预言性的基于事例推理技术

预言性的基于事例推理技术（英文）

Predictive Case Based Reasoning

Despite heavy investment in data management and monitoring platforms, the financial services industry still lacks real-time operational intelligence to enable better business decision-making and prevent systems and service failures and catastrophic trading errors. These outages expose institutions to undue risk and compliance violations that can cost organizations millions of dollars in financial losses and regulatory fines. They also undermine investor confidence and damage firm reputation.

Modern financial markets have become more complex that ever fueled by the globalization of capital markets, including a variety of new securities, derivatives and indexes, the evolution of high-frequency trading platforms with millisecond execution windows, more stringent regulations and higher levels of interconnection among different players. This increased complexity is overwhelming legacy systems, resulting in overlooked information and missed opportunities to uncover hidden patterns, relationships and dependencies. Markets can quickly and easily be destabilized by external or internal shocks that spread rapidly through massive electronic communication and transaction systems. The fact that participants often try to disguise their strategies only adds to market complexity.

The availability of intricate layers of market data has markedly lifted the quality of trading and risk management capabilities in recent years, enabling clearer identification of problems and faster resolution of market exposures considered far too unacceptable just a few years ago. Yet despite the wealth of data and content, most business users seek to gain even easier access to information they need in a timely fashion to stay ahead of their competition.

The sum-of-the-parts approach of traditional analytics methodologies that breaks systems into their component parts for individual analysis is ineffective. Analyzing or mitigating risk in only one component of the system sometimes feels like progress but does little to prevent truly disastrous events or failures. In fact, errors or failures can be amplified, as one component affects another and then another, spreading risk throughout the system or market. To better understand the properties of the components and their place in the overall system requires a higher-level assessment of the relationships to each other as well as to the wider system and environment.

Getting Predictive with Case-Based Reasoning

Predictive analytics platforms enable organizations to leverage all enterprise data – from historical structured market data to newer forms of unstructured big data – to drive faster, more informed decision-making and provide preemptive warnings of systems failure. Users build sophisticated mathematical models to explore the relationships among these variables to uncover previously hidden patterns in the data, identify classifications, make associations and perform segmentation. While many of these techniques are not new, advances in underlying technologies – from multi-core and parallel processing to faster and larger data stores that can keep entire databases in-memory – are enabling real-time analysis on massive data sets of current and past activity to predict future scenarios.

Case-based reasoning (CBR) is a type of predictive analytics that uses machine learning to solve current problems with knowledge gained from past experience. A CBR-driven predictive analytics engine seeks patterns by automatically and continuously comparing real-time data streams of multiple heterogeneous data types. To pre-emptively direct the user to the most appropriate decision or action, a self-learning case library adapts past solutions to help solve a current problem and recognizes patterns in data that are similar to past occurrences.

Using CBR, systems can learn from the past and become more adaptive. For example, if a system begins exhibiting a pattern of anomalous trading behavior, CBR searches for past cases of similar patterns and issues an alert for action before the pattern escalates into a full-blown hazardous event. A deeper understanding of past cases provides the context for markets and participants to prevent technical failures instead of just responding to them.

CBR technology has multiple use cases in financial services. The most significant opportunity may be for CBR to serve as an early warning system for market operators and participants to prevent disruptions caused by outages, trading errors, improper systems oversight or other compliance violations. Capital markets institutions can deploy CBR to monitor and deter abnormal client behavior, detect risk exposures through internal or external fraudulent activities (in areas such as trading or client interaction), improve IT operational efficiency in the back office and uncover customer-facing opportunities to generate new revenue.

译文：

预言性的基于事例推理技术

我们为了更好的进行业务决策和防范因为信息系统和服务系统出现的故障，从而导致灾难性错误，因此对各种数据管理和信息监控平台进行了大量的投资，但是整个金融服务行业依然缺乏实时性的智能运营能力。一旦发生错误或者是因为公司违规操作而造成的意外中断，可能会带来的是数百万美金的损失以及监管机构的开出的巨额罚单。并且还会损坏公司的形象以及破坏了投资者的信心。

