Quantexa

大資料服務提供商, 使用物體決議, 關系分析和人工智能技術幫助客戶進行資料處理和預防金融犯罪.

企業概覽

• 2016年成立, 當前規模500人
• 服務特色是場景決策智能CDI(contextual decision intelligence)
• 落地場景主要是金融機構的反洗錢反金融詐騙監控, 資料管理, 風控
• 解決的問題: 監管合規, 提高警告準確率, 降低成本, 提高行業競爭力
• 面向的主要客戶是銀行, 保險, 支付機構, 運營商(CSP)和政府機構, 已知客戶有匯豐銀行, 渣打銀行, 丹斯克銀行(丹麥), 紐約&梅隆銀行, OFX(澳洲支付機構)

時間軸

2016

• 2016-03
  • Founded, 15 people(6 financial crime experts). Work for anti financial crimes for HSBC, services: AML, people traffic, solve the data problems
• 2016-09
  • SWIFT Innotribe Chanllenge Winner

2017

• 2017-03 3.3m in Series A investment
• 2017-10 Microsoft Accelerator Programme Winner
• 2017-? Synechron became a customer

2018

• 2018-04 Featured in Financial Times
• 2018-04 Named in Tech Nation Future 50
• 2018-04 HSBC became a customer
• 2018-07 Open US office in NY and Boston
• 2018-08 30m in Series B investment
• 2018-09 100 employees
• 2018-? Danske Bank a successful pilot

2019

• 2019-02 Featured in The Times
• 2019-02 Host QuanCon
• 2019-03 Appeared on CNN(TV)
• 2019-05 Named "Cool Vendor" by Gartner
• 2019-07 Appeared on Sky(TV)
• 2019-09 200 employees

2020

• 2020-07 64.7m in Series C funding. The round was led by Evolution Equity Partners,
• 2020-09 Engagement with BNY Mellon

2021

• 2021-07 153m in Series D funding from Warburg Pincus and a growing group of blue-chip investors
• 2021-09 BNY Mellon has completed a strategic investment in Quantexa.
• 2021-10 Quantexa 2 release - easier deployment, simplify navigation, introducing contextual search for unstructured data

2022

• 2022-04 Quantexa 2.1 release, introducing Geospatial Search

Quantexa使客戶能夠從資料中做出更好的決策, 根據其網站介紹, 分為監控和調查兩個方向, 可能是同一個產品的兩個不同側重的說明.

場景監控 contextual monitoring

結合內部資料和外部資料構建關系網路，降低誤報, 提高速度和準確率, 并識別之前未發現的風險

• Enhance detection rates with advanced models that leverage network-based context to reduce false positives and generate more accurate alerts.
• Generate more meaningful alerts with context for investigators, leading to faster, trusted decisions.
• Find new, previously unknown risk from external sources to optimize future alert generation.

調查 investigations

借助可視化功能快速回應警報和資訊請求, 對每個客戶和交易對手創建單獨畫像以及實時的關聯和行為圖譜, 更快識別金融犯罪和欺詐風險.

• Automate manual work, and free up experts to focus on real risk.
• Create a true single view of each customer or counterparty, and a real-time network of relevant connections and behaviors.
• Go deeper and wider in your data to identify financial crime and fraud risks and typologies, faster.

涉及的服務明細

反洗錢 KYC & AML

KYC和AML是大部分國家都存在的金融業監管要求

• 交易監控 Transaction Monitoring, 對例外的賬戶交易發出預警
• 重點監控名單 Watch List
• 身份校驗 Identity Verification, 保管客戶的身份以及機構資訊，確保實際受益人資訊的準確性以及有效性
• 案例管理 Case Management
• 行為分析 Behavioral Analytics
• 風險評估 Risk Assessment, 交易是否涉及敏感國家或地區
• 客戶是否包括擔任重要公職的人員 PEP Screening, 受制裁或涉及任何負面新聞/媒體資訊
• 可疑行為報告 SARs (suspicious activity report)
• 調查管理 Investigation Management
• 合規報告 Compliance Reporting

欺詐檢測 Fraud Detection

• 自定義欺詐引數 Custom Fraud Parameters
• 模式識別, 銀行業/保險業 Pattern Recognition: for Banking, for Insurance Industry
• 調查記錄 Investigator Notes
• 支票欺詐監控 Check Fraud Monitoring
• 內部欺詐監控 Internal Fraud Monitoring
• 權限安全管理 Access Security Management
• 針對電商和數字貨幣的交易審核 Transaction Approval: for eCommerce, for Crypto

資料管理 Master Data Management

• 關系映射 Relationship Mapping
• 資料屏蔽 Data Masking
• 流程管理 Process Management
• 可視化 Visualization
• 匹配和合并 Match & Merge
• 層級管理 Hierarchy Management
• 資料源集成 Data Source Integrations
• 多領域/多模型 Multi-Domain
• 資料治理 Data Governance
• 元資料管理 Metadata Management

