Big Data Push Processing Method and System Based on Internet Finance

Abstract

The present invention provides a big data push processing method and system based on Internet finance. By obtaining the financial product interaction data stream of a target user on an Internet finance platform and extracting the discrete interaction trajectory vectors of each interaction data unit, a fine-grained interaction model between the user and the financial product is effectively constructed. Further, by generating transfer interaction trajectory vectors to capture the global interaction pattern of the financial product interaction data stream and combining the discrete interaction trajectory vectors of the x-th interaction data unit to generate target interaction trajectory vectors, a dynamic and global understanding of the user's interaction behavior is achieved, and the x-th interaction data unit can be accurately predicted for interest points, generating corresponding interest point prediction labels, and based on this, accurate information push is performed on the target user. Thus, the pertinence and effectiveness of information push are improved, the user experience is enhanced, and at the same time, a more intelligent user service strategy is provided for the Internet finance platform.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular, to a big data push processing method and system based on Internet finance. Background Art

[0002] With the rapid development of Internet finance, the types and quantities of financial products have increased day by day, and the interaction behaviors of users on Internet finance platforms have become more complex and diverse. In order to improve user experience and enhance user stickiness, Internet finance platforms need to more accurately understand the interaction behaviors of users and perform personalized information push based on the interest points of users.

[0003] Traditional information push methods often rely on the static attributes or historical behavior data of users, and it is difficult to capture the real-time interaction behaviors and interest changes of users at the current moment. Especially in the field of Internet finance, the interaction behaviors of users often have the characteristics of temporality and phased updates, and traditional static analysis methods are difficult to accurately reflect the dynamic interaction patterns of users. Summary of the Invention

[0004] In view of the problems mentioned above, in combination with the first aspect of the present invention, embodiments of the present invention provide a big data push processing method based on Internet finance, and the method includes:

[0005] Obtain the financial product interaction data stream of the target user on the Internet finance platform, where the financial product interaction data stream includes x interaction data units that are phased updated in time series, and x is a positive integer;

[0006] Obtain the discrete interaction trajectory vectors corresponding to each of the x interaction data units in the financial product interaction data stream, where the discrete interaction trajectory vector is the independent interaction trajectory vector of each interaction data unit;

[0007] Generate a transfer interaction trajectory vector based on the discrete interaction trajectory vectors corresponding to the first y interaction data units in the financial product interaction data stream, where the transfer interaction trajectory vector is the global interaction trajectory vector of the financial product interaction data stream, and y is a positive integer less than x;

[0008] Generate the target interaction trajectory vector corresponding to the xth interaction data unit based on the transfer interaction trajectory vector and the discrete interaction trajectory vector corresponding to the xth interaction data unit;

[0009] Perform interest point prediction on the xth interaction data unit based on the target interaction trajectory vector corresponding to the xth interaction data unit, generate an interest point prediction label corresponding to the xth interaction data unit, and perform information push on the target user based on the interest point prediction label corresponding to the xth interaction data unit.

[0010] In another aspect, an embodiment of the present invention further provides a big data push processing system based on Internet finance, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, the embodiments of the present application effectively construct a fine-grained interaction model between users and financial products by obtaining the financial product interaction data stream of the target user on the Internet finance platform and extracting the discrete interaction trajectory vectors of each interaction data unit. Further, by generating and transmitting interaction trajectory vectors to capture the global interaction pattern of the financial product interaction data stream, and generating target interaction trajectory vectors in combination with the discrete interaction trajectory vectors of the x-th interaction data unit, a dynamic and global understanding of the user's interaction behavior is achieved, and the x-th interaction data unit can be accurately predicted for the point of interest, generating corresponding point of interest prediction labels, and accordingly, accurate information push is performed on the target user. Thereby, the pertinence and effectiveness of information push are improved, the user experience is enhanced, and at the same time, a more intelligent user service strategy is provided for the Internet finance platform, enhancing the user stickiness and competitiveness of the platform. Description of the Drawings

[0012] Figure 1 is a schematic execution flow diagram of the big data push processing method based on Internet finance provided by an embodiment of the present invention.

[0013] Figure 2 is a schematic hardware architecture diagram of the big data push processing system based on Internet finance provided by an embodiment of the present invention. Detailed Embodiments

[0014] The present invention will be specifically described below in conjunction with the accompanying drawings of the specification. Figure 1 is a schematic flow diagram of the big data push processing method based on Internet finance provided by an embodiment of the present invention. The big data push processing method based on Internet finance will be introduced in detail below.

[0015] Step S110, obtain the financial product interaction data stream of the target user on the Internet finance platform. The financial product interaction data stream includes x interaction data units that are updated in stages in time sequence, and x is a positive integer.

[0016] In this embodiment, it is assumed that on an Internet finance platform named "Investment Platform A", there is a target user named Mr. G. Mr. G has participated in various financial product interaction activities on this platform. These financial products include fund investment, time deposit, wealth management product purchase, etc.

[0017] When Mr. G first logs in to the platform and views the detail pages of various fund products, this operation is recorded as an interaction data unit. This interaction data unit contains information such as the time of the operation (e.g., 10:00 on January 1, 2023), the type of the operation (viewing fund details), the name of the fund product involved (e.g., "Growth Fund A"), and the result of the operation (no purchase or investment operation).

