Big data-based customer consumption behavior analysis method and system

Abstract

The application provides a customer consumption behavior analysis method and system based on big data, belongs to the technical field of customer portrait in big data analysis, and comprises a data access and shunting unit, a long-term portrait processing unit, an instant intention monitoring unit, a situation impact identification unit, a dual-state portrait arbitration unit and a portrait execution and output unit; the system receives data flow and constructs a historical stable portrait and an instant intention portrait in parallel; the core is that the situation impact identification unit determines whether to activate a situation arbitration activation signal; the dual-state portrait arbitration unit determines whether to use the stable historical portrait or switch to the sensitive instant intention portrait as the current effective portrait according to the signal; the application solves the technical conflict between long-term stability and instant sensitivity through the dual-state portrait arbitration architecture, and takes into account stable operation and instant response.

Description

Technical Field

[0001] This invention relates to the field of customer profiling technology in big data analytics, specifically to a method and system for analyzing customer consumption behavior based on big data. Background Technology

[0002] With the development of big data analytics, customer profiling is becoming increasingly important in business decision-making. Currently, traditional single-profiling architectures struggle to balance stability and sensitivity. If the system prioritizes long-term stability, it cannot capture high-value immediate intents, leading to missed critical conversion windows. If the system prioritizes immediate sensitivity, long-term profiles are contaminated by transient noise data, causing profile drift or inaccuracy. Therefore, accurately distinguishing between high-priority contextual intents and transient noise, and resolving the technical conflict between long-term profile stability and immediate intent responsiveness, has become a pressing issue in this field. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for analyzing customer consumption behavior based on big data, so as to solve at least one of the technical problems existing in the background art.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] In a first aspect, the present invention provides a customer consumption behavior analysis system based on big data, comprising:

[0006] The data access and splitting unit is used to receive customer behavior data streams and split the data streams into historical data and real-time data streams.

[0007] The long-term profile processing unit is used to receive the historical data, construct a historical stable profile, and send the historical stable profile to the contextual impact recognition unit and the dual-state profile arbitration unit.

[0008] The real-time intent monitoring unit is used to receive the real-time stream data, construct a real-time intent profile, and send the real-time intent profile to the contextual impact recognition unit and the dual-state profile arbitration unit;

[0009] The contextual impact recognition unit is used to receive the real-time intent profile, the historical stable profile, and optional external event signals, determine whether the contextual activation conditions are met, and generate a contextual arbitration activation signal when the conditions are met, and send the contextual arbitration activation signal to the dual-state profile arbitration unit and the data access and splitting unit.

[0010] The dual-state profile arbitration unit is used to receive the historical stable profile, the real-time intent profile, and the contextual arbitration activation signal. When the contextual arbitration activation signal is not received, the historical stable profile is selected as the currently effective profile. When the contextual arbitration activation signal is received, the real-time intent profile is selected as the currently effective profile.

[0011] The portrait execution and output unit is used to receive and output the currently effective portrait.

[0012] As a further limitation of the first aspect of the present invention, the data access and splitting unit is specifically used for:

[0013] Upon receiving the context arbitration activation signal, the routing logic is dynamically adjusted to stop routing real-time stream data that has been identified as high-priority context intents to the long-term profile processing unit, and instead only to the real-time intent monitoring unit, so as to achieve isolation and protection of historical stable profiles.

[0014] As a further limitation of the first aspect of the present invention, the long-term image processing unit is specifically used for:

[0015] By employing a long-term window and batch processing model, the historical data is aggregated, cleaned, and modeled to calculate the historical stable profile that reflects the long-term preference status of customers.

[0016] As a further limitation of the first aspect of the present invention, the real-time intent monitoring unit is specifically used for:

[0017] Using a short response latency and streaming model, the real-time streaming data is rapidly analyzed to identify short-term behavioral clusters that deviate from the historical stable profile, in order to construct the real-time intent profile.

[0018] As a further limitation of the first aspect of the present invention, the contextual impact recognition unit is specifically used for:

[0019] Based on the context activation determination logic, when both the intent intensity determination and cross-domain deviation determination are satisfied, the instant intent profile is determined to be a high-priority context intent, and the context arbitration activation signal is generated.

