Intelligent Verify Code

Based on the biological behavior characteristics and the environment self-development in-depth learning algorithm, the security issues in such scenarios as login and registration, activity seckill, likes giving as well as message posting are solved, credible, warning and malicious behaviors are accurately identified. In this way, the business security is protected to the largest extent, while the users’ ultimate experience is guaranteed.

Product Introduction

  • Nonperceptive Verification

    Users’ Nonperceptive Verification Experience

    Nonperceptive Verification

    The intelligent nonperceptive experience leads to efficient defense based on multi-dimensional information including environment, and credible users can pass by nonperceptive verification

  • One-click Verification

    Security verification can be achieved by one click

    One-click Verification

    The intelligent one-click verification leads to efficient defense based on multi-dimensional information including environment, and credible users can be verified by one click

  • Sliding Block Verification

    Puzzles are slid conveniently, achieving higher security and efficiency

    Sliding Block Verification

    Verification can be completed by easy sliding with intelligent sliding block verification. Such second user verification experience is efficient and safe

  • Semantics Verification

    Intelligent Semantics Final Verification

    Semantics Verification

    The final verification solution is provided for malicious attacks. With such extreme protection, security is several times higher

Typical Scenario

Registration Scenario

Typical Scenario: In the scenario of new user introduction and registration of enterprises, unreal users with operations completed by machines may easily flood in and bring risks to relevant businesses. Combining with device environment, regional location, behavior trace and other dimensions, JD Intelligent Verify Code gets prepared fast and identifies machine behaviors through characteristic rules and deep learning algorithms, and practically and efficiently stop abnormal batch registration behaviors from sources.

Login Scenario

Typical Scenario: Login risks are mainly the users’ interest losses caused by card and account thefts, brute force attacks, credential stuffing attacks and malicious login situation. On the premise of effective identification of abnormal login and reduction of high-quality user experience disturbance, JD Intelligent Verify Code accurately excavates abnormal hidden risks based on personal identification, behavior trace identification and other dimensions with machine learning, deep reinforcement learning and other modeling, and improves the security strength of account login.

Marketing Scenario

Typical Scenario: In scenarios of "panic buying", "seckill", "coupon obtaining" and other activities, participating users increase faster than usual, and a large number of econnoisseurs are also attracted to gather, creating an illusion that the platform gathers a large popularity. However, "econnoisseurs" are basically not loyalty to the platform, and may leave when no interests can be pursued, and the marketing expenses become a waste, to the disadvantage of long-term development of the platform. JD Intelligent Verify Code effectively withstand malicious scalping and getdeal based on blacklist, complicated community network, unsupervised graph model and other technologies, in order to protect business security and real customer benefits.

Interaction Scenario

Typical Scenario: In the interactive scenarios including "voting", "posting" and "liking", there are a large number of unreal users, seriously affecting the fairness of activities. JD Intelligent Verify Code may effectively withstand automatic like, malicious spams, posting, voting and other problems based on community relationship network, machine identification, abnormal agglomeration identification and other advanced artificial intelligence technologies, and maintain the fairness of rules.


  • Security Defense

    Based on the man-machine identification technology of the in-depth learning algorithm of self-development, the recent results of JD’s mass data mining technologies are shared, to intelligently deal with various risk challenges

  • Ultimate Experience

    It reduces the users’ difficulties in operation in addition to achieving safe verification. So, the issue of poor user experience of traditional verification codes is solved

  • Intelligent Grading

    The verification methods are judged and switched over intelligently by risk levels, the dynamic semantics verification is created, the solution for ultimate verification is provided, and the business security is guaranteed

  • Rapid Response

    Millisecond service response and elastic dynamic expansion are achieved based on the JD Cloud server

  • Data Encryption

    The patented private encryption algorithm is adopted for such data as the user’s behavior traces, achieving safer data transmission

  • Rich Interaction Forms

    Multiple product interaction configurations such as embedded, popup, trigger, associate are supported, meeting the demands of various scenarios

  • Customized Configuration Supported

    Image configurations and picture display time are spliced by customizable sliding blocks, more effectively defending business risks

Solution Architecture

System Composition

1. Verification code client (JS and SDK): The JS file and SDK connected from the frontend are mainly provided; JS mainly provides the verification code products integrated by the front-end code 2. Verification code server: The client and algorithm model are connected, the user data submitted by the client are received, data are subject to preprocessing for judgment, the algorithm model is called, it is judged whether the user has passed the verification by rules; the background interface business party is provided to verify the user’s operation result 3. Verification code algorithm model: User data are reported through the front end, and man-machine is identified by the machine learning algorithm 4. The verification code management backstage consists of: Verify the management of the intelligent decision-making rules engine, management of access business, statistics & analysis of various business requests and throughputs