CN-121997161-A - Anti-fraud method, device, equipment and storage medium based on hotel reservation

CN121997161ACN 121997161 ACN121997161 ACN 121997161ACN-121997161-A

Abstract

The application relates to a hotel reservation-based anti-fraud method apparatus, device, and storage medium. The method comprises the steps of constructing a booking-canceling time sequence sub-graph of a user to which a target order belongs in a preset time period, extracting a static behavior baseline vector and a dynamic time sequence adjacent matrix based on the booking-canceling time sequence sub-graph, coupling the dynamic time sequence adjacent matrix with a market price difference index to generate an environment-aware embedded vector, inputting the static behavior baseline vector and the environment-aware embedded vector into a fraud intention recognition model to obtain fraud probability score of the target order, and executing a corresponding anti-fraud strategy based on the fraud probability score. The application can improve the accuracy of the identification of the fraudulent activity.

Inventors

JIN ZHEN
WU WEILUE
Ye Wenzhao
CAO YUE

Assignees

深圳市道旅旅游科技有限公司

Dates

Publication Date: 20260508
Application Date: 20251203

Claims (10)

1. A method of anti-fraud based on hotel reservations, the method comprising: constructing a reservation-cancellation time sequence subgraph of a user to which a target order belongs in a preset time period, and extracting a static behavior baseline vector and a dynamic time sequence adjacency matrix based on the reservation-cancellation time sequence subgraph; coupling the dynamic time sequence adjacent matrix with a market price index to generate an environment perception embedded vector; Inputting the static behavior baseline vector and the environment perception embedded vector into a fraud intention recognition model to obtain fraud probability scores of the target orders; Based on the fraud probability score, a corresponding anti-fraud policy is executed.
2. The hotel reservation-based anti-fraud method of claim 1, wherein the constructing a reservation-cancellation timing subgraph of a user to whom the target order belongs within a preset time period, extracting a static behavior baseline vector and a dynamic timing adjacency matrix based on the reservation-cancellation timing subgraph, comprises: Extracting an effective event set corresponding to a target order from an order database; Constructing the time sequence subgraph based on the effective event set, wherein nodes of the time sequence subgraph represent single events, and edges represent time sequence relations among the events and house source incidence relations; Calculating the time attenuation weight of each edge to generate a dynamic time sequence adjacent matrix; And inputting the dynamic time sequence adjacency matrix and the node characteristic vector into a graph rolling network, and extracting a static behavior baseline vector representing the behavior stability of the user.
3. The hotel reservation-based anti-fraud method of claim 2, wherein said extracting a set of valid events corresponding to the target order from the order database comprises: extracting all reservation and cancellation events of a user to which a target order belongs in a preset time period from an order database, and constructing an original event sequence comprising a time stamp, a house source identifier and an operation type; And preprocessing the original event sequence to obtain an effective event set containing the target hotel and the related house sources.
4. The hotel reservation-based anti-fraud method of claim 2, wherein said coupling said dynamic time series adjacency matrix with a market price index generates an environment-aware embedded vector comprising: determining a target hotel corresponding to the target order, and acquiring bid price data of a business district where the target hotel is located; Calculating the deviation degree of the real-time price of the target hotel and the price data of the bid, and generating a standardized market price difference index; Normalizing the dynamic time sequence adjacent matrix according to rows to obtain an edge weight matrix representing the transition probability of each node; taking the market price difference index as a global adjusting factor, multiplying the market price difference index with the edge-out weight matrix element by element to generate a price difference perception adjacency matrix; And inputting the price difference perception adjacent matrix into a bilateral bilinear pooling layer, fusing time sequence topological characteristics and market price fluctuation characteristics, and outputting an environment perception embedded vector.
5. The hotel reservation-based anti-fraud method of claim 4, wherein said inputting the price-difference-aware adjacency matrix into a bilateral bilinear pooling layer, fusing time-series topological features with market price fluctuation features, outputting an environment-aware embedded vector, comprises: Singular value decomposition is carried out on the valence difference perception adjacent matrix, so that a topological feature matrix representing a time sequence connection mode and a weight feature matrix representing valence difference influence are obtained; Inputting the topological feature matrix and the weight feature matrix into two independent channels of a bilateral bilinear pooling layer, and executing element-by-element outer product operation to generate a high-dimensional coupling feature tensor; and carrying out global maximum pooling and average pooling double-path aggregation on the high-dimensional coupling characteristic tensor along the characteristic dimension to generate a compressed environment perception embedded vector.
6. The hotel reservation-based anti-fraud method of claim 1, wherein said entering the static behavior baseline vector and the context awareness embedded vector into a fraud intent recognition model to obtain a fraud probability score for the target order comprises: Non-linear activation is carried out on the static behavior baseline vector, and a behavior pattern enhancement vector is generated; performing adaptive channel attention weighting on the environment-aware embedded vector to generate an environment risk enhancement vector; Performing bilinear feature intersection on the behavior pattern enhancement vector and the environment risk enhancement vector to generate a fusion feature vector; And inputting the fusion feature vector into a fully-connected network to obtain the fraud probability score of the target order.
7. A hotel reservation-based anti-fraud method according to claim 1, characterized in that said executing a respective anti-fraud policy based on said fraud probability score comprises: if the fraud probability score is in a first interval, carrying out identity verification on the user to which the target order belongs; And if the fraud probability score is in the second interval, freezing the target order and pushing the target order to a pre-audit terminal for manual audit.
8. An anti-fraud device based on hotel reservations, the device comprising: The extraction module is used for constructing a reservation-cancellation time sequence subgraph of a user to which the target order belongs in a preset time period, and extracting a static behavior baseline vector and a dynamic time sequence adjacency matrix based on the reservation-cancellation time sequence subgraph; The generation module is used for coupling the dynamic time sequence adjacent matrix with the market price index to generate an environment perception embedded vector; The identification module is used for inputting the static behavior baseline vector and the environment perception embedded vector into a fraud intention identification model to obtain fraud probability scores of the target orders; and the execution module is used for executing corresponding anti-fraud strategies based on the fraud probability scores.
9. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus; A memory for storing a computer program; a processor for implementing the hotel reservation-based anti-fraud method of any of claims 1 to 7 when executing a program stored on a memory.
10. A computer readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements a hotel reservation-based anti-fraud method according to any of claims 1 to 7.