[0020] The intent intensity determination is defined as follows: the frequency, depth, and continuity of the real-time behavior reflected by the real-time stream data exceed a preset transient noise baseline.

[0021] The cross-domain deviation is determined when the correlation between the domain to which the instant intent profile belongs and the domain to which the historical stable profile belongs is lower than a preset threshold.

[0022] As a further limitation of the first aspect of the present invention, the contextual impact recognition unit is also used for:

[0023] Access external event signals;

[0024] The context activation determination logic also includes external context association determination, which is: the instant intent profile is matched with the cross-domain context impact event in the external event signal.

[0025] As a further limitation of the first aspect of the present invention, when the dual-state image arbitration unit receives the context arbitration activation signal, it is specifically used for:

[0026] Temporarily suspend the effective status of the historical stable profile to prevent it from being used for subsequent business operations;

[0027] The instant intent profile is executed first, making it the currently active profile.

[0028] As a further limitation of the first aspect of the present invention, the dual-state image arbitration unit is also used for:

[0029] When the situation impact recognition unit detects that the situation activation condition is no longer met and stops sending the situation arbitration activation signal, it automatically releases the suspended state of the historical stable portrait and switches back to the default state, selecting the historical stable portrait as the currently effective portrait.

[0030] Secondly, the present invention provides a customer consumption behavior analysis method based on big data, including:

[0031] The data access and distribution unit receives customer behavior data streams and distributes the data streams into historical data and real-time data streams.

[0032] The long-term profile processing unit receives the historical data, constructs a historical stable profile, and sends the historical stable profile to the contextual impact recognition unit and the dual-state profile arbitration unit;

[0033] The real-time intent monitoring unit receives the real-time stream data, constructs a real-time intent profile, and sends the real-time intent profile to the contextual impact recognition unit and the dual-state profile arbitration unit;

[0034] The context impact recognition unit receives the real-time intent profile, the historical stable profile, and optional external event signals, determines whether the context activation conditions are met, and generates a context arbitration activation signal when the conditions are met, and sends the context arbitration activation signal to the dual-state profile arbitration unit and the data access and splitting unit.

[0035] The dual-state profile arbitration unit selects between the historical stable profile and the immediate intent profile based on whether the context arbitration activation signal is received, in order to determine the currently effective profile;

[0036] The portrait execution and output unit receives and outputs the currently effective portrait.

[0037] Thirdly, the present invention provides a non-transitory computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the customer consumption behavior analysis method based on big data as described in the second aspect.

[0038] Fourthly, the present invention provides a computer device including a memory and a processor, wherein the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the customer consumption behavior analysis method based on big data as described in the second aspect.

[0039] Fifthly, the present invention provides an electronic device, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the customer consumption behavior analysis method based on big data as described in the second aspect.

[0040] The beneficial effects of this invention are as follows: By constructing a dual-state profile arbitration architecture, it overcomes the technical limitations of traditional single profiles and resolves the technical conflict between long-term profile stability and real-time intent sensitivity, enabling the system to simultaneously ensure stable operation and real-time response. By dynamically adjusting the routing logic of data access and distribution units during context activation, it achieves isolation and protection of historical stable profiles. It ensures that real-time intent data with high deviations does not pollute the historical dataset, preventing long-term profiles from drifting or becoming inaccurate from the data source and maintaining their long-term accuracy. It proposes a multi-dimensional context activation judgment logic, which can accurately distinguish between high-value cross-domain context intents and massive transient browsing noise by combining judgments of intent strength, cross-domain deviation, and even external events, greatly improving the accuracy of arbitration and avoiding false activation of the system due to noise interference. It achieves a smooth and robust dual-state switching mechanism. When a high-value context is identified, the system can immediately suspend the long-term profile and prioritize the execution of the real-time intent profile, ensuring correct intervention during the critical conversion window. After the context ends, it can automatically and seamlessly switch back to the stable profile, avoiding resource misallocation and missed business opportunities.

[0041] The advantages of additional aspects of the invention will be set forth more clearly in the following description or will be learned by practice of the invention. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a functional principle block diagram of the customer consumption behavior analysis system based on big data as described in an embodiment of the present invention.

