learning about the underlying specialty and result in insightful and easy to understand labels, it requires a substantial level of domain knowledge and it tends to be time-consuming and cognitively demanding because of the synthetic work required over a diverse range of individual publications.

Traditionally, researchers often identify the nature of a cocitation cluster based on common themes among its members. ... The emphasis on common areas is a practical strategy; otherwise, comprehensively identifying the nature of a specialty can be too complex to handle manually.

Many researchers have studied the structural and dynamic properties of specialties in information science in terms of clusters, multivariate factors, and principle components (Morris & Van der Veer Martens, 2008; Persson, 1994; Tabah, 1999; White & Griffith, 1982).

A recent study of information science (Ibekwe-SanJuan, 2009) mapped the structure of information science at the term level using a text analysis system TermWatch and a network visualization system Pajek, but it did not address structural patterns of cited references.

Researchers also studied the structure of information science qualitatively, especially with direct inputs from domain experts. For example, Zins conducted a Critical Delphi study of information science, involving 57 leading information scientists from 16 countries (Zins, 2007a, 2007b, 2007c, 2007d).

Janssens, Leta, Glänzel, and De Moor (2006) studied the full-text of 938 publications in five library and information science journals with latent semantic analysis (LSA; Deerwester, Dumais, Landauer, Furnas, & Harshman, 1990) and agglomerative clustering. They found an optimal 6-cluster solution in terms of a local maximum of the mean silhouette coefficients (Rousseeuw, 1987) and a stability diagram (Ben-Hur, Elisseeff, & Guyon, 2002). Their clusters were labeled with single-word terms selected by tf*idf (p. 1625), which are not as informative as multiword terms for cluster labels.

Klavans, Persson, and Boyack (2009) recently raised the question of the true number of specialties in information science. They suspected that the number is much more than the 11 or 12 as reported in ACA studies such as (White & McCain, 1998) and (Zhao & Strotmann, 2008a, 2008b), but significantly fewer than the 72 reported in their own study, which is also based on the 12 journals between 2001 and 2005.

The 12-journal Information Science dataset, retrieved from the Web of Science, contains 10,853 unique bibliographic records, written by 8,408 unique authors from 6,553 institutions and 89 countries. These articles cited 129,060 unique references for a total of 206,180 times. They cited 58,711 unique authors and 58,796 unique sources.

The traditional procedure of cocitation analysis for both DCA and ACA comprises the following steps:

1. Retrieve citation data from sources such as the Science Citation Index (SCI), Social Science Citation Index (SSCI), Scopus, and Google Scholar.

2. Construct a matrix of cocited references (DCA) or authors (ACA).

3. Represent the cocitation matrix as a node-and-link graph or as a multidimensional scaling (MDS) configuration with possible link pruning using Pathfinder network scaling or minimum spanning tree algorithms.

4. Identify specialties in terms of cocitation clusters, multivariate factors, principle components, or dimensions of a latent semantic space using a variety of algorithms for clustering, community finding, factor analysis, principle component analysis, or latent semantic indexing.

5. Interpret the nature of cocitation clusters.

The interpretation step is the weakest link. It is time-consuming and cognitively demanding, requiring a substantial level of domain knowledge and synthesizing skills. In addition, much of attention routinely focuses on cocitation clusters per se, but the role of citing articles that are responsible for the formation of such cocitation clusters may not be always investigated as an integral part of a specialty.

Our new method extends and enhances traditional cocitation methods in two ways: (a) by integrating structural and content analysis components sequentially into the new procedure and (b) by facilitating analytic tasks and interpretation with automatic cluster labeling and summarization functions. The new procedure is highlighted in yellow in Figure 2, including clustering, automatic labeling, summarization, and latent semantic models of the citing space (Deerwester et al., 1990).

Our new procedure adopts several structural and temporal metrics of cocitation networks and subsequently generated clusters.

Structural metrics include betweenness centrality, modularity, and silhouette.

Temporal and hybrid metrics include citation burstness and novelty

The betweenness centrality metric is defined for each node in a network. It measure the extent to which the node is in the middle of a path that connects other nodes in the network (Brandes, 2001; Freeman, 1977). High betweenness centrality values identify potentially revolutionary scientific publications (Chen, 2005) as well as gatekeepers in social networks.

