Vocational recommendation system with collaborative filtering method

Career choice is an important decision point in the process of determining people's future careers. Currently, tests that measure the personality and abilities of high school students are generally performed to help them choose a career. In addition to these tests, counselors can also make suggestions based on the student's academic success. In this thesis, a different recommendation system has been developed that can help young people in choosing a career. This system aims to perform similarity analyzes using the answers given to survey questions by professional people, university students and high school students. According to these analyses, the professions of the adults who gave similar answers at the highest rate (if they are happy to pursue those professions) are offered as suggestions to the young people who answered the survey. Among the questions in the surveys, in addition to the questions in personality determination tests such as the RIASEC test and questions to understand which of the courses taught in high school they like, there are also additional questions for university students to understand whether they are satisfied with the department they study, for professionals, whether they are satisfied with the profession they do, and how they make their choices. Currently, if the personality traits determined according to the RIASEC test are realistic (R), for example, some predetermined professions that may be suitable for this personality type are recommended. In the collaborative filtering-based recommendation system we have developed, a career recommendation is made according to the "highest similarity" by using a model based on cosine similarity between "high school students" and "university students + professionals" based on both the questions in the RIASEC test and the answers to questions about which courses they like. In this thesis, a hybrid vii method that can be used in the "cold start" situation where data is scarce is also proposed. In this method, the professions suitable for the personality determined according to the results of the RIASEC test were kept in one set, the professions of people with high similarity obtained through cosine similarity, based on the data obtained from questions about which of the courses they liked in high school, were stored in another set, and the professions at the intersection of these two sets were recommended. Although the analysis results could not be obtained with the desired accuracy due to the lack of high participation in the survey (such as lack of participation from some professions or only 1-2 people), this Profession Recommendation System provided a valuable contribution that can form the basis for future research and applications.