[0044] Figure 2 This is a flowchart of the customer consumption behavior analysis method based on big data as described in an embodiment of the present invention. Detailed Implementation

[0045] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0046] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0047] It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as here.

[0048] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0049] To facilitate understanding of the present invention, the present invention will be further explained and described below with reference to the accompanying drawings and specific embodiments. However, the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0050] Those skilled in the art should understand that the accompanying drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily essential for implementing the present invention.

[0051] Example 1

[0052] In this embodiment 1, as Figure 1 As shown, a customer consumption behavior analysis system based on big data is provided, including:

[0053] The data access and distribution unit is used to receive customer behavior data streams and distribute the data streams into historical data and real-time data streams.

[0054] The long-term profile processing unit is used to receive historical data, construct historical stable profiles, and send the historical stable profiles to the contextual impact recognition unit and the dual-state profile arbitration unit.

[0055] The real-time intent monitoring unit is used to receive real-time streaming data, construct a real-time intent profile, and send the real-time intent profile to the contextual impact recognition unit and the dual-state profile arbitration unit.

[0056] The contextual impact recognition unit is used to receive real-time intent profiles, historical stable profiles, and optional external event signals, determine whether the contextual activation conditions are met, and generate a contextual arbitration activation signal when the conditions are met, and send the contextual arbitration activation signal to the dual-state profile arbitration unit and the data access and splitting unit.

[0057] The dual-state profile arbitration unit is used to receive historical stable profiles, real-time intent profiles, and contextual arbitration activation signals. When no contextual arbitration activation signal is received, the historical stable profile is selected as the currently effective profile. When a contextual arbitration activation signal is received, the real-time intent profile is selected as the currently effective profile.

[0058] The portrait execution and output unit is used to receive and output the currently active portrait.

[0059] In one specific embodiment, the above system architecture is described in detail;

[0060] The data access and distribution unit is designed to perform initial data processing and distribution, providing independent data sources with different characteristics for subsequent dual-state processing. This unit serves as the system's unified entry point, receiving all customer behavioral data streams. Its internal processing logic involves, by default, distributing data streams based on data timeliness and processing requirements: on one hand, routing data requiring long-term aggregation and analysis as historical data; on the other hand, routing immediate, short-window data as real-time stream data. The historical data generated by this unit's distribution is sent to the long-term profiling unit, while the real-time stream data is sent to the real-time intent monitoring unit.

[0061] The long-term profile processing unit is tasked with building a baseline profile that reflects the customer's long-term and stable preferences. This unit receives historical data from the data access and distribution unit. Its core processing logic uses a long time window and batch processing model to calculate the customer's most likely stable state. This unit calculates and outputs a structured historical stable profile, which is then sent to the contextual impact recognition unit and the dual-state profile arbitration unit.

[0062] The real-time intent monitoring unit is tasked with capturing customers’ short-term, real-time needs that may deviate from their long-term preferences. This unit receives real-time streaming data from the data access and distribution unit. Its core processing logic adopts a short response latency and streaming processing model, which aims to quickly identify short-term behavior clusters. This unit analyzes and outputs a temporary real-time intent profile, which is simultaneously sent to the contextual impact recognition unit and the dual-state profile arbitration unit.

[0063] The contextual impact recognition unit is designed to solve the core technical challenge of distinguishing between transient noise and high-priority contextual intent. This unit receives an immediate intent profile, a historical stable profile, and optional external event signals. Its core processing logic is one of the key aspects of this system. It compares and analyzes the two and makes a judgment based on a preset contextual activation determination logic. This determination logic mainly comprehensively evaluates whether the intensity of the immediate intent exceeds the transient noise baseline and whether the cross-domain deviation between the immediate intent profile and the historical stable profile is lower than a preset threshold. The specific determination conditions and calculation process are described in detail in subsequent embodiments. When the determination conditions are met, the unit outputs a key contextual arbitration activation signal. This signal is a high-priority control command that is simultaneously sent to the dual-state profile arbitration unit and the data access and splitting unit.

