ارایه یک سامانه توصیه‌گر مکانی شخصی سازی شده برای خدمات ورزشی

امروزه سامانه های توصیه گر مکانی، امکان تحلیل و توصیه خدمات مکانی را برای کاربران مختلف ایجاد کرده اند. قابلیت شخصی سازی ارائه خدمات به کاربران متنوع سیستم قابلیت بکارگیری و کارایی اینگونه سامانه ها را افزایش داده است. شخصی سازی توصیه ها در در حوزه مکان های ورزشی از جمله خدمات شهری است که در افزایش سطح سلامت شهروندان و جامعه می تواند بسیار موثر باشد. هدف از تحقیق حاضر طراحی یک سامانه ‌‌توصیه‌گر دانش مبنای شخصی سازی شده‌ای است تا فرد بر اساس نیاز‌‌ها و شرایط خود مناسب‌ترین و بهینه‌ترین موقعیت ها برای دریافت خدمات ورزشی را دریافت نمایند. جهت اجرای طرح پیشنهادی افرادمختلف براساس ویژگی‌‌ها و شرایط شخصی به کمک الگوریتم شبکه عصبی نگاشت خودسازمان ده خوشه بندی شدند و در ادامه جهت مدل سازی عدم قطعیت، ترجیحات افراد در موتور استنتاج فازی شهودی وارد و خروجی صریح و دقیقی به دست آمده و به کاربر ارائه گردید. تعداد 98 نفر در سامانه ثبت نام کردند و به 7 خوشه متفاوت تقسیم شدند. برای هر خوشه فرایند شخصی سازی برای ارائه خدمات انجام شد. در نهایت سامانه بر اساس سه پارامتر دقت (0.74)، حساسیت (81 /0) و f-score (77 /0) مورد ارزیابی قرار گرفت و نتایج حاصل حاکی از عملکرد خوب و قابل قبول رویکرد پیشنهادی دارد. 

developing a personalized location-based recommender system for sports services

today, location recommender systems have made it possible to analyze and recommend location services for different users. the ability to personalize the provision of services to various system users has increased the applicability and efficiency of such systems. the personalization of recommendations in the field of sports venues is one of the urban services that can be very effective in increasing the level of health of citizens and society. the purpose of the current research is to design a personalized knowledge-based recommender system so that people can receive the most suitable and optimal positions for receiving sports services based on their needs and conditions. in order to implement the proposed plan, different people were clustered based on their characteristics and personal conditions with the help of the self-organizing neural network algorithm, and in order to model the uncertainty, the preferences of the elderly were entered into the intuitive fuzzy inference engine and a clear and accurate output was obtained and given to the user. in this research, the knowledge recommender system has been designed as a personalized basis for providing sports recommendation services so that citizens can receive the most appropriate and optimal services based on their needs and conditions. in order to implement the proposed plan, the characteristics and personal conditions of people were obtained based on various online questionnaires and based on the unsupervised neural network of self-organizer mapping, people with similar characteristics were placed in specific groups, and in this way, and the basic knowledge base for the recommender system was created. finally, through the descriptive combination of the obtained outputs from the information of the users, optimal positions have been extracted for recommending and displaying the output of the system on the map, django library, which is a web-based programming platform, has been used. from the open-layer location library and j-query library, the desired location should be shown on the osm map as a map under mobile. after that, through the designed user interface, the personal characteristics of the people are received, then, the individual cluster is determined, and in order to model the uncertainty, the information received from the citizens is entered into the fuzzy inference engine of the second type. the outputs obtained from the designed algorithm will be displayed as a set of positions and the user can see its position on the map by selecting any of these positions. then, an explicit and accurate output of the user’s preferences has been achieved and a recommendation is made from the inflectional combination of the outputs obtained from the two designed sections, and suitable outputs are recommended to the user. finally, the system was evaluated based on three parameters: accuracy (0.74), sensitivity (0.81), and f-score (0.77), and the results indicate a good and acceptable performance of the system.