[0018] Over time, Mr. G may perform more operations. For example, he deposited a sum of money into his time deposit account at 14:00 on January 5, 2023. This operation is also recorded as an interaction data unit, which contains information such as the operation time, the type of the operation (depositing into time deposit), and the deposit amount (e.g., 5,000 yuan).

[0019] Suppose that within a certain period of time, Mr. G has performed a total of 10 such different types of financial product interaction operations on the platform. Then x here is equal to 10. These interaction data units are arranged in chronological order, which constitutes the financial product interaction data stream of Mr. G on the "Investment Platform A" Internet financial platform. Each interaction data unit reflects an interaction state of Mr. G with the financial product at a specific time point, and the ordered set of these interaction data units completely describes the financial product interaction behavior trajectory of Mr. G on this platform.

[0020] Step S120, obtain the discrete interaction trajectory vectors corresponding to each of the x interaction data units in the financial product interaction data stream, where the discrete interaction trajectory vector is the independent interaction trajectory vector of each of the interaction data units.

[0021] Continuing with the above example of Mr. G's interaction on the "Investment Platform A" platform. For the first interaction data unit, that is, the operation of viewing fund details at 10:00 on January 1, 2023. First, perform interference feature cleaning on the entire financial product interaction data stream. In this example, there may be some interference features, such as the platform's advertisement push records, some system prompt records that are irrelevant to the financial product interaction itself, etc. After removing these interference feature data, the financial product interaction data stream after interference feature cleaning is obtained.

[0022] Then, based on this cleaned data stream, obtain the interaction state transition features corresponding to this interaction data unit. For the operation of viewing fund details, the interaction transfer nodes may include entering the fund detail page, browsing the historical return chart of the fund, viewing the information of the fund management team, etc. Suppose that in the data stream after interference feature cleaning, it is found that the interaction transfer event corresponding to the interaction transfer node of browsing the historical return chart of the fund is the largest (e.g., the longest stay time, the highest operation frequency, etc.). Then this node is determined as the target interaction transfer node.

[0023] Taking this target interaction transfer node as the reference transfer node, set A = 2 interaction transfer nodes in the forward direction (for example, entering the fund details page and clicking to view the fund risk level), and B = 1 interaction transfer node in the backward direction (for example, viewing the investment strategy of the fund), thus generating the set of interaction transfer nodes corresponding to this interaction data unit.

[0024] Next, decompose the interaction transfer events corresponding to this set of interaction transfer nodes to generate multiple interaction data sub-units. Assume that 3 interaction data sub-units are obtained after decomposition, and each sub-unit contains the interaction transfer events corresponding to C = 2 interaction transfer nodes. For each interaction data sub-unit, obtain the interaction state transition characteristics based on the corresponding interaction transfer events. For example, for the interaction data sub-unit containing the two interaction transfer events of entering the fund details page and clicking to view the fund risk level, its interaction state transition characteristics may include the time interval from entering the page to clicking to view the risk level, the changes in page elements before and after clicking to view the risk level, etc.

[0025] For each interaction data sub-unit in this interaction data unit, obtain its respective corresponding gain coefficient. Assume that the gain coefficient of the first interaction data sub-unit is 0.3, the second is 0.4, and the third is 0.3. Then, based on these gain coefficients, perform a weighted calculation on the interaction state transition characteristics corresponding to each interaction data sub-unit to generate the first weighted feature.

[0026] For each interaction data sub-unit, obtain the time memory flow characteristics between it and each target interaction data sub-unit (here referring to subsequent interaction data sub-units) corresponding to this interaction data sub-unit. Taking the first interaction data sub-unit as an example, determine its first boundary information, including the start and end ranges of the interaction transfer events it contains (for example, starting from entering the fund details page to ending with clicking to view the fund risk level), and the definition information of the relevant attributes involved in these interaction transfer events (for example, the page elements involved, operation permissions, etc.). For the target interaction data sub-unit, determine its second boundary information, including the interaction transfer event range, the definition information of relevant attributes, and the relative relationship information with the first interaction data sub-unit. Based on these boundary information, determine the associated range information, then analyze the interaction transfer events within the interaction data sub-unit, identify the position and importance of each interaction transfer event in the entire interaction data sub-unit, and construct the first event memory structure. At the same time, based on the associated range information, identify the position and importance of each interaction transfer event within the target interaction data sub-unit, and construct the second event memory structure. Through a series of calculations and analyses, obtain the quantified time memory flow parameters, construct the time memory flow feature model, and extract the time memory flow characteristics.

[0027] Finally, fuse the second weighted features corresponding to each interactive data subunit in the interactive data unit to generate a third weighted feature. Fusing the first weighted feature and the third weighted feature generates the discrete interactive trajectory vector corresponding to this interactive data unit. In the same way, the discrete interactive trajectory vectors corresponding to all 10 interactive data units of Mr. G on the platform can be obtained.

[0028] Step S130: Generate a transfer interactive trajectory vector based on the discrete interactive trajectory vectors corresponding to the first y interactive data units in the financial product interactive data stream. The transfer interactive trajectory vector is the global interactive trajectory vector of the financial product interactive data stream, where y is a positive integer less than x.

