Recommender system for an academic supervisor with a matrix normalization approach

Kazakovtsev, Vladimir; Oreshin, Svyatoslav A.; Serdyukov, Alexey; Krasheninnikov, Egor; Muravyov, Sergey; Bezvinnyi, Albert; Panfilov, Alexander; Glukhov, I.A.; Kaliberda, Yulia; Masalskiy, Daniil; Podolenchuk, Timofey; Khlopotov, Maksim

doi:10.1145/3437802.3437817

Cited by 5 publications

(3 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Their paper provides a system for selecting an academic supervisor that is based on a comparison of scholar interests and the scienti c contributions of a potential supervisor from the faculty members. They took a novel strategy to calculate similarity by focusing on Scopus quality criteria rather than coauthorship networks (Kazakovtsev et al, 2020). Each scientist is pro led based on his accomplishments in several elds of science.…”

Section: Literature Reviewmentioning

confidence: 99%

A Machine Learning Approach for Advisors to Discover in Higher Education

Ahmed

Rehman

2022

Preprint

View full text Add to dashboard Cite

In higher education, an institute thesis is under the supervision of a scholar with a comparable research interest. An appropriate match has to be made between the topics of scholars and the area of scholar expertise and skills. The determination of the final thesis supervisor is an important factor in the work of the scholar thesis. A system is needed for the information of the scholar and scholar matching for the determination of the supervisors. As currently, the HEC supervisor directory is there, but it is not up to date. There are a lot of supervisor recommendation systems like the cosine similarity technique uses keywords from a text as a metric to calculate the similarity of two scholars in different categories and many others. But theses recommendation system is working on the parameter of supervisor personality matching, criteria and sub-criteria of advisor system and some other demission’s relevance, connectedness, and quality. But, they do not focus on the similarity matching of their area of interest, w.r.t their publications and citations. Thus, a system is needed for the recommendation of the supervisor matching the relevancy according to their domain area, number of citations and publications. This paper explores a system of “SRAF” that assists scholars in finding the best-matched scholar. This proposed framework is different from the existing models because the dataset implemented has some additional features through which it matches the appropriate supervisor like number of area of specializations, publications, and citations. The framework matches the area of interest recommended w.r.t supervisors’ publications and citations to find the best supervisors according to the scholars’ research topic. The success of scholar research not only determined by the supervisor chosen but is also greatly influenced by the performance of the scholar and intellect, aspects which in turn affect higher education.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

A Machine Learning Approach for Advisors to Discover in Higher Education

Ahmed

Rehman

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Proper supervisor is a very important factor that will determine the success of the student [5]. Many studies have been done to create a recommendation system for selecting supervisors like in [6][7][8][9][10][11][12]. All of these researches only take student preferences without considering the supervisor's preferences, which we called a one-way recommendation system.…”

Section: Introductionmentioning

confidence: 99%

Two-way Recommendation System for Supervisor Selection using Historical Data and Skyband-view Queries

Annisa Annisa,

Muhammad Rayhan Adyatma,

Global Ilham Sampurno

et al. 2023

ARASET

View full text Add to dashboard Cite

Completing the final project on time is one indicator of success in university. Unfortunately, many students have not been able to complete their studies on time. One of the factors that influence this is the failure in choosing the right supervisor. This study aims to create a recommendation system that can help students choose suitable supervisors. Unlike other studies, this research builds a two-way recommendation system that takes into account the preferences of students and supervisors. The previous study used the skyline views query concept to recommend dominant objects. However, the skyline view query concept has a major limitation: only skyline objects will be recommended to the user. Thus, students who are not skyline objects may not get a supervisor's recommendation, and vice versa. In this research, we use the concept of a skyband view query to overcome the limitation of the skyline view query. In addition to answering eight important queries from both parties, students and supervisors, the skyband view query concept is also able to overcome the shortcomings in previous research. Historical data from alumni is used to construct students' and supervisors' interests. This research succeeded in expanding the choice of research topics given in previous studies, as well as increasing the number of recommended supervisors and students.

show abstract

“…The relevance of the development of this method is proved by the insufficient theoretical and practical basis for solving the problem of choosing partners, taking into consideration a complex structure of scientific projects in certain scientific subject space. In particular, paper [3] describes the system of selection of a scientist, taking into consideration scientometric indicators of the Scopus base using the weighed influence of a scientist in a particular field of knowledge. The use of scientometric indicators makes it possible to take into consideration the dynamics of scientists' productivity in specified subject space.…”

Section: Introductionmentioning

confidence: 99%

Devising an individually oriented method for selection of scientific activity subjects for implementing scientific projects based on scientometric analysis

Kuchansky

Гладка

2021

EEJET

View full text Add to dashboard Cite

The main factors influencing the choice of individual subjects of the scientific activity or potential partners and executors for scientific and educational projects were analyzed. The specific features of choosing project executors of different categories were indicated. The functional responsibilities of project participants in accordance with the project structure were described. The individually oriented method for choosing subjects of scientific activity as executors of scientific and educational projects was developed, taking into account the productivity of their scientific activities in the past and considering the structure of projects. To determine the merits of the subjects of scientific activity, which are included in the relevant scientific subject spaces, it is necessary to apply the procedure of their productivity assessment. In addition, it is necessary to predict a change in productivity in the future based on retrospective data for this subject. Next, it is required to solve the multi-criteria problem of the choice among the subjects of scientific activity who are quite productive in the opinion of the project manager. The use of the developed method reduces the subjective impact on making a decision regarding the choice of project executors. This is due to the fact that they are chosen by automated calculation of scientometric indicators of subjects, guided only by open sources of information. The individually oriented method for the selection of subjects of scientific activity was verified on the example of the formation of three applications of research projects. As a result, the average percentage of scientists who meet the requirements of project managers for each scientific subject space was about 46.55 %. The percentage of those involved in the project from those who were selected is about 24.07 %. The probability of cooperation is higher among those who have an average H-index.

show abstract

Recommender system for an academic supervisor with a matrix normalization approach

Cited by 5 publications

References 9 publications

A Machine Learning Approach for Advisors to Discover in Higher Education

A Machine Learning Approach for Advisors to Discover in Higher Education

Two-way Recommendation System for Supervisor Selection using Historical Data and Skyband-view Queries

Devising an individually oriented method for selection of scientific activity subjects for implementing scientific projects based on scientometric analysis

Contact Info

Product

Resources

About