產品介紹

以上服務和解決方案的載體為 Quantexa Syneo 平臺. 當前(2022.04)最新版本為2.1

產品明細

Quantexa利用大資料和人工智能技術，發現潛在的客戶聯系和行為，以解決金融犯罪、客戶洞察和資料分析方面的需求

快速資料匯入 Rapid data ingestion

• 可擴展, 高性能的資料訂閱(匯入), 不需要復雜的ETL; 對現有的資料和結構進行自動判斷, 配置, 清洗, 決議和標準化; 開箱即用, 帶默認的物體定義和屬性設定, 帶預先訓練好的模型

• 可以接受結構化, 非結構化和半結構化的輸入資料; 匯入時驗證資料欄位, 識別問題; 提供UI使用戶能夠進行操作并解決問題

• Quantexa 為其客戶提供了許多分析模型, 目前可用的模型包括資本市場反洗錢(包括外匯、股票和貴金屬), 融情報機構評分, 減少誤報, 貿易反洗錢, 客戶畫像評分, 證券反洗錢檢測, 貿易融資欺詐, 信用卡申請欺詐等

• Quantexa 還提供定制建模和技能培訓服務.

• Use Quantexa Fusion to model complex source data and ingest it fast with no-code, scalable, high performance data preparation and ingestion – and no complex ETL.

• Automatically infer, configure, cleanse, parse and standardize potential linking attributes from existing data schema.

• Get started quickly with out of the box, state-of-the-art AI-tuned models. Define entities and their attributes.

物體決議 Entity Resolution

• Quantexa的物體決議功連接內部和外部資料得到更好的準確率, 甚至對于沒有唯一關鍵詞的資料也能得到較好效果; 定義和創建人, 業務, 地址等各種資料資產并輸出給批量和流水線處理

• 最終用戶可以深入到一個物體中，查看不同的資料記錄如何以及為什么被匹配到同一個物體中. 用戶可以動態調整決議匹配邏輯.

• Connect internal and external data sources with unprecedented accuracy, even from poor quality data without unique match keys.

• Create data assets for people, businesses, addresses and more, and expose them through batch and real-time data pipelines.

關系圖譜 Network Generation

使用圖展示物體之間的真實關聯, 這些關聯包括供應鏈, 合作伙伴, 法律層級, 社會關系等; 基于動態物體決議為不同的場景, 并生成不同的關聯; 挖掘用戶, 機構, 地址和交易之間的關聯

• Use to generate graphs that link entities into relevant, real world networks representing supply chains, associates, legal hierarchies, social connections and more.
• Build on dynamic entity resolution to generate different networks for different use cases.
• Reveal the context of how people, organizations, places, and transactions relate to each other.

關聯(場景)分析 Contextual analytics

• 使用Quantexa Assess(可能是Syneo內部的一個資料資產管理模塊, 外部并無單獨介紹)創建和維護資料關系模型; 為機器學習和AI服務的物體圖譜分析工具.

• 客戶能夠匯入外部檢測模型或使用他們自己喜歡的分析環境, 如KNIME, R或Python. 建模方法促進了透明性和可解釋性，并且可以批量或實時運行.

• Use Quantexa Assess to empower data scientists to build and maintain their own contextual models with ease.

• Productively engineer features for machine learning and AI with native support for entity graphs and networks to build robust features for machine learning and AI.

Quantexa支持的機器學習演算法和適用場景

可視化和探查 Visualization and exploration

• 調查人員可以搜索平臺獲取的各種客戶和交易資料

• 界面支持上千用戶同時操作, 進行快速和精確的合作決策. 界面支持可視化探索和分析, 創建標簽, 高亮感興趣的資料; 同時提供API給第三方系統如CRM等進行集成

• 資料隱私合規: Quantexa具有限制對客戶資料訪問的能力，以允許其客戶遵守當地的資料隱私要求，當調查人員與物體和圖譜互動時，他們只能根據用戶的權限查看資料.

• Support thousands of users with faster, more accurate, collaborative decisioning using Quantexa’s UI to search, visualize and explore context; investigate and thematically analyze; and review analytically created flags within their context, highlighting points of interest.

• Or, use Quantexa’s APIs for external application platforms including CRM and case management.

作業流程

資料匯入和管理

場景分析和調查

產品技術堆疊

語言

• Scala
  Quantexa Syneo的主要開發語言
• Python
  資料作業者常用語言, 用于機器學習以及資料處理
• R
  資料作業者常用語言, 函式豐富, 常用于科學計算, 統計和資料分析, 作圖

存盤

• PostgreSQL
  中小型關系資料存盤
• Oracle
  大中型關系資料存盤, 商業軟體
• Hadoop/Hive
  大型分布式存盤和處理, 用于時效性要求不高的計算任務, 猜測在這個產品中主要用于給Spark Streaming提供存盤
• Elastic
  資料檢索引擎, 支持分布式集群
• Apache Spark, Spark Streaming
  資料處理引擎, 支持容錯的高吞吐量實時流資料處理, 可以運行在Hadoop或Google Cloud, Kubernetes之上, 使用記憶體計算, 速度較快
• Apache Kafka
  訊息佇列, 流式資料管道, 用于在Spark前接收和暫存資料