[0029] Suppose in Mr. G's 10 interactive data units, y = 5. For the discrete interactive trajectory vectors corresponding to the first 5 interactive data units, first obtain the gain coefficient corresponding to each discrete interactive trajectory vector. For example, the gain coefficient corresponding to the discrete interactive trajectory vector of the first interactive data unit is 0.15, the second is 0.18, the third is 0.2, the fourth is 0.22, and the fifth is 0.25.

[0030] Then perform feature space transformation on these 5 discrete interactive trajectory vectors to generate the corresponding mapped interactive trajectory vectors. This feature space transformation may involve converting the original discrete interactive trajectory vector from one representation space to another more suitable for fusion calculation. For example, converting the discrete interactive trajectory vector based on multiple dimensions such as time and operation type to a vector space representation under a certain mathematical model.

[0031] Based on the gain coefficients obtained above, fuse the mapped interactive trajectory vectors corresponding to these 5 interactive data units. The specific fusion method may be weighted summation or the like. Suppose after calculation, a new vector is generated, and this vector is the transfer interactive trajectory vector, which synthesizes the information of the first 5 interactive data units and reflects the financial product interactive trajectory characteristics of Mr. G at this stage from a global perspective.

[0032] Step S140: Generate the target interactive trajectory vector corresponding to the x-th interactive data unit based on the transfer interactive trajectory vector and the discrete interactive trajectory vector corresponding to the x-th interactive data unit.

[0033] For example, in Mr. G's example, x = 10, and the discrete interaction trajectory vector corresponding to the 10th interaction data unit has been obtained in step S120. Now integrate the transfer interaction trajectory vector generated in step S130 and the discrete interaction trajectory vector corresponding to the 10th interaction data unit. The integration method may be to combine or operate the two vectors on the corresponding dimensions. For example, addition operations are performed on certain dimensions, weighted average operations are performed on other dimensions, and so on. After such an integration operation, the new vector generated is the target interaction trajectory vector corresponding to the 10th interaction data unit. This target interaction trajectory vector contains both the global interaction trajectory information of the first 5 interaction data units and the independent interaction trajectory information of the 10th interaction data unit itself.

[0034] Step S150, based on the target interaction trajectory vector corresponding to the xth interaction data unit, predict the point of interest for the xth interaction data unit, generate a point of interest prediction tag corresponding to the xth interaction data unit, and push information to the target user based on the point of interest prediction tag corresponding to the xth interaction data unit.

[0035] Taking Mr. G’s 10th interaction data unit as an example, the corresponding target interaction trajectory vector contains rich information. The point of interest prediction label may be predicted based on multiple factors. Assume that on this Internet financial platform, the logic of the point of interest prediction label is constructed as follows:

[0036] If some dimensions in the target interaction trajectory vector indicate frequent viewing of high-risk and high-yield financial products, and a tendency to invest a small amount of money in high-risk products in recent operations, and more consultation interactions with platform customer service about high-risk products, then the point of interest prediction label may be "interested in high-risk and high-yield financial products and in the risk assessment and learning stage."

[0037] Assume that if the target interaction trajectory vector shows that the user frequently switches between time deposits and low-risk financial products and is more sensitive to interest rate fluctuations, then the predicted label of the point of interest may be "pay attention to low-risk stable return financial products and seek the optimal interest rate."

[0038] According to the target interaction trajectory vector corresponding to the 10th interaction data unit, after complex calculations and logical judgments, the corresponding point of interest prediction label is predicted to be "interested in high-risk and high-yield financial products and in the risk assessment and learning stage."

[0039] Based on the prediction of tags for this interest point, the platform can push information to Mr. G. For example, the platform can push articles on investment strategies for high-risk financial products, recommendations for high-risk funds, and some risk control training courses to Mr. G. These information pushes aim to meet the interest points shown by Mr. G during the interaction with financial products, improve the user experience, and also help the platform better serve users, promote the sales of financial products, and users' financial management.

[0040] Based on the above steps, the embodiment of the present application effectively constructs a fine-grained interaction model between users and financial products by obtaining the financial product interaction data stream of the target user on the Internet financial platform and extracting the discrete interaction trajectory vectors of each interaction data unit. Further, by generating and transmitting interaction trajectory vectors to capture the global interaction pattern of the financial product interaction data stream, and combining the discrete interaction trajectory vectors of the xth interaction data unit to generate the target interaction trajectory vector, a dynamic and global understanding of the user's interaction behavior is achieved, and the interest point prediction of the xth interaction data unit can be accurately made, generating corresponding interest point prediction tags, and based on this, precise information push to the target user is carried out. Thus, the pertinence and effectiveness of information push are improved, the user experience is enhanced, and at the same time, a more intelligent user service strategy is provided for the Internet financial platform, enhancing the user stickiness and competitiveness of the platform.

[0041] In a possible implementation manner, step S120 may include:

[0042] Step S121, cleaning the interference features of the financial product interaction data stream to generate a financial product interaction data stream after cleaning the interference features, and the interference feature data is removed from the financial product interaction data stream after cleaning the interference features.

[0043] In this embodiment, taking the interaction of Mr. G on the Internet financial platform of "Investment Platform A" as an example, the relevant steps are further described in detail.

[0044] First, it is to clean the interference features of the financial product interaction data stream. In Mr. G's interaction data, the interference features may include pop-up click records implanted on the page by the platform to promote its own business, which are irrelevant to Mr. G's actual financial product interaction, or records of some maintenance prompt information automatically carried out by the system, etc. After removing these interference feature data from the financial product interaction data stream, a financial product interaction data stream after cleaning the interference features is obtained. This cleaned data stream more purely reflects Mr. G's interaction with financial products.