现代社会，因为资本市场的全球化工作不断推进，包括各种新的证券、衍生性金融商品和金融指数的产生，各种高频交易平台的窗口期都以毫秒级速度在执行，外加更加严格的法规制度以及更高水平的参与者，让现代金融市场越发的复杂。这日益复杂的市场中，产生的信息对旧有系统产生压倒性的挑战，往往导致投资者和公司都因为忽略了某些信息和信息中隐藏的一些模式、关系和依赖情况，从而错过了机会。市场通常是很敏感的，来自内部或者外部的冲击，就会很容易的导致整个市场的不稳定，并且能够通过大规模的信息通信和交易系统迅速的蔓延波及到整个市场。事实上，参与者还会试图对自己进行伪装以增加市场的复杂性。

近年来，因为利用数据在技术层面对复杂的市场在分析和识别，有了很大的进步，所以对交易和风险管理的能力有了较大的提高，并且能够快速的解决一些市场风险，这些在几年前是几乎不太可能实现的。然而尽管大多数企业都迫切需要获取更加及时的信息，并且它们也能够轻易获取到非常丰富的数据和内容，以期保持领先于对手，但是他们的竞争对手确更容易获取这些信息。

传统的分析通常是采用分类加总估值法（一种给多元化控股公司估值的方法，将公司同时经营的不同业务分别选择合适的估值方法估值，再根据持股比例加权汇总得出该多元化控股公司的总价值。）来进行，但是把组成系统的各个部件都打破来进行个体分析，本身就是一种效果很差的方法。用分析来减缓系统的某一个组成部分的风险，有时候感觉像是像是一种进步，但是丝毫无助于防止整个系统发生灾难性的事件和故障。事实上，错误或者故障可能被无限放大，因为一个部分可能会影响另一个部分，然后是继续影响其他的部分，风险在整个系统或者市场内是传播和分散的。为了更好的理解每一个部分在其整个系统中发挥的作用和特性以及他们之间的相互关系，就需要在一个更广泛的系统和环境之间构建一个更高级别的评估方法。

基于案例推力的预测方法

预测分析平台使企业能够充分利用所有的企业数据——从历史的结构化市场数据到新形式的非结构化大数据，来驱动我们更快更明智进行决策，并且还可以针对系统故障进行预先的告警。用户通过构建各种复杂的数学模型来探讨各种变量之间的关系，以揭示以前隐藏在数据中的一些模式，识别各种分类，进行关联分析和执行精准细分。虽然其中的很多技术都不是最新的，但是这些基本技术也在不断进步——利用多核和并行处理技术，可以将整个数据库加载到内存中，以实行更大的数据存储和更快的分析——这种技术使得能够在当前和过去生成的大量数据集上进行实时分析，以预测和分析未来的场景。

基于案件推理的技术（CBR）是使用机器学习从过去的经验中获取相关知识，以解决当前问题的一种预测分析技术。案例驱动预测分析引擎通过自动的访问和比较由多个异构数据类型组成的实时数据流，以发现一种模式来解决问题。自学习案例库自动去适配过去的解决方案，并认识到这个问题类似于过去出现的一些模式，可以帮助用户来解决当前的问题，而抢先引导用户做出做适当的决定或者是行动。

使用CBR技术，系统可以从过去的信息中去学习，以更加适应当前的情况。例如：如果一个系统开始表现出了反常的交易行为模式，那么CBR通过搜索符合这种模式的过去的案例，那么在这个行为升级为一场全面的危机之前，就发出警报以采取行动。这种为市场机构和参与者提供对过去更为深入的分析和了解的技术，可以避免旧系统只是简单的回应是否出现了技术故障这种的问题。

CBR技术已经在金融领域里面有多个实际案例。最著名的案例是采用CBR作为市场监管者和参与者服务的防止因意外停电，交易失误，不恰当的系统监管或者其他违反规定而意外中断的早期预警系统。资本市场的机构可以通过部署CBR系统，来监控和阻止不正常的客户操作行为，检测任何通过内部或外部进行欺诈活动的风险敞口（风险敞口（risk exposure）指未加保护的风险，即因债务人违约行为导致的可能承受风险的信贷余额， 指实际所承担的风险，一般与特定风险相连。）（诸如在交易的过程或者客户私下互动的活动中），可以提高IT的运营效率，在交易后台就可以为客户发现新的机遇以产生新的收益。