Recommendations Of Educational Resource Crowdfunding Projects That Integrate User Interests And Ratings

Crowdfunding is a business model composed of project sponsors,the public and intermediaries.The educational resource crowdfunding project aims to raise the required resources for publishers.The sponsor is the platform user,the public is the other users of the platform,and the intermediary The organization is the platform for this article.With the increase of crowdfunding projects,many crowdfunding projects lack sufficient attention,and it is difficult for users to find suitable projects,and the success rate of the projects will inevitably decrease.Therefore,by creating a recommendation system to match user preferences and improve the success rate of the project,but the crowdfunding platform data is relatively sparse,the traditional collaborative filtering algorithm is difficult to work,and there will be cold start problems.In addition,educational resource crowdfunding projects need to consider the user’s ability to complete the project.After an educational resource crowdfunding project is recommended to users,it is not just for users to see it,but to help publishers raise resources and hope users upload resources;Also consider the limited time of the project.Crowdfunding projects are time limited.Try to help the publisher gather resources within the time limit.In response to the above problems,this paper proposes a personalized recommendation strategy that integrates user interests and ratings,and implements a personalized recommendation system based on this strategy.The main work of the thesis includes:(1)Introduce the user interest model and divide the user interest model into explicit user interest model and implicit user interest model.The explicit user interest model is established through the user’s historical behavior and item characteristics,which can better reflect user preferences.The implicit user interest model is established by text mining and feature extraction on the text information left by the user.As long as the user leaves enough text information such as comments and retrieval in other modules of this platform,the crowdfunding module can be recommended to improve "Cold start" problem.In addition,the introduction of the interest model can not only expand the statistical range of user similarity,thereby improving the "data sparse" problem,but also modify the similarity calculated by the original single scoring model to make the recommendation more reasonable.(2)Propose a collaborative filtering recommendation algorithm that integrates user interests and ratings,integrate the above models and rating models according to certain rules,and use the fused model as the basis for collaborative filtering recommendation,and design experiments based on public data sets to compare traditional collaboration The performance of the filtering algorithm and the collaborative filtering recommendation algorithm that integrates user interests and ratings,so as to verify the feasibility of the algorithm.(3)Design and implement a recommendation system for crowdfunding projects based on the above strategies,which can measure the user’s ability to complete the project based on the user’s uploading resource behavior and the publisher’s adopting resource behavior,and this is reflected in the scoring model;the introduction of crowdfunding project recommendation coefficients can be Flexible recommendations are made according to the remaining time of the project.The specific modules include a predictive user rating module,a platform data statistics module,and a recommendation business logic module.