[0045] Step S122, based on the financial product interaction data stream after cleaning the interference features, obtain the interaction state transition features corresponding to each of the x interaction data units.

[0046] Step S123: Based on the interaction state transition features corresponding to each of the x interaction data units, obtain the discrete interaction trajectory vectors corresponding to each of the x interaction data units.

[0047] In a possible implementation, the interaction data unit includes interaction transfer events corresponding to multiple transmitted interaction transfer nodes.

[0048] Step S122 includes:

[0049] Step S1221: Based on the financial product interaction data stream after cleaning by the interference feature, obtain the target interaction transfer nodes corresponding to each of the x interaction data units, where the target interaction transfer node is used to represent the interaction transfer node with the largest interaction transfer event among the multiple interaction transfer nodes corresponding to the interaction data unit.

[0050] Step S1222: For each of the interaction data units, using the target interaction transfer node corresponding to the interaction data unit as the reference transfer node, obtain A interaction transfer nodes in the forward direction and B interaction transfer nodes in the backward direction, and generate an interaction transfer node set corresponding to the interaction data unit, where A and B are preset positive integers.

[0051] Step S1223: Based on the interaction transfer events corresponding to each interaction transfer node in the interaction transfer node set corresponding to the interaction data unit, obtain the interaction state transition feature corresponding to the interaction data unit.

[0052] In a possible implementation, Step S1223 may include:

[0053] Step S1223-1: Decompose the interaction transfer events corresponding to the interaction transfer node set to generate multiple interaction data sub-units, where each interaction data sub-unit includes the interaction transfer events corresponding to C interaction transfer nodes.

[0054] Step S1223-2: For each of the interaction data sub-units, based on the interaction transfer events corresponding to the interaction data sub-unit, obtain the interaction state transition feature corresponding to the interaction data sub-unit.

[0055] Step S1223-3: Based on the interaction state transition features corresponding to each of the multiple interaction data sub-units, obtain the interaction state transition feature corresponding to the interaction data unit.

[0056] In this embodiment, for each interaction data unit, after obtaining the interaction state transition characteristics, it is necessary to further determine the gain coefficients corresponding to each interaction data subunit. The determination of these gain coefficients may be based on factors such as the importance of the interaction transfer events in the interaction data subunit to the entire interaction data unit. For example, for an interaction data subunit containing important operations (such as confirming a purchase), its gain coefficient may be relatively high. Based on the gain coefficients corresponding to each interaction data subunit in the interaction data unit, weighted calculations are performed on the interaction state transition characteristics corresponding to each interaction data subunit in the interaction data unit to generate a first weighted feature.

[0057] For each interaction data subunit in the interaction data unit, obtain the time memory flow characteristics between it and each target interaction data subunit (here referring to subsequent interaction data subunits) corresponding to this interaction data subunit. Taking the interaction data subunit containing clicking into the wealth management product page and performing a risk assessment as an example, determine its first boundary information, including the start and end ranges of these two interaction transfer events (such as starting from clicking into the wealth management product page to the end of completing the risk assessment), and the definition information of the relevant attributes involved in these interaction transfer events (such as the risk level options involved in the risk assessment process, etc.). For the target interaction data subunit (such as the interaction data subunit containing inputting the purchase amount and confirming the purchase), determine its second boundary information, including the interaction transfer event range, the definition information of the relevant attributes, and the relative relationship information with the previous interaction data subunit (such as the sequential relationship, etc.). Based on these boundary information, determine the associated range information, and then analyze the interaction transfer events within the interaction data subunit, identify the position and importance of each interaction transfer event in the entire interaction data subunit, and construct a first event memory structure. At the same time, based on the associated range information, identify the position and importance of each interaction transfer event within the target interaction data subunit, and construct a second event memory structure. Through a series of calculations and analyses, obtain the quantified time memory flow parameters, construct a time memory flow feature model, and extract the time memory flow characteristics.

[0058] Fuse the second weighted features corresponding to each interaction data subunit in the interaction data unit to generate a third weighted feature. Finally, fuse the first weighted feature and the third weighted feature to generate the discrete interaction trajectory vector corresponding to this interaction data unit. According to such a process, the discrete interaction trajectory vectors corresponding to all x interaction data units of Mr. G on the platform can be obtained.

[0059] Taking an interaction data unit of Mr. G as an example, such as a wealth management product purchase operation he carried out on February 1, 2023. This interaction data unit contains interaction transfer events corresponding to multiple transmitted interaction transfer nodes. These interaction transfer nodes may include operations such as logging in to the platform, searching for wealth management products, viewing the details of wealth management products, conducting risk assessments, entering the purchase amount, and confirming the purchase. Each operation is an interaction transfer event. First, the target interaction transfer node corresponding to this interaction data unit needs to be obtained. In this example, it is assumed that the interaction transfer event corresponding to the interaction transfer node of viewing the details of wealth management products is the largest. Here, "the largest interaction transfer event" can mean that in this interaction data unit, the data traffic involved in the operation of viewing the details of wealth management products is the largest (for example, the amount of data loaded during page stay is the largest) or the operation duration is the longest, etc. Then the interaction transfer node of viewing the details of wealth management products is determined as the target interaction transfer node.