[0064] The dual-state profile arbitration unit functions as the system's context arbitration mechanism or control center, executing the dynamic switching of dual-state profiles. This unit receives historical stable profiles, current intent profiles, and key context arbitration activation signals; its processing logic switches states based on these activation signals.

[0065] When no situational arbitration activation signal is received, the internal logic of the arbitration unit selects the historical stable profile as the currently effective profile;

[0066] When a situational arbitration activation signal is received, the internal logic of the arbitration unit immediately switches and selects the immediate intent profile as the currently effective profile.

[0067] The profile execution and output unit is positioned to apply the arbitrated profile that best reflects the current state of the customer to downstream business; this unit receives the currently effective profile from the dual-state profile arbitration unit; this unit then outputs the profile to the downstream recommendation system or marketing system;

[0068] Through the collaborative work of the above six units, this invention constructs a novel dual-state arbitration architecture. This invention overcomes the technical dogma of portrait uniqueness in the traditional single portrait architecture and resolves the technical conflict between stability and sensitivity. This enables the system to maintain the stability of long-term portraits and, under specific situational shocks, prioritize responses to high-value immediate intentions, thereby avoiding resource misallocation and missing critical conversion windows.

[0069] In one specific embodiment, the data access and splitting unit is specifically used for:

[0070] Upon receiving a context arbitration activation signal, the routing logic is dynamically adjusted to stop routing real-time stream data that has been identified as high-priority contextual intents to the long-term profile processing unit, and instead only to the real-time intent monitoring unit, in order to achieve isolation and protection of historical stable profiles.

[0071] In addition to customer behavior data streams, the input to the data access and distribution unit also includes situation arbitration activation signals from the situation impact recognition unit;

[0072] By default, the data stream is distributed to the long-term profiling unit and the real-time intent monitoring unit as usual.

[0073] When the unit receives a context arbitration activation signal, the signal acts as a high-priority instruction, triggering the unit to dynamically adjust its routing logic.

[0074] This unit will stop routing currently occurring real-time stream data related to baking intent to the long-term profiling unit; at the same time, this data will only be routed to the real-time intent monitoring unit for continuous short-term intent tracking.

[0075] This dynamic routing and isolation action achieves the isolation and protection of historical stable profiles; it ensures that temporary, highly skewed intent data will not contaminate the historical dataset used to build long-term preferences; thus, it prevents long-term profiles from drifting or becoming inaccurate due to short-term situational shocks from the source of data, and greatly maintains the long-term accuracy of historical stable profiles.

[0076] In one specific embodiment, the long-term image processing unit is specifically used for:

[0077] By employing a long time window and batch processing model, historical data is aggregated, cleaned, and modeled to calculate a historically stable profile that reflects the long-term preferences of customers.

[0078] The long-term profile processing unit receives historical data; its processing logic is configured to use a long time window; the specific duration of the long time window is preferably set to 6 to 18 months, and this range is chosen to balance the stability of the profile with its adaptability to changes in macro trends; this unit adopts a batch processing model, which can be implemented using technologies such as MapReduce, Hive, or SparkSQL, and its execution cycle is preferably... or This unit aggregates, cleans, and models massive amounts of historical data. Its output, a historical stable profile, is a data structure reflecting the long-term, macro-level, and gradually changing preferences of customers. Specifically, this long-time-window and batch processing model includes a data cleaning module, a feature extraction module, and a preference weight calculation module. Its working logic is as follows: First, the data cleaning module removes abnormal interaction data and invalid browsing records. Then, the feature extraction module extracts the customer's category preferences and consumption frequency features. Finally, the preference weight calculation module uses a time-decay weighted algorithm to calculate the customer's stable state. The formula for calculating the long-term preference weight of a customer for a certain category is: ,in For customers in time Effective interaction behavior score, For the current time, The preset time decay coefficient is used to generate a key-value pair data structure containing the weights of various preferences, which constitutes a structured historical stability profile.

[0079] By employing a long time window and batch processing model, the macroscopic accuracy and stability of historical stable profiles are ensured, making them less susceptible to interference from short-term, random behavior, thus providing a reliable anchor point for the system.

