In the last century, the number of researchers and published scientific works has greatly increased. Tracking scientific discoveries and determining their impact is becoming increasingly difficult. In addition, researchers are increasingly competing in gaining a reputation in science. Therefore, there was a need for an automatic and reliable measure for ranking scientific works.
Citation networks provide a good insight into the development of science. With the help of measures such as the node intermediacy, we can determine which scientific works played a more important role in the development of a given scientific field.
In the master's thesis we present the node intermediacy measure and extend it to weighted graphs. We then evaluate several ways of mapping the weights of connections to probabilities. We also present the Monte Carlo algorithm for calculating the measure and implement a web application. Finally, we examine how the different properties of the articles affect the extended measure of node intermediacy and evaluate them empirically. Compared to the original measure, our gives slightly better results.
|