[0060] For this interaction data unit, taking the target interaction transfer node of viewing the details of wealth management products as the reference transfer node, assume that the preset A = 3 and B = 2. Then obtain 3 interaction transfer nodes in the forward direction, namely logging in to the platform, searching for wealth management products, and clicking to enter the wealth management product page; and 2 interaction transfer nodes in the backward direction, namely conducting risk assessment and entering the purchase amount. In this way, the interaction transfer node set corresponding to this interaction data unit is generated.

[0061] Based on the interaction transfer events corresponding to each interaction transfer node in this interaction transfer node set, further obtain the interaction state transition characteristics corresponding to this interaction data unit. First, decompose the interaction transfer events corresponding to the interaction transfer node set to generate multiple interaction data sub-units. Assume C = 2, that is, each interaction data sub-unit includes the interaction transfer events corresponding to 2 interaction transfer nodes. For example, one interaction data sub-unit contains the interaction transfer events of logging in to the platform and searching for wealth management products, and another interaction data sub-unit contains the interaction transfer events of clicking to enter the wealth management product page and conducting risk assessment, etc. For each interaction data sub-unit, obtain the interaction state transition characteristics based on its corresponding interaction transfer events. For the interaction data sub-unit containing the interaction transfer events of logging in to the platform and searching for wealth management products, the interaction state transition characteristics may include the time interval from logging in to the platform to starting to search for wealth management products, the matching degree between the keywords used when searching for wealth management products and the wealth management products recommended by the platform, etc.

[0062] Then, based on the interaction state transition features corresponding to each of the multiple interaction data subunits, the interaction state transition feature corresponding to the entire interaction data unit is obtained. For example, the interaction state transition features corresponding to each interaction data subunit are weighted and aggregated, and the determination of the weights may be based on factors such as the importance of each interaction data subunit in the entire interaction data unit. In this way, the interaction state transition feature corresponding to this interaction data unit is obtained. In the same way, all x interaction data units of Mr. G on the platform can be processed to obtain the interaction state transition features corresponding to each interaction data unit respectively.

[0063] In a possible implementation manner, step S123 includes:

[0064] Step S1231, for each of the interaction data units, obtain the gain coefficient corresponding to each interaction data subunit in the interaction data unit.

[0065] In this embodiment, for example, Mr. G has an interaction data unit on the platform regarding the operation process of purchasing a certain fund product. This interaction data unit is decomposed into multiple interaction data subunits. One interaction data subunit includes two interaction transfer events: logging in to the platform and searching for fund products, and another interaction data subunit includes two interaction transfer events: viewing fund details and assessing risks, etc. For the interaction data subunit including logging in to the platform and searching for fund products, since logging in to the platform is the starting operation of the entire interaction, which is relatively basic and important, it may be given a relatively low gain coefficient, such as 0.2; while the operation of searching for fund products is more critical for locating the target fund and may be given a gain coefficient of 0.3. For the interaction data subunit including viewing fund details and assessing risks, viewing fund details can obtain key information about the fund and may be given a gain coefficient of 0.35, and assessing risks is very important for investment decisions and may be given a gain coefficient of 0.45. The determination of these gain coefficients is obtained by comprehensively considering various factors such as the importance of the interaction transfer events in the entire interaction data unit and the impact on subsequent operations.

[0066] Step S1232, based on the gain coefficients corresponding to each interaction data subunit in the interaction data unit, perform weighted calculation on the interaction state transition features corresponding to each interaction data subunit in the interaction data unit to generate a first weighted feature.

[0067] Continuing with the above example of the interactive data unit for purchasing fund products, assume that the interactive state transition characteristics of the interactive data subunit containing the interactive data subunits of logging in to the platform and searching for fund products include the time interval from logging in to searching (assumed to be 5 seconds), the matching degree between the search keyword and the platform-recommended fund (assumed to be 80%), etc. These interactive state transition characteristics are weighted and calculated according to the gain coefficient. For example, the weighted value of the time interval is 5×0.2 = 1, and the weighted value of the matching degree is 80%×0.3 = 0.24. Similar weighted calculations are also performed on the interactive state transition characteristics of the interactive data subunit containing viewing fund details and assessing risks, and then these weighted results are combined to generate the first weighted feature.

[0068] Step S1233: For each interactive data subunit in the interactive data unit, obtain the time memory flow characteristics between the interactive data subunit and each corresponding target interactive data subunit. Herein, the target interactive data subunit is used to represent the interactive data subunit after the interactive data subunit in the interactive data unit.

[0069] Step S1234: Fuse the time memory flow characteristics between the interactive data subunit and each corresponding target interactive data subunit to generate the second weighted feature corresponding to the interactive data subunit.

[0070] Step S1235: Fuse the second weighted features corresponding to each interactive data subunit in the interactive data unit to generate the third weighted feature.

[0071] Step S1236: Fuse the first weighted feature and the third weighted feature to generate the discrete interactive trajectory vector corresponding to the interactive data unit.