[0080] In one specific embodiment, the real-time intent monitoring unit is specifically used for:

[0081] Employing a short response latency and streaming model, real-time streaming data is rapidly analyzed to identify short-term behavioral clusters that deviate from historical stable profiles, thereby constructing real-time intent profiles.

[0082] The real-time intent monitoring unit receives real-time streaming data; its processing logic is configured to use a short response latency, strictly controlled within the range of 500 milliseconds to 5 seconds, to ensure intent capture is completed within the user's golden window. This unit employs a streaming processing model, which can be implemented using technologies such as Flink, Spark Streaming, or Kafka Streams. This unit performs rapid analysis of real-time data, with its core objective being to identify short-term behavioral clusters that significantly deviate from historical stable profiles. Its output real-time intent profile is a temporary profile reflecting current needs. Specifically, the short response latency and streaming processing model include a sliding window module, a real-time feature encoding module, and a density clustering module. Its working logic is as follows: the sliding window module captures real-time streaming data at a preset time step; the real-time feature encoding module converts the captured data into a multi-dimensional intent vector; and then the density clustering module uses a streaming clustering algorithm to classify the intent vector in real-time. For the identification of short-term behavior clusters, the system calculates the core density of the intent vector cluster within the current sliding window. When the core density exceeds the set clustering threshold Dmin, it is identified as a short-term behavior cluster, and the central features of the cluster are extracted as an instant intent profile.

[0083] By employing a short response latency and a streaming model, the system is able to sensitively capture customers' immediate intentions, especially context-driven new demands that are inconsistent with long-term preferences, providing crucial input for subsequent context identification and arbitration.

[0084] In one specific embodiment, the contextual impact recognition unit is specifically used for:

[0085] Based on the context activation determination logic, when both the intent intensity determination and cross-domain deviation determination are satisfied, the immediate intent profile is determined to be a high-priority context intent, and a context arbitration activation signal is generated.

[0086] Among them, the determination of intent intensity is: the frequency, depth and continuity of real-time behavior reflected by real-time stream data exceed the preset transient noise baseline;

[0087] Cross-domain deviation is determined when the correlation between the domain to which the real-time intent profile belongs and the domain to which the historical stable profile belongs is lower than a preset threshold.

[0088] The contextual impact recognition unit is also used for:

[0089] Access external event signals; external event signals refer to objective event data that are independent of individual customer history and can trigger large-scale changes in the consumption intentions of customer groups. Specifically, these include: sudden extreme weather change signals, major social news event signals, public health event signals, or major promotional holiday signals across the entire network.

[0090] The context activation determination logic also includes external context association determination, which is: matching the real-time intent profile with cross-domain context impact events in external event signals.

[0091] The contextual impact recognition unit accurately distinguishes between transient noise and high-priority contextual intent;

[0092] Real-time intent profiling, historically stable profiling, and optional external event signals;

[0093] Context activation determination logic; upon receiving the immediate intent profile, this unit does not immediately trigger arbitration, but instead executes a series of strict determinations:

[0094] Intent strength determination: This unit calculates a transient intent strength score in real time; the score is determined by a preset weighting algorithm after taking into account the frequency, depth and continuity of the immediate behavior.

[0095] The transient intent intensity score is compared with a preset transient noise baseline;

[0096] Transient noise baseline is a quantitative indicator used to characterize the intensity of random browsing behavior without genuine intent. The system statistically analyzes the intensity scores of random, aimless browsing behavior from a massive amount of historical users and takes the baseline. quantiles As a baseline;

[0097] This criterion is satisfied only when the transient intent strength score is higher than the baseline.

[0098] Cross-domain deviation determination: This unit calculates the correlation between the real-time intent profile and the historical stable profile;

[0099] The method for calculating the relevance can be based on the well-known profile tag system in this field, for example: by calculating the conditional probability of co-occurrence of two domain tags in historical user behavior or the semantic vector similarity;

[0100] This correlation will be compared with a preset threshold; if a normalized correlation of 0 to 1 is used, the threshold can preferably be set in the range of 0.05 to 0.2.

