Ανάπτυξη ενός συστήματος συστάσεων κριτών επιστημονικών άρθρων με βάση τα προφίλ κριτή-άρθρου και χρήση μεθόδων πολυκριτήριας ανάλυσης και μηχανικής μάθησης

Abstract

One of the most important problems is the development of valuable user profiles (consumers, products, paper reviewers, ...) with huge fields of application in e-business, e-governance, etc. On the other hand, Artificial Intelligence (AI) in combination with Big Data, has developed rapidly, making it possible to utilize appropriate classifiers and large amounts of data to develop user profiles.

Within the framework of the thesis, software was developed that implements AI methods and special algorithms, such as machine learning, data mining, text analysis, natural language processing, etc., and multi-criteria decision analysis, in order to form the appropriate profiles of papers’ reviewers and papers, and then proceed to evaluate the paper-reviewer combinations and formulate the proposal of suitable reviewers per paper.

Τo create profiles of user and papers, a hybrid methodology has been developed that is based on direct and indirect feedback, that is, on information provided by both the user himself and his digital footprint (personal websites, published papers, etc.).

Finally, the above methodology was applied to a series of cases to evaluate its effectiveness.