While organizing conference programs, it is necessary to include articles that do not have common subject in the same sessions, to hold sessions containing articles on the same subject in parallel, etc. situations decrease the efficiency of the conference in terms of participants. It also may create a bad impression for conference owners and may affect the participation in future conferences. Therefore, it is very important that conference programs are created correctly in terms of content and scheduling. Currently, conference programs are organized manually by the organizers. This process is difficult and time consuming considering the number of articles in the conference and the breadth of the conference scope. Considering that today many problems are solved with software and hardware models and that dependency to human is minimized, this process, which is carried out manually, has been seen as a problem that needs to be solved automatically.
In order to facilitate the conference program preparation process and to create more efficient programs, two approaches are proposed in this thesis. The program organization process has been carried out in four stages: finding article similarities, creating sessions, determining session titles and scheduling. With the community discovery-based approach, sessions with different numbers of articles have been planned. In the second approach, sessions with an equal number of articles have been created with the proposed clustering algorithm. In addition, a conference program has been organized in which parallel sessions will not conflict with the proposed scheduling approach. The results have been compared in detail with the real programs and it has been shown that more efficient and effective conference programs are created automatically with the proposed approach. |