深度学习视角的FinTech风险知识获取与平台治理模型

Based on the context of intelligent era, this project puts forward a scientific problems：How to identification the risk knowledge and designing models of governance platforms of FinTech. The problem is a hot and important digital finance and intelligent knowledge management issues in the knowledge management analytics with a wide range of actual background. In order to improve the model of risk knowledge and platforms governance of FinTech, the project study on the mechanism of risk knowledge identifications in FinTech from the DL. Firstly, the project analysis the common features of source of FinTech markets and risk pattern of the characteristics of FinTech platforms source，the project specifically aims to address the gap of the relationship between risk and platforms of FinTech by DL; Secondly, the project research the risk knowledge identifications model and designing for risk artifact for FinTech with models of risk sensor, sentiment sensor, trust sensor, safety sensor of digital finance relevance based on risk knowledge map; Thirdly, present a relationship model for risk, sentiment, trust, safety in FinTech platforms and then answer the question: what is the mechanism of FinTech platforms governance. Finally, We propose a model of FinTech platforms governance which can identify risk knowledge based on DL; which can find risk knowledge and the mechanism of FinTech platforms governance. We verify these findings which will choice three types of FinTech platforms based on DL. The research contributions can provide the new methodologies of risk knowledge identification and the mechanism of designing a model of platforms governance; it can provide a apply model for FinTech knowledge management based on DL of intelligent context

本项目提出“智能时代如何识别金融科技风险与平台治理”的科学问题。该问题是金融科技与智能知识管理的热点问题，具有广泛的实际背景。为了揭示金融科技风险识别机理，首先，研究金融科技市场数据特征与风险特征，具体探索利用深度学习解释金融科技风险与金融科技平台之间的关系机理；其次，研究金融科技风险识别的知识模型：用设计科学设计金融风险传感模型，研究模型中的风险感知器、情绪感知器、信任感知器、安全感知器，用来获取金融科技风险知识与知识图谱；再次，探索金融科技中风险、情绪、信任、安全之间的相互影响机理；最后，构建金融科技平台治理模型。在研究中，拟选择金融科技中3类平台来验证理论成果。项目的价值：从深度学习视角探索金融科技风险知识识别方法；揭示金融科技风险知识管理机理和金融科技创新商务应用模式。

项目背景:智能时代如何识别金融科技风险与平台治理问题。.主要研究内容:研究了金融科技风险特征，金融科技风险与金融科技平台之间关系，金融科技风险识别知识模型：研究了金融科技中风险、情绪、信任、安全之间的相互影响机理与治理模型。.重要结果:成果提出了基于文本情感分析的风险分析模型，该模型用于识别在线借贷平台上问答文本的风险。成果基于无监督机器学习算法，从语义和内容角度，探索信息效应和风险效应对IPO市场的短期市场回报与长期风险规避的交叉作用机制。构建了移动支付领域的领域词库，并建立了能对移动支付文本内容进行自动摘要模型。研究成果提出了一种基于知识图谱和政策效应的移动支付风险研究方法，以获得更细粒度的移动支付风险知识。成果设计了借贷平台风险预警模型，并建立了基于多维度信息投资者情绪指数。成果在国内外重要期刊上发表论文34篇论文，包含SEPS、JKM、IJPR、ESA、AOR、ECR、计算机工程与科学、金融研究等。重要学术合作与交流合作交流30次。培养了年轻学者；培养了18名研究生毕业，2位研究生获得国家奖学金。.关键数据:投资问答对数据、借贷平台数据、移动支付政策数据和IPO数据.科学意义:从深度学习视角探索金融科技风险知识识别方法；揭示金融科技风险知识管理机理和金融科技创新商务应用模式。