Crowdsourcing Data Quality Control And Result Recommendation

With the rapid development of Internet technologies, services of convenient query and sufficient selection make the online travel industry occupy a place in the Internet market. Nowadays, however, in the era of pursuing "personalization", online travel sites which mainly provide tickets and hotel booking services merely have lost a significant competitive advantage in the market. Most mainstream online travel sites could not escape the fate of performance losses, and it has become the irrefutable fact that online travel industry is confronted with the bottleneck of development. At present a growing number of tourists are eager to get rid of the restrictions of package tour, and not willing to spend extra time planning travel routes. The tourists would prefer recommendation from enthusiastic natives or even company. Therefore, in order to meet the demand of tourists which is increasingly personalized and dynamic, and to provide new direction of development for the online travel market, the thesis attempts to apply crowdsourcing, the new model in the Internet, to the service of the online travel, trying to innovate the business process. As a kind of new production organization pattern, the crowdsourcing model is based on personal choice, collecting the public’s knowledge, skills, information and technique to solve the complex and diverse problems. Now, there are a lot of successful examples of application, such as Wikipedia and so on.In recent years, more and more scholars both at home and abroad have paid much attention to the researches on crowdsourcing model, and made a certain progress. The state-of-the-art research mainly focuses on the problem of quality control and recommendation of crowdsourcing results. Taking the personalized online travel services as an example, the thesis designs a novel personalized tours system based on the crowdsourcing model, called CrowdTravel. Tourists can obtain the travel plan that satisfies their personalized needs through the system, and the travel plan is provided by at least one workers. The worker can be a part-time local guider, or taxi driver, etc. Meanwhile, with the purpose of making the best of mass resources on the Internet, getting rid of the negative factors existing in the traditional tourism service, and improving users’ experiences, based on the application field of crowdsourcing, the thesis designs a kind of tourism crowdsourcing tasks, aiming to achieve the goal of crowdsourcing quality control. And, it also designs a personalized recommendation method based on implicit behaviors of tourists with combined crowdsourcing results to make effective use of the results. In addition, an optimized recommendation scheme based on super sphere in multi-dimensional space is devised to improve the efficiency of personalized recommendation for crowdsourcing results. At last, through the results of simulation experiments, we experimentally prove the effectiveness of the quality control strategies, personalized recommendation method for crowdsourcing results and its optimized strategy.