More or less heard of or experienced the risk control decision engine. If you ask, the biggest use of the decision engine is to liberate the pressure of development students and reduce the complexity of the system. When it comes to decision-making engines, big data is indispensable everywhere.
The decision engine originated from rule engine . The rule engine originated from the rules-based expert system, and the rules-based expert system is one of the branches of the expert system. The expert system belongs to the category of artificial intelligence . It imitates human reasoning methods, uses tentative methods to reason, and uses terms that humans can understand to explain and prove its reasoning conclusions.
The development of the entire decision-making engine, with the development of big data technology in recent years, the processing, analysis, mining and decision-making of massive data has become an urgent need. Based on the traditional rule engine, functions such as data measurement, data modeling, simulation, data visualization, champion/challenger (AB testing) have emerged. Some decision engines have even added machine learning functions, making it necessary for data Scientists and deep modeling tools have also become more inclusive. The decision-making engine evolved "machine replaces manual operation" to the realm of "machine replaces manual decision-making".
. In terms of the entire risk control segmentation content, we can divide the entire risk control decision process into four modules: minimum feature factor, risk control rules, strategy set, and engine. The following are the relevant contents:
Age
Gender
AABon number
BBBon number
20Age 60
Gender ==Male and 20Age 55
Gender ==Female and 20-age 60
B long number 5A long number 6
3. Strategy set:
New household access strategy set
Old household access strategy set
Anti-fraud strategy set
Device fingerprint strategy set
Knowledge graph strategy set
Four .Engines include:
Pre-credit engine
Anti-fraud engine
Quota increase engine
Is it possible to market engine
Whether multiple loans are allowed
Whether finance is risk control, and the core of risk control is data, whether it is traditional risk control or large Intelligent data risk control, risk control data are the basis of risk control. In the development of big data intelligent risk control, three-party risk control data plays an important role. At present, the application of big data intelligent risk control technology under the Internet finance wave is quite mature, and the three-party risk control data manufacturers and risk control data types are rich. Risk control data covers basic information, verification information, historical credit, consumption behavior, fulfillment ability, social information, public information, asset and liability information, black/grey list, relationship information, equipment information, etc. (Here, mainly lists human risk control data) type.
Rich three-party risk control data requires full-process system management from access, testing, use and operation. Only by doing the full-process management of data from the beginning of data access can the access of risk control data be changed More orderly, the risk control effect of risk control data is as great as possible.
. Among the policy rules introduced above, strong rules, weak rules, many rules, and rules are related layer by layer. There is also a Hunter-like strategy that everyone has rarely heard of, and its function is equivalent to soft rules and strategies, that is, they are different from HC (Hard Check) and access, the specific strategy contents are:
Edit and switch to center
Add image annotation, no more than 140 words (optional)
How to implement deployment and design in the specific decision engine, Hunter How much weight should a class strategy occupy, how to integrate Hunter-type strategies into specific risk control processes, how to design risk control decision flows to better block high-risk customers, and how to use decisions to further improve efficiency and reduce efficiency System pressure.
Regarding these contents, Tomato Risk Control has prepared a video course of nearly 40 minutes for all students during the holiday to help everyone understand the relevant content. The details are as follows:
edit switch to center
Add image annotation, no more than 140 Word (optional)
(Detailed video can be seen in the next article) In addition, regarding the rules for actual deployment, this Tomato Risk Control has also prepared a relevant decision-making engine platform for everyone to help everyone launch the real rule model. The specific deployment rules include : : : : : : : : : : : : : : : (Optional)
③Other anti-fraud and other related rules , please refer to the decision engine practical platform
practical platform related pages as follows:
Add image annotation, no more than 140 words (optional)
About For this practical platform, if you are interested, you can follow the video practical tutorial:
edit switch to center
Add image annotation, no more than 140 words (optional)
...
~Original article
~Original article
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-