Research On Multi-objective Optimization Method Of Ridesharing Matching Based On Social Network

The way of ridesharing is greatly convenient for people's lives,and the popularity of location-based social networks has also injected new ideas into the development of ridesharing.Rich social resources in LBSN provides diverse data support for ridesharing,helping users to match the satisfaction of their shared partners,thereby the quality of the group's travel in ridesharing will be improved and the attraction of ridesharing can be increased indirectly.Ridesharing matching with social attributes needs to consider the experience constraints,consequently,leading to decrease in ridesharing matching rate.Meanwhile,the total experience and the matching rate of ridesharing need to be considered comprehensively in the ridesharing group discovery.However,the existing methods are difficult to achieve optimal match.Based on the fact illustrated above,this paper aims to improve the quality of group's experience in ridesharing and increase the matching rate of ridesharing.Correspondingly,the multi-objective optimization of rideshairng matching can be achieved.This paper conducts research from three aspects which can be listed as follows:(1)A social similarity group discovery method based on interest and trajectory is proposed.The method integrates user's social interest and trajectory information,and characterizes user's multiple characteristics by constructing a social heterogeneous information network.As a result,the user's social similarity can be calculated and the similar groups can be found.Experiments show that this method is of precision,meanwhile,the accuracy of the recommendation in destination and the total experience of ridesharing can be improved.(2)A potential destination recommendation method based on location prediction is proposed.This method learns user's history POI trajectory based on V-PPM location prediction algorithm.By predicting POI-type of the next destination and combining the preferences of the social similar group,this method can recommend time-sensitive destinations for users.Experiments show that the method increases the matching rate of ridesharing by a certain acceptance rate.(3)A multi-objective optimization of ridesharing matching method with social attributes is proposed.The method combines the result of discovery in social similarity group and the outcome from potential destination recommendation.Also the candidate ridesharing group can be found under the constraints in space-time and experience.Then a multi-objective function will be constructed,at the same time,optimal ridesharing group can be generated through SA optimization algorithm.Experiments show that the method improves the quality of total experience and increases the matching rate of ridesharing.The final result of ridesharing group matching will be optimized with the method mentioned above.