基于动静态表征的众筹协同预测方法

众筹是一个新兴的互联网金融平台,项目的发起者可以通过使用互联网,征求大量平台用户的资金来资助他们的项目.但是由于众筹平台所具有的独特规则,只有在特定时间内收集了足够的资金,项目的筹资才会成功进行交易.为了防止项目发起者和投资者在可能失败的项目上浪费时间和精力,动态追踪众筹项目的筹资过程以及估算其融资成功概率便极为重要.然而,现有的一些工作既没有针对动态预测跟踪机制的研究,也没有考虑平台上的项目发起者和投资者之间的动态行为交互.为了解决这些问题,基于长短期记忆网络设计了一种新颖的动静态协同预测模型.该模型着重分析了用户行为,包括评论的情绪倾向以及融资过程中的动态增量信息,从而将融资项目与投资人之间的交互行为进行深度挖掘分析.首先,针对平台上的静态特征和动态用户行为数据,通过不同的Embedding方法得到他们的深度表征.在此基础上,进一步设计了基于注意力机制的协同预测模型,以便了解项目融资的时序信息对最终结果的影响程度.最后,在真实的众筹数据集上进行的大量实验结果表明,所提出的动静态表征预测方法相比其他预测方法更为有效.

Cooperative Prediction Method Based on Dynamic and Static Representation for Crowdfunding

Crowdfunding is an emerging finance platform for creators to fund their efforts by soliciting relatively small contributions from a large number of individuals using the Internet. Due to the unique rules, a campaign succeeds in trading only when it collects adequate funds in a given time. To prevent creators and backers from wasting time and efforts on failing campaigns, dynamically estimating the success probability of a campaign is very important. However, existing crowdfunding systems neither have the mechanism of dynamic predictive tracking, nor consider the dynamic interaction between project sponsors and investors on the platform. To address these issues, we design a novel dynamic and static collaborative prediction model based on long and short-term memory network. This model focuses on user behavior, including the emotional tendency of reviews and the dynamic incremental information in the financing process, so as to deeply mine and analyze the interaction between financing projects and investors. Firstly, for the static features and dynamic user behavior data on the platform, their deep characterization is obtained by different embedding methods. On this basis, a collaborative prediction model based on attention mechanism is further designed to understand the impact of timing information of project financing on the final results. Finally, experiments on real crowdfunding datasets show that the proposed dynamic and static representation prediction method is more effective than other prediction methods.