The Research Of Literature Value Prediction Algorithm And Author Impact Assessment Algorithm In Academic Network

The evaluation problem of academic influence is to use some methods to evaluate the impact of publications, authors, journals or organizations. The evaluation problem of academic influence can be group into four groups:(1) Literature Value Assessment and Prediction;(2) Author Impact Assessment;(3) Journal (or Conference) Impact Assessment;(4) Organization Impact Assessment. These problems focus on different aspects of academic influence assessment. The methods of solving these problems can help researchers to master the key point and hot research problem in a certain field, find experts, find high impact journals (or conferences) and high impact organizations.This paper focuses on the first two problems:Literature Value Assessment and Prediction and Author Impact Assessment and we have finished the following works:(1) We have proposed a literature value prediction algorithm, PRank. We have studied the literature value prediction algorithm, such as FutureRank and Liu’s proposed method. These two algorithms have been modified and improved in this work and a literature value prediction algorithm has been proposed. The proposed algorithm takes literatures’published time and citation information into consideration. In order to verify the validity of the proposed algorithm, we have compared it with other two different algorithms on three different datasets. The results show that the time consumption has been reduced and the predication accuracy has been improved.(2) We have proposed an author impact assessment algorithm, named as PR-index, through combining PageRank and h-index. In this work, the Author Impact Assessment algorithm can be classified into two classes:Statistical Method and Methods based on Academic Network. In this paper, we have combined the h-index indicator with PageRank algorithm to avoid some limitations of both. In our method, the PageRank is taken as a standard for evaluation literature’s impact. Then, we transform the PageRank score of each publication into a score which similar to the citations of each publication. Finally, we calculate authors’ h-index according to their amounts of publications and publications’PageRank score. What’s more, we have taken the dataset which collected from the website of Microsoft Academic Research by using Data Mining keyword. Besides, the average rank of SIGKDD Innovation award winners from2000to2012has been taken as a reference stand. We have compared PR-index with other indicators (such as h-index and PageRank), the results shows that PR-index shows a higher average rank to the winners than others indicators.