[0101] This criterion is satisfied only when the correlation is below this threshold;

[0102] External context association determination: When the system receives an external event signal, this unit will also determine whether the instant intent profile matches the currently known cross-domain context impact event; this matching process can be achieved through keyword Boolean matching or short text semantic similarity comparison;

[0103] Only when both the intent strength determination and the cross-domain deviation determination are satisfied will the unit ultimately determine the instant intent profile as a high-priority contextual intent, and immediately generate a contextual arbitration activation signal, which will be sent to the dual-state profile arbitration unit and the data access and diversion unit.

[0104] Through the aforementioned multi-dimensional context activation judgment logic, this embodiment provides a non-obvious, robust differentiation mechanism. It no longer relies on a single threshold, but rather uses a combination of intensity, deviation, and external context to accurately identify truly high-value cross-domain contextual impacts from massive amounts of real-time data. At the same time, it filters out a large amount of transient noise, greatly improving the accuracy and effectiveness of subsequent arbitration and avoiding false activation of the system.

[0105] When the dual-state image arbitration unit receives the context arbitration activation signal, it is specifically used for:

[0106] Temporarily suspend the effective status of the historical stability profile to prevent it from being used for subsequent business operations;

[0107] And prioritize the execution of the immediate intent profile, making the immediate intent profile the currently effective profile.

[0108] The dual-state image arbitration unit is also used for:

[0109] When the context impact recognition unit detects that the context activation conditions are no longer met and stops sending context arbitration activation signals, it automatically releases the suspended state of the historical stable profile and switches back to the default state, selecting the historical stable profile as the currently effective profile.

[0110] The dual-state image arbitration unit enables the orderly switching between two image states;

[0111] Historical stable profile, real-time intent profile, and context arbitration activation signal from the context impact recognition unit;

[0112] When this unit receives the context arbitration activation signal, its internal logic immediately executes two key actions:

[0113] The system temporarily suspends the effective status of historical stability profiles to prevent these profiles from being used for subsequent business operations.

[0114] The system then prioritizes the execution of the real-time intent profile; this real-time intent profile becomes the only high-priority currently active profile that the system outputs to the profile execution and output unit.

[0115] This unit continuously monitors the status of the context arbitration activation signal; when a high-priority context intent disappears, the unit's logic automatically triggers recovery.

[0116] The system automatically releases the suspended state of the historical stable profile and switches back to the default state, reselecting the historical stable profile as the currently active profile;

[0117] Through this arbitration mechanism of suspending A and prioritizing B, and the closed-loop logic of signal cessation and automatic restoration of A, this embodiment achieves a smooth and robust dual-state switching. It ensures that the system can make the correct intervention based on the real-time intent profile during the golden window period when the user's intent is strongest. At the same time, after the situation ends, it can seamlessly switch back to the historical stable profile and continue long-term operation, realizing dynamic coordination and orthogonal decoupling between stability and sensitivity of the system, rather than sacrificing one for the other.

[0118] Please see Figure 2 In this implementation, a customer consumption behavior analysis method based on big data was implemented using the above-mentioned system, including:

[0119] The data access and distribution unit receives customer behavior data streams and distributes the data streams into historical data and real-time data streams.

[0120] The long-term profile processing unit receives historical data, constructs a historical stable profile, and sends the historical stable profile to the contextual impact recognition unit and the dual-state profile arbitration unit;

[0121] The real-time intent monitoring unit receives real-time stream data, constructs a real-time intent profile, and sends the real-time intent profile to the contextual impact recognition unit and the dual-state profile arbitration unit;

[0122] The context impact recognition unit receives real-time intent profiles, historical stable profiles, and optional external event signals, determines whether the context activation conditions are met, and generates a context arbitration activation signal when the conditions are met, and sends the context arbitration activation signal to the dual-state profile arbitration unit and the data access and splitting unit.

[0123] The dual-state profiling arbitration unit selects between the historical stable profile and the immediate intent profile based on whether a contextual arbitration activation signal is received, in order to determine the currently effective profile.

[0124] The portrait execution and output unit receives and outputs the currently active portrait.

[0125] This step is performed by the data access and distribution unit. It receives all behavioral data streams from the customer and distributes them into historical data and real-time data streams.