容器

• Redhat Openshift (Kubernetes)

第三方服務

• Google Cloud Storage
• Google Cloud SQL
• AWS
• Azure
• Salesforce

界面展示

暫時只能搜索到圖譜分析部分的界面

這兩個是版本2.1中新增的地理位置分析功能

市場驅動

監管需求 Regulatory requirements

for financial firms’ ability to detect money laundering continue to mount. The price of failure is hefty fines (banks worldwide have paid several billion dollars in fines for AML lapses since 2010), embarrassing headlines, and potential liability for the firm’s chief AML officer in the form of personal fines and even jail time.

創新需求 Innovation

in financial services is creating an ever-growing attack surface. Faster payments and the increasing electronification of payment flows create utility for businesses, but criminals benefit from these innovations as well.

客戶期望 Customers’ expectations

for a smooth and easy experience put pressure on firms to reduce lag time and friction across the customer life cycle. These expectations start at the onboarding process and extend throughout the customer journey.

歷史遺留技術升級壓力 Legacy technology

that produces high volumes of alerts, false positives, and often false negatives compounds the challenges that banks face. Banks often have to throw bodies at the problem to keep up with alert volume. This is not only expensive but often problematic in terms of finding skilled analysts to fill these positions.

輿論壓力 Social pressure

from citizens who feel that banks, as trusted custodians, have an ethical obligation to detect and intercede in money laundering, human trafficking, and fraud incidents

市場趨勢 Trends

針對銀行的犯罪攻擊技術在不斷升級 Escalating criminal attacks on banks use advanced technology.

Organized crime rings, rogue nations, and terrorists are all leveraging automation and artificial intelligence in their attacks on the financial ecosystem. These sophisticated attacks, combined with the growing volume of electronic payments, make it ifficult for FIs to keep pace with the rising tide of alerts.

監管機構希望金融機構升級技術協助其更好提升情報能力 Regulators are encouraging FIs to use more sophisticated detection techniques.

Especially in the AML arena, concern over regulatory response to the use of advanced analytics has been an inhibitor to adoption. The new openness among regulators is encouraging FIs to invest in technology that can help them extract intelligence from their customer data.

銀行希望提高運營效率 Banks are looking for operational efficiencies.

While many FIs initially turned to outsourcing first- and secondlevel alert triage to less expensive offshore locations, the benefits of these strategies were short-lived, as alert volumes continue to multiply. Many banks are now focused on tackling the source of the issue—dirty source data and high levels of false-positive alerts.

新技術的采用給銀行等金融企業創造競爭優勢 Adoption of next-generation financial crime technology is creating competitive differentiation.

Firms that use advanced technologies to vet customers’ identities and transactions differentiate themselves from their competitors, as they provide more responsive and streamlined customer interactions, improve their operational efficiency, and meet regulatory requirements.

參考

1. Official site https://www.quantexa.com/
2. 2019-08-05 Jamie Hutton, chief technology officer at Quantexa, about building a culture of compliance within the banking industry.
   https://www.youtube.com/watch?v=X5vaAGfytA8
3. 2020-03-02 Ian Lees is the Head of Research and Development at Quantexa, he gave an introduction to Quantexa (our hosts) at the start of this months Scala in the City, Lightbend Edition
   https://www.youtube.com/watch?v=f5A1R_JCvqA
4. 2020-07 Quantexa Raises $64.7M to Drive Growth in Big Data and Analytics Ecosystem
   https://www.datanami.com/this-just-in/quantexa-raises-64-7m-to-drive-growth-in-big-data-and-analytics-ecosystem/
5. 2021-03-09 Jennifer Calvery, Head of Financial Crime HSBC. How HSBC Uses Technology To Combat Crime. See how HSBC is using technology to manage its data effectively and improve financial crime detection to tackle horrific crimes, from terrorist financing and human trafficking.
   https://www.youtube.com/watch?v=JmnI2K6OVNg
6. Follows a successful 12-month engagement with BNY Mellon using Quantexa's platform and includes an expanded relationship focused on data fabric innovation at the bank
   https://www.prnewswire.com/news-releases/bny-mellon-invests-in-quantexa-technology-301388579.html
7. OFX with Quantexa: OFX is an Australian foreign exchange and payments company https://cloud.google.com/customers/ofx-quantexa
8. Case of using Quantexa https://thefinancialcrimenews.com/why-illegal-trafficking-in-organs-is-growing-fastbut-few-are-talking-about-itby-steve-farrer/
9. Dun & Bradstreet partner with Quantexa https://www.dnb.com/solutions/partner/quantexa-partners-detail.html
10. Positive, PR service provider for Quantexa https://www.positivemarketing.com/case-studies/quantexa/

標籤：大數據

上一篇：SceneKit–遞回地克隆一個節點

下一篇：Druid SQL和Security在美團點評的實踐