Please tell us what you think of this issue!  Feedback

Bulletin, April/May 2009


Whatís New?
 
Selected Abstracts from JASIST

Authors who choose to do so prepare and submit these summaries to the editor of the Bulletin.

From JASIST v. 60 (1) 
Neuhaus, C., Marx, W., & Daniel, H.D.
(2009). The publication and citation impact profiles of Angewandte Chemie and the Journal of the American Chemical Society based on the sections of Chemical Abstracts: A case study on the limitations of the Journal Impact Factor. (176-183).

Study and Results: Taking Angewandte Chemie International Edition and the Journal of the American Chemical Society as examples, the study examines the publication and citation impact profiles of both journals across the sections of the bibliographic database Chemical Abstracts. The findings suggest that a single measure of journal citation impact such as the Journal Impact Factor published by Thomson Reuters is insufficient for characterizing the significance and performance of multidisciplinary and wide-scope journals.

Whatís New? The findings show that the information available in the Science Citation Index is a rather unreliable indication of the document type and is therefore inappropriate for comparative analysis of journals. The findings further suggest that the composition of the journal in terms of contribution types, the length of the citation window and the thematic focus of the journal in terms of the sections of Chemical Abstracts has a significant influence on the overall journal citation impact. For the comparison of multidisciplinary and wide-scope journals more sophisticated methods such as publication and citation impact profiles across subject headings of bibliographic databases (for example, .the sections of Chemical Abstracts) are therefore valuable.

Limitations: A high quality of indexing information is a prerequisite for the applicability of a subject classification scheme to comparative analysis.

From JASIST v. 60 (2) 
Bose, I., & Chen, X.
(2009). A method for extension of generative topographic mapping for fuzzy clustering. (363-372).

Study and Results: We developed a clustering method (GTMFCM) that combined generative topographic mapping (GTM) and fuzzy c-means algorithm (FCM). It is observed that this method performs better than FCM and Gustafson-Kessel algorithms in terms of values of clustering validity indexes. In business applications, this new method can be used to explore segments of customers to create a complete and vivid profile of customersí behavioral patterns. Businesses can benefit from such knowledge by aligning their marketing strategies with customersí preferences.

Whatís New? Clustering algorithms explore the hidden behavioral patterns of customers. However the presentation and interpretation of clustering results are difficult for practitioners. Fuzzy clustering techniques are believed to be more capable of revealing information about customersí behavioral patterns because they assign data to clusters with probabilities. On the other hand, techniques such as GTM visualize the distribution of customers but cannot group them into the desired number of clusters. By combining the two techniques, we created a tool which can do visualization and clustering at the same time with acceptable performance. 

Limitations: The paper is limited in that the proposed method has been validated using benchmark and simulated data sets but needs further validation using customer data collected from real business situations. 

Pera, M. S., & Ng, Y.-K. (2009). SpamED: A spam email detection approach based on phrase similarity. (393-411).

Study and Results: We introduce a novel spam-email detection approach, denoted SpamED, which is designed for solving todayís problem on increasing influx of spam emails that reach userís inboxes, leading to monetary loss and waste of computational resources. The premise of our investigation is to correctly identify incoming (non)-spam emails based solely on performing exact and partial similarity matching among the phrases in an incoming email E and the ones in a user-identified spam email S, which determines how similar E and S are and subsequently establishes the likelihood of E being (non-)spam. Experimental results compiled by using known spam-detection corpuses on SpamED prove the effectiveness of SpamED with a 96% accuracy in correctly classifying incoming emails.

Whatís New? SpamED is computational inexpensive, since the word-correlation factors used for establishing the degrees of similarity among emails are (i) pre-computed, (ii) are efficient in detecting (non)-spam emails, (iii) require little user intervention (for labelling incoming spam emails), (iv) minimize the number of misclassified legitimate emails which contain information users cannot afford to lose, (v) allow users the feasibility in expressing and modifying their preference on what constitutes spam and (vi) outperform existing spam-detection approaches in terms of accuracy.

Limitations: SpamED is designed for processing text-based emails.