[0072] In this embodiment, assume that there are three interactive data subunits in the interactive data unit for purchasing fund products. After calculating the respective second weighted features, weights are determined according to factors such as the importance of each interactive data subunit in the entire interactive data unit, and the third weighted feature is obtained through fusion calculation. Finally, the first weighted feature and the third weighted feature are fused to generate the discrete interactive trajectory vector corresponding to the interactive data unit. This discrete interactive trajectory vector synthesizes various information such as the interactive state transition characteristics and time memory flow characteristics of each interactive data subunit in the interactive data unit, and comprehensively describes the characteristics of this interactive data unit in the entire financial product interaction process. In the same way, other interactive data units of Mr. G on the platform can be processed to obtain the discrete interactive trajectory vectors corresponding to each interactive data unit.

[0073] In a possible implementation manner, step S1233 includes:

[0074] Step S1233-1, determining first boundary information of the interaction data subunit, where the first boundary information includes the start and end ranges of the interaction transfer events included in the interaction data subunit, and the definition information of the relevant attributes involved in the interaction transfer events.

[0075] Step S1233-2, determining second boundary information of the target interaction data subunit, where the second boundary information includes the interaction transfer event range and the definition information of relevant attributes of the target interaction data subunit, and the relative relationship information with the interaction data subunit.

[0076] Step S1233-3, based on the first boundary information and the second boundary information, determining the association range information between the interaction data subunit and each corresponding target interaction data subunit, where the association range information covers various data connection information existing from the interaction data subunit to the target interaction data subunit, and the data connection information includes direct interaction transfer event association information and indirect connection information constructed by relevant attributes.

[0077] Step S1233-4, according to the association range information, analyzing the interaction transfer events in the interaction data subunit, and identifying the position and importance of each interaction transfer event in the entire interaction data subunit.

[0078] Step S1233-5, according to the logical relationship between the interaction transfer events, constructing a first event memory structure in the interaction data subunit, where the first event memory structure is presented in the form of a logical hierarchy diagram, where nodes represent interaction transfer events, and edges represent the logical connections between interaction transfer events. For each interaction transfer event, determine the position of the interaction transfer event on the time axis, and analyze the time interval between different interaction transfer events to obtain the structure information of the first event memory structure.

[0079] Step S1233-6, according to the association range information, identifying the position and importance of each interaction transfer event in the target interaction data subunit, constructing a second event memory structure by analyzing the logical relationship between each interaction transfer event, determining the position of each interaction transfer event on the time axis, analyzing the time interval between interaction transfer events, and the time sequence relationship with the events in the interaction data subunit, and generating the structure information of the second event memory structure.

[0080] Step S1233-7: Based on the structural information of the first event memory structure, the structural information of the second event memory structure, and the association range information, search for the nodes and edges connected to the event memory structure of the target interaction data subunit to obtain the basic connection part. Identify the memory transfer path along the basic connection part, and sort and classify the identified memory transfer paths according to the length, complexity, total time span, and time interval distribution of the memory transfer path, and divide the memory transfer paths into different categories.

[0081] Step S1233-8: For each memory transfer path, determine the key nodes on the memory transfer path. For each key node, calculate the time delay of the key node during the memory transfer process. The time delay refers to the time difference experienced from a starting interaction transfer event in the interaction data subunit to this key node.

[0082] Step S1233-9: Analyze the time interval distribution between different key nodes on the memory transfer path, and determine the time interval distribution characteristics of the memory transfer time interval between different key nodes.

[0083] Step S1233-10: Calculate the overall time span of the memory transfer path. The overall time span is the total time experienced from the starting interaction transfer event of the interaction data subunit to the final interaction transfer event of the target interaction data subunit. The overall time span reflects the overall time cost of the memory from the interaction data subunit to the target interaction data subunit.

[0084] Step S1233-11: Quantitatively represent the calculated time delay, time interval distribution characteristics, and overall time span to form quantified time memory flow parameters.

[0085] Step S1233-12: Determine the basic structure of the target model according to the quantified time memory flow parameters. The basic structure is a model based on a vector space, where each dimension represents a defined time memory flow parameter.

[0086] Step S1233-13: Normalize each time memory flow parameter, and fill the normalized time memory flow parameters into the corresponding dimensions of the target model to construct a time memory flow feature model.

[0087] Step S1233-14: Extract time memory flow features from the constructed time memory flow feature model according to the predefined feature extraction rules.

[0088] Specifically, taking the interactive data subunit that includes viewing fund details and assessing risks as an example, its first boundary information is determined. The start of the interactive transfer events included in this interactive data subunit is the click on the button to view fund details, and the end is the completion of the risk assessment submission. The defined information of the relevant attributes involved includes various types of information contained in the fund details page viewed (such as the investment portfolio of the fund, historical returns, etc.), the types of risk assessments (such as different levels of assessments like conservative, stable, etc.). For the target interactive data subunit (assumed to be the interactive data subunit of inputting the purchase amount and confirming the purchase), its second boundary information is determined. The range of its interactive transfer events is from the start of inputting the purchase amount to the end of confirming the purchase. The defined information of the relevant attributes includes the numerical range of the purchase amount, the payment method at the time of confirming the purchase, etc. The relative relationship information with the interactive data subunit is an order relationship, that is, the purchase operation can only be carried out after the risk assessment is completed.