[0126] This step involves two parallel processing paths:

[0127] This step is performed by the long-term profile processing unit. It receives historical data, constructs a historical stable profile through a batch processing model, and sends it to the contextual impact recognition unit and the dual-state profile arbitration unit.

[0128] This step is executed by the real-time intent monitoring unit. It receives real-time streaming data, constructs a real-time intent profile through a streaming processing model, and sends it to the contextual impact recognition unit and the dual-state profile arbitration unit.

[0129] This step is performed by the contextual impact recognition unit. It receives the real-time intent profile and the historical stable profile, and makes a judgment based on the contextual activation judgment logic. If the intent is determined to be a high-priority contextual intent, a contextual arbitration activation signal is generated and sent to the subsequent unit.

[0130] This is executed by the dual-state image arbitration unit, which determines which image to output based on whether a context arbitration activation signal is received.

[0131] If no activation signal is received, a historical stable profile will be selected as the currently active profile.

[0132] If an activation signal is received, a switch is executed, and the immediate intent profile is selected as the currently effective profile;

[0133] This step is executed by the profiling and output unit. It receives and outputs the currently effective profiling determined by the arbitration unit for use by downstream business systems.

[0134] This method provides a complete, closed-loop data processing flow. Through parallel processing, contextual judgment, and dynamic arbitration, the data processing model for customer profiles is transformed from a traditional single-state machine into a dual-state or multi-state arbitration state machine. This enables the entire analysis system to intelligently switch between long-term stability and immediate response modes based on the context, thereby achieving accurate capture of customer intent and effective protection of long-term profiles.

[0135] Example 2

[0136] This embodiment 2 provides a non-transitory computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, they implement the customer consumption behavior analysis method based on big data as described above. The method includes:

[0137] Acquire customer behavior data streams, including historical data and real-time streaming data;

[0138] Based on the historical data, construct a historical stable profile; based on the real-time stream data, construct a real-time intent profile.

[0139] Based on the real-time intent profile, the historical stable profile, and optional external event signals, determine whether the context activation condition is met, and generate a context arbitration activation signal when the condition is met.

[0140] Based on the contextual arbitration activation signal, a selection is made between the historical stable profile and the immediate intent profile to determine the currently effective profile.

[0141] Example 3

[0142] This embodiment 3 provides a computer device, including a memory and a processor, wherein the processor and the memory communicate with each other, and the memory stores program instructions that can be executed by the processor. The processor calls the program instructions to execute the customer consumption behavior analysis method based on big data as described above, the method including:

[0143] Acquire customer behavior data streams, including historical data and real-time streaming data;

[0144] Based on the historical data, construct a historical stable profile; based on the real-time stream data, construct a real-time intent profile.

[0145] Based on the real-time intent profile, the historical stable profile, and optional external event signals, determine whether the context activation condition is met, and generate a context arbitration activation signal when the condition is met.

[0146] Based on the contextual arbitration activation signal, a selection is made between the historical stable profile and the immediate intent profile to determine the currently effective profile.

[0147] Example 4

[0148] This embodiment 4 provides an electronic device, including: a processor, a memory, and a computer program; wherein, the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device performs the customer consumption behavior analysis method based on big data as described above, including:

[0149] Acquire customer behavior data streams, including historical data and real-time streaming data;

[0150] Based on the historical data, construct a historical stable profile; based on the real-time stream data, construct a real-time intent profile.

[0151] Based on the real-time intent profile, the historical stable profile, and optional external event signals, determine whether the context activation condition is met, and generate a context arbitration activation signal when the condition is met.

[0152] Based on the contextual arbitration activation signal, a selection is made between the historical stable profile and the immediate intent profile to determine the currently effective profile.

[0153] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0154] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0155] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0156] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment, whereby a series of operational steps are performed to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0157] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that, based on the technical solutions disclosed in the present invention, various modifications or variations that can be made by those skilled in the art without creative effort should be included within the scope of protection of the present invention.