Description

Anti-fraud method, device, equipment and storage medium based on hotel reservation Technical Field The present application relates to the field of fraud identification technologies, and in particular, to a hotel reservation-based anti-fraud method, device, apparatus, and storage medium. Background In the online hotel distribution industry, there is a house-cutting fraudulent activity, a fraudulent party uses an illegal credit card to subscribe a hotel room which can be canceled for free, resells a real client at a price slightly lower than the market price, and cancels the original subscription cash register after the client is checked in, so that the hotel is paid after the client is checked in rent, the client complaints and the platform reputation are damaged. The prior art mainly relies on credit card verification, order time limit cancellation rules and manual spot check for precaution, but the means have certain defects that the single-dependent payment verification cannot describe time sequence behavior characteristics of a fraudulent party, risk detection is lagged, a problem is usually exposed after a customer enters and complains, loss cannot be recovered, and the conventional general anti-fraud model cannot effectively correlate and analyze user behavior abnormality and external market price difference causes, so that the accuracy rate of fraud behavior identification is not high under a high price difference scene. Therefore, how to improve the recognition accuracy of the fraudulent activity has become a technical problem to be solved by those skilled in the art. Disclosure of Invention In view of the above, the present application provides a hotel reservation-based anti-fraud method, device, apparatus and storage medium, which aims to solve the above technical problems. In a first aspect, the present application provides a method of anti-fraud based on hotel reservations, the method comprising: constructing a reservation-cancellation time sequence subgraph of a user to which a target order belongs in a preset time period, and extracting a static behavior baseline vector and a dynamic time sequence adjacency matrix based on the reservation-cancellation time sequence subgraph; coupling the dynamic time sequence adjacent matrix with a market price index to generate an environment perception embedded vector; Inputting the static behavior baseline vector and the environment perception embedded vector into a fraud intention recognition model to obtain fraud probability scores of the target orders; Based on the fraud probability score, a corresponding anti-fraud policy is executed. In a second aspect, the present application provides a hotel reservation-based anti-fraud device comprising: The extraction module is used for constructing a reservation-cancellation time sequence subgraph of a user to which the target order belongs in a preset time period, and extracting a static behavior baseline vector and a dynamic time sequence adjacency matrix based on the reservation-cancellation time sequence subgraph; The generation module is used for coupling the dynamic time sequence adjacent matrix with the market price index to generate an environment perception embedded vector; The identification module is used for inputting the static behavior baseline vector and the environment perception embedded vector into a fraud intention identification model to obtain fraud probability scores of the target orders; and the execution module is used for executing corresponding anti-fraud strategies based on the fraud probability scores. In a third aspect, the present application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus; A memory for storing a computer program; a processor configured to implement the steps of the hotel reservation-based anti-fraud method according to any of the embodiments of the first aspect when executing a program stored on a memory. In a fourth aspect, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the hotel reservation-based anti-fraud method according to any of the embodiments of the first aspect. Compared with the prior art, the technical scheme provided by the embodiment of the application has the following advantages: The sequential rhythm characteristics of 'intensive booking-centralized cancellation' of a user are characterized by constructing booking-canceling sequential subgraphs and extracting static behavior baseline vectors and dynamic sequential adjacency matrixes, the dynamic sequential adjacency matrixes and market price difference indexes are coupled to generate environment perception embedded vectors, dynamic fusion of external benefit inducement and internal behavior anomal