[0089] Based on the first boundary information and the second boundary information, the associated range information between the interactive data subunit and each corresponding target interactive data subunit is determined. In this example, the direct interactive transfer event association information is that the purchase operation can only be carried out after the risk assessment is passed. The indirect connection information constructed by the relevant attributes includes the potential association between the risk assessment result and the purchase amount (such as a higher risk tolerance may lead to a larger purchase amount). According to the associated range information, the interactive transfer events within the interactive data subunit are analyzed. For the event of viewing fund details, it is a basic operation to obtain information in the entire interactive data subunit, which is relatively important and is at a lower level in the logical hierarchy diagram; assessing risks is a decision-making operation based on viewing the details, which is more important and is at a higher level in the logical hierarchy diagram. According to the logical relationship between the interactive transfer events, the first event memory structure within the interactive data subunit is constructed. For the event of viewing fund details, it occurs earlier on the time axis, and the time interval between it and assessing risks is assumed to be 30 seconds. Information such as the time intervals between different interactive transfer events constitutes the structural information of the first event memory structure.

[0090] According to the associated range information, the positions and importance of each interactive transfer event within the target interactive data subunit are identified. Inputting the purchase amount is a pre-operation for the purchase operation, and confirming the purchase is the final operation, and confirming the purchase is more critical. By analyzing the logical relationship between each interactive transfer event, the second event memory structure is constructed, and the position of each interactive transfer event on the time axis is determined. The time intervals between the interactive transfer events are analyzed, as well as the time sequence relationship with the events in the interactive data subunit, to generate the structural information of the second event memory structure. For example, the time interval between inputting the purchase amount and confirming the purchase is assumed to be 10 seconds.

[0091] Based on the structural information of the first event memory structure, the structural information of the second event memory structure, and the association range information, search for the nodes and edges connected to the event memory structure of the target interaction data subunit to obtain the basic connection part. Identify the memory transfer path along the basic connection part. For example, from viewing the fund details to entering the purchase amount and then to confirming the purchase might be a memory transfer path. Sort and classify the identified memory transfer paths according to the length of the memory transfer path (here it is 3 interactive transfer events), complexity (due to different types of operations involved such as risk assessment and amount input, the complexity is relatively high), total time span (assuming a total of 60 seconds from viewing the details to confirming the purchase), and time interval distribution (30 seconds from viewing the details to assessing the risk, 10 seconds from entering the amount to confirming the purchase).

[0092] For each memory transfer path, determine the key nodes on the memory transfer path. In the above memory transfer path, viewing the fund details, assessing the risk, and confirming the purchase might be the key nodes. For each key node, calculate the time delay of this key node during the memory transfer process. For the starting key node of viewing the fund details, the time delay is 0 seconds; for the key node of assessing the risk, the time difference from the starting interactive transfer event of viewing the fund details to assessing the risk is 30 seconds; for the key node of confirming the purchase, the time difference from viewing the fund details to confirming the purchase is 60 seconds. Analyze the time interval distribution between different key nodes on the memory transfer path to determine the time interval distribution characteristics of the memory transfer time interval between different key nodes. For example, the time interval from viewing the details to assessing the risk is relatively long, and the time interval from entering the amount to confirming the purchase is relatively short. Calculate the overall time span of the memory transfer path, that is, the total time experienced from the starting interactive transfer event (viewing the fund details) of the interactive data subunit to the final interactive transfer event (confirming the purchase) of the target interactive data subunit is 60 seconds. This overall time span reflects the overall time cost of the memory from the interactive data subunit to the target interactive data subunit.

[0093] Quantify the calculated time delay, time interval distribution characteristics, and overall time span to form quantified time memory flow parameters. For example, the time delay can be directly represented by the number of seconds, the time interval distribution characteristics can be represented by the proportional relationship of time intervals, and the overall time span can be represented by the total number of seconds. Based on the quantified time memory flow parameters, determine the basic structure of the target model. This basic structure is a vector space-based model, where each dimension represents a defined time memory flow parameter. For example, one dimension represents the time delay from viewing the fund details to assessing the risk, and one dimension represents the time interval from inputting the amount to confirming the purchase. Normalize each time memory flow parameter and fill the normalized time memory flow parameters into the corresponding dimensions of the target model to construct a time memory flow feature model. Extract time memory flow features from the constructed time memory flow feature model according to the predefined feature extraction rules.

[0094] In a possible implementation manner, step S130 includes:

[0095] Step S131, based on the discrete interaction trajectory vectors corresponding to the first y interaction data units, obtain the gain coefficients corresponding to the first y interaction data units.

[0096] For example, assume that there are 10 interaction data units in Mr. G's financial product interaction data stream on the platform, where y = 5. For the first interaction data unit, its corresponding discrete interaction trajectory vector contains comprehensive feature information about Mr. G's operations related to browsing various financial products (such as funds, bonds, etc.) after his first login to the platform. Determining the gain coefficient of this interaction data unit requires considering multiple factors, such as the fundamental contribution of the operations in this interaction data unit to the overall financial product interaction behavior. If the operations in this interaction data unit are the first comprehensive browsing of the platform's financial products, which lays the foundation for subsequent in-depth interactions, it may be assigned a relatively high gain coefficient, such as 0.2.

