Search

In microblogging platforms, hashtags are used to annotate the microblogs for a more convenient categorization and analysis of the published contents. Due to the fast growth of the social network, the hashtag recommendation field has attracted the researchers’ attention most recently. In this study, a review of existing works in the hashtag recommendation filed is presented. After collecting all the papers in this field, the author keywords are exploited in order to extract popular topics and explore the evolution of them since their inception. In this regard, statistical analysis of the keywords, keyword-pairs co-occurrences, and the cluster analysis through the co-word data (co-word analysis) are performed. The obtained results demonstrate that there are four evolved thematic areas in this research field, including “SIMILARITY”, “HASHTAG-RECOMMENDATION”, “MACHINE-LEARNING”, and “POPULARITY-PREDICTION”. Besides, there are some popular themes in each thematic area, such as the “DEEP_LEARNING”, which has excellent future development potential. Similarly, the “SIMILARITY” and “TOPIC-MODEL” are two motor themes that have gained increased interest from researchers in recent studies. Eventually, the analysis results of the related works in the hashtag recommendation domain are utilized to extract the main approaches in this research area involving “DEEP LEARNING”, “TOPIC MODELING”, “SIMILARITY”, “CLASSIFICATION”, and “TOPICAL TRANSLATION”. The results’ implications and the future research directions determined that the researchers’ interest in the field of hashtag recommendation will increase rapidly.

In this paper, we explore the longitudinal research collaboration network of `mammography performance' over 30 years by creating and analysing a large collaboration network data using Scopus. The study of social networks using longitudinal data may provide new insights into how this collaborative research evolve over time as well as what type of actors influence the whole network in time. The methods and findings presented in this work aim to assist identifying key actors in other research collaboration networks. In doing so, we apply a rank aggregation technique to centrality measures in order to derive a single ranking of influential actors. We argue that there is a strong correlation between the level of degree and closeness centralities of an actor and its influence in the research collaboration network (at macro/country level).

Although co-authorship in scientific research has a long history the analysis of co-authorship network to explore scientific collaboration among authors is a relatively new research area. Studies of current literature about co-authorship networks mostly give emphasis to understand patterns of scientific collaborations, to capture collaborative statistics, and to propose valid and reliable measures for identifying prominent author(s). However, there is no such study in the literature which conducts a longitudinal analysis of co-authorship networks. Using a dataset that spans over 20 years, this paper attempts to explore efficiency and trend of co-authorship networks. Two scientists are considered connected if they have co-authored a paper, and these types of connections between two scientists eventually constitute co-authorship networks. Co-authorship networks evolve among researchers over time in specific research domains as well as in interdisciplinary research areas. Scientists from diverse research areas and different geographical locations may participate in one specific co-authorship network whereas an individual scientist may belong to different co-authorship networks. In this paper, we study a longitudinal co-authorship network of a specific scientific research area. By applying approaches to analyze longitudinal network data, in addition to known methods and measures of current co-authorship literature, we explore a co-authorship network of a relatively young and emerging research discipline to understand its trend of evolution pattern and proximity of efficiency.

At global and local levels, we are observing an increasing range and rate of disease outbreaks that show evidence of jumping from animals to humans, and from food to humans. Zoonotic infections (i.e. Hendra, swine flu, anthrax) affect animal health and can be deadly to humans. The increasing rate of outbreaks of infectious diseases transferring from animals to humans (i.e. zoonotic diseases) necessitates detailed understanding of the education, research and practice of animal health and its connection to human health. These emerging microbial threats underline the need to exploring the evolutionary dynamics of zoonotic research across public health and animal health. This study investigates the collaboration network of different countries engaged in conducting zoonotic research. We explore the dynamics of this network from 1980 to 2012 based on large scientific data developed from Scopus. In our analyses, we compare several properties of the network including density, clustering coefficient, giant component and centrality measures over time. We also map the network over different time intervals using VOSviewer. We analyzed 5182 publication records. We found United States and United Kingdom as the most collaborative countries working with 110 and 74 other countries in 1048 and 599 cases, respectively. Our results show increasing close collaboration among scientists from the United States, several European countries including United Kingdom, Italy, France, Netherland, Switzerland, China and Australia with scientists from other parts of the world.