In the context of this study, the modularity Q measures the extent to which a network can be divided into independent blocks, i.e., modules (Newman, 2006; Shibata, Kajikawa, Taked, & Matsushima, 2008).

The silhouette metric (Rousseeuw, 1987) is useful in estimating the uncertainty involved in identifying the nature of a cluster.

Burst detection determines whether a given frequency function has statistically significant fluctuations during a short time interval within the overall time period.

Sigma is introduced in (Chen, et al., 2009a) as a measure of scientific novelty. ... In this study, Sigma is defined as (centrality + 1)burstness such that the brokerage mechanism plays more prominent role than the rate of recognition by peers.

We adopt a hard clustering approach such that a cocitation network is partitioned to a number of nonoverlapping clusters.

In this article, cocitation similarities between items i and j are measured in terms of cosine coefficients.

A good partition of a network would group strongly connected nodes together and assign loosely connected ones to different clusters. This idea can be formulated as an optimization problem in terms of a cut function defined over a partition of a network. Technical details are given in relevant literature (Luxburg, 2006; Ng, Jordan, & Weiss, 2002; Shi & Malik, 2000).

Spectral clustering is an efficient and generic clustering method (Luxburg, 2006; Ng et al., 2002; Shi & Malik, 2000). It has roots in spectral graph theory. Spectral clustering algorithms identify clusters based on eigenvectors of Laplacian matrices derived from the original network.

Spectral clustering has several desirable features compared to traditional algorithms such as k-means and single linkage (Luxburg, 2006):
 • It is more flexible and robust because it does not make any assumptions on the forms of the clusters,
• it makes use of standard linear algebra methods to solve clustering problems, and
• it is often more efficient than traditional clustering algorithms.

Candidates of cluster labels are selected from noun phrases and index terms of citing articles of each cluster. These term are ranked by three different algorithms. In particular, noun phrases are extracted from titles and abstracts of citing articles. The three term ranking algorithms are tf*idf (Salton, Yang, & Wong, 1975), log-likelihood ratio (LLR) tests (Dunning, 1993), and mutual information (MI).

Each cocitation cluster is summarized by a list of sentences selected from the abstracts of articles that cite at least one member of the cluster.

In this study, sentences are ranked by Enertex (Fernandez, SanJuan, & Torres-Moreno, 2007). Given a set S of N sentences, let M be the square matrix that for each pair of sentences gives the number of nominal words in common (nouns and adjectives).

In this study, summarization sentences were also ranked by two new functions gtf and gftidf , which are further simplified approximations of the energy function E.

The ACA and DCA studies described in this article were conducted using the CiteSpace system (Chen, 2004; Chen, 2006). CiteSpace is a freely available Java application for visualizing and analyzing emerging trends and changes in scientific literature.

CiteSpace supports a unique type of cocitation network analysis—progressive network analysis—based on a time slicing strategy and then synthesizing a series of individual network snapshots defined on consecutive time slices. Progressive network analysis particularly focuses on nodes that play critical roles in the evolution of a network over time. Such critical nodes are candidates of intellectual turning points.

In summary, (a) spectral clustering and factor analysis identified about the same number of specialties, but they appeared to reveal different aspects of cocitation structures and (b) cluster labels chosen from citers of a cluster tend to be more specific terms than those chosen by human experts.

We found the comparison with the study of Zhao and Strotmann very valuable. It offered us an opportunity to compare the analysis conducted by human experts to the interpretation cues provided by our automatic labeling and summarization methods.

Spectral clustering for the purpose of network decomposition is exclusive in nature although in reality it is often sensible to allow overlapping clusters because of multiple roles individual entities may play.

Spectral clustering of cocitation networks tends to generate distinct clusters with high precision, whereas human experts tend to aggregate entities into broadly defined clusters.

In conclusion, the new cocitation analysis procedure has the following advantages over the traditional one:
• It can be consistently used for both DCA and ACA.
• It uses more flexible and efficient spectral clustering to identify cocitation clusters.
• It characterizes clusters with candidate labels selected by multiple ranking algorithms from the citers of these clusters and reveals the nature of a cluster in terms of how it has been cited.
• It provides metrics such as modularity and silhouette as quality indicators of clustering to aid interpretation tasks.
• It provides integrated and interactive visualizations for exploratory analysis.

Modularity and silhouette metrics provide useful quality indicators of clustering and network decomposition.