Claims

1. A method for analyzing customer consumption behavior based on big data, characterized in that, include: Receive customer behavior data streams and split the data streams into historical data and real-time stream data; Receive the historical data and construct a historical stable profile; Receive the real-time streaming data and construct a real-time intent profile; Based on the real-time intent profile, the historical stable profile, and optional external event signals, determine whether the context activation condition is met, and generate a context arbitration activation signal when the condition is met. Based on the historical stable profile, the immediate intent profile, and the contextual arbitration activation signal, when the contextual arbitration activation signal is not received, the historical stable profile is selected as the currently effective profile; when the contextual arbitration activation signal is received, the immediate intent profile is selected as the currently effective profile.

2. The customer consumption behavior analysis method based on big data according to claim 1, characterized in that, Using a long time window and batch processing model, the historical data is aggregated, cleaned, and modeled to calculate the historical stable profile that reflects the long-term preference status of customers; Using a short response latency and streaming model, the real-time streaming data is rapidly analyzed to identify short-term behavioral clusters that deviate from the historical stable profile, in order to construct the real-time intent profile.

3. The customer consumption behavior analysis method based on big data according to claim 1, characterized in that, Based on the context activation determination logic, when both the intent intensity determination and cross-domain deviation determination are satisfied, the instant intent profile is determined to be a high-priority context intent, and the context arbitration activation signal is generated; wherein, the intent intensity determination is: the frequency, depth and continuity of the instant behavior reflected by the instant stream data exceed the preset transient noise baseline; the cross-domain deviation determination is: the correlation between the domain to which the instant intent profile belongs and the domain to which the historical stable profile belongs is lower than a preset threshold.

4. The customer consumption behavior analysis system based on big data according to claim 3, characterized in that, The context activation determination logic also includes external context association determination, which is: the instant intent profile is matched with the cross-domain context impact event in the external event signal.

5. The customer consumption behavior analysis system based on big data according to claim 1, characterized in that, Upon receiving the aforementioned situational arbitration activation signal, the effective status of the historical stable profile is temporarily suspended to prevent the historical stable profile from being used for subsequent business execution. The instant intent profile is executed first, making it the currently active profile.

6. The customer consumption behavior analysis system based on big data according to claim 5, characterized in that, When the scenario activation condition is no longer met and the scenario arbitration activation signal is stopped, the suspended state of the historical stable profile is automatically released, and the default state is switched back, and the historical stable profile is selected as the currently effective profile.

7. A customer consumption behavior analysis system based on big data, characterized in that, include: The data access and distribution unit is used to receive customer behavior data streams and distribute the data streams into historical data and real-time data. When the context arbitration activation signal is received, the routing logic is dynamically adjusted to stop routing real-time data that has been identified as high-priority context intents to the long-term profile processing unit, and instead only to the real-time intent monitoring unit, so as to achieve isolation and protection of historical stable profiles. The long-term profile processing unit is used to receive the historical data, construct a historical stable profile, and send the historical stable profile to the contextual impact recognition unit and the dual-state profile arbitration unit. The real-time intent monitoring unit is used to receive the real-time stream data, construct a real-time intent profile, and send the real-time intent profile to the contextual impact recognition unit and the dual-state profile arbitration unit; The contextual impact recognition unit is used to receive the real-time intent profile, the historical stable profile, and optional external event signals, determine whether the contextual activation conditions are met, and generate a contextual arbitration activation signal when the conditions are met, and send the contextual arbitration activation signal to the dual-state profile arbitration unit and the data access and splitting unit. The dual-state profile arbitration unit is used to receive the historical stable profile, the real-time intent profile, and the contextual arbitration activation signal. When the contextual arbitration activation signal is not received, the historical stable profile is selected as the currently effective profile. When the contextual arbitration activation signal is received, the real-time intent profile is selected as the currently effective profile. The portrait execution and output unit is used to receive and output the currently effective portrait.

8. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is used to store computer instructions, which, when executed by a processor, implement the customer consumption behavior analysis method based on big data as described in any one of claims 1-6.

9. A computer device, characterized in that, The device includes a memory and a processor, the processor and the memory communicating with each other, the memory storing program instructions that can be executed by the processor, and the processor calling the program instructions to execute the customer consumption behavior analysis method based on big data as described in any one of claims 1-6.

10. An electronic device, comprising: A processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory, characterized in that, when the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the customer consumption behavior analysis method based on big data as described in any one of claims 1-6.

Images