[0097] For the second interaction data unit, assume that Mr. G conducted a detailed review and preliminary evaluation of a specific fund product. This operation has a certain guiding role in the entire interaction process and has a greater impact on subsequent operations. After comprehensive consideration, it may be assigned a gain coefficient of 0.22. The third interaction data unit is that Mr. G further consulted the platform customer service about some questions regarding the fund product. This interaction data unit plays an important role in obtaining accurate information and may be assigned a gain coefficient of 0.23. The fourth interaction data unit is that Mr. G made a small amount of fund transfer operation, which serves as the fund preparation for potential subsequent financial product purchases. Considering its contribution to fund preparation, it may be assigned a gain coefficient of 0.18. The fifth interaction data unit is that Mr. G reviewed the recent earnings of the fund product again. This operation helps with the final investment decision and may be assigned a gain coefficient of 0.17.

[0098] Step S132: Perform a feature space transformation on the discrete interaction trajectory vectors corresponding to the first y interaction data units respectively to generate the mapped interaction trajectory vectors corresponding to the first y interaction data units respectively.

[0099] Taking the first interaction data unit as an example, its discrete interaction trajectory vector is a feature representation formed based on a series of previous operations (such as the time and order of browsing different financial product pages, the frequency of clicking on different sections, etc.). When performing the feature space transformation, the original representation based on these operation features may be transformed into a representation in another mathematical space. For example, the original discrete interaction trajectory vector is in a space with operation type, operation time, etc. as coordinate axes. Through a certain mathematical transformation (such as a linear transformation or a non-linear transformation, specifically it may be based on a conversion algorithm preset by the platform, and this algorithm may consider factors such as the logical relationship between different operations and the distribution characteristics of the overall interaction data), it is transformed into a space with new features as coordinate axes, thereby generating the corresponding mapped interaction trajectory vector. For the second interaction data unit, its discrete interaction trajectory vector includes the operation features of viewing different sections (such as investment portfolio, management team, etc.) of the specific fund product details page. When performing the feature space transformation, it also follows the algorithm preset by the platform to transform these features into a new space and generate a new mapped interaction trajectory vector. The third, fourth, and fifth interaction data units also generate corresponding mapped interaction trajectory vectors respectively according to the feature content of their respective discrete interaction trajectory vectors through the same feature space transformation mechanism.

[0100] Step S134: Based on the gain coefficients corresponding to the first y interaction data units respectively, fuse the mapped interaction trajectory vectors corresponding to the first y interaction data units respectively to generate the transfer interaction trajectory vector.

[0101] Finally, according to the gain coefficients determined above, multiply the mapped interaction trajectory vector corresponding to the first interaction data unit by its gain coefficient 0.2, multiply the mapped interaction trajectory vector corresponding to the second interaction data unit by 0.22, multiply the mapped interaction trajectory vector corresponding to the third interaction data unit by 0.23, multiply the mapped interaction trajectory vector corresponding to the fourth interaction data unit by 0.18, and multiply the mapped interaction trajectory vector corresponding to the fifth interaction data unit by 0.17. Then fuse these five product results, and the fusion method may be in the way of vector addition (of course, it may also be other fusion methods defined based on platform algorithms, and this algorithm will consider factors such as the status and mutual relationship of different mapped interaction trajectory vectors in the overall interaction). Through such a fusion operation, the finally generated transfer interaction trajectory vector synthesizes the information of the first y interaction data units and reflects the financial product interaction trajectory characteristics of Mr. G in this stage (the interaction stage covered by the first y interaction data units) from a more macroscopic perspective. This transfer interaction trajectory vector will serve as an important basic data for subsequent processing steps and be used for operations such as correlation analysis with other interaction data units.

[0102] In a possible implementation manner, step S140 includes: integrating the transfer interaction trajectory vector and the discrete interaction trajectory vector corresponding to the x-th interaction data unit to generate the target interaction trajectory vector corresponding to the x-th interaction data unit.

[0103] In this embodiment, during the financial product interaction process of Mr. G on the platform, the transfer interaction trajectory vector has been obtained through the processing of the first y interaction data units before. This vector synthesizes the information of the first y interaction data units after conversion, weighted fusion with different operation characteristics, and reflects the interaction trajectory characteristics of a certain stage as a whole. Now consider the x-th interaction data unit, and its corresponding discrete interaction trajectory vector is an independent interaction trajectory representation formed based on the operation characteristics of this interaction data unit itself. For example, the x-th interaction data unit is a discrete interaction trajectory vector composed of the interaction information related to Mr. G's final purchase operation of a certain specific financial product, including detailed characteristics such as purchase amount input, purchase confirmation, and payment method selection.

[0104] To generate the target interaction trajectory vector corresponding to the x-th interaction data unit, it is necessary to integrate the transmitted interaction trajectory vector and the discrete interaction trajectory vector corresponding to the x-th interaction data unit. This integration process is an operation that organically combines the information of the two vectors. Specifically, during integration, operations may be performed according to the preset algorithm for each dimension of the vector. For example, for some dimensions representing operation types, if the transmitted interaction trajectory vector indicates the operation tendencies of viewing and evaluating various financial products in this dimension, while the discrete interaction trajectory vector corresponding to the x-th interaction data unit indicates the operation of purchasing the current financial product in this dimension, then during the integration process, the information of the two in this dimension may be merged according to certain rules, which may be weighted merging or integration based on logical relationships. For the dimension representing the operation time, the time information in the transmitted interaction trajectory vector reflects the time distribution of previous interaction operations, and the time information in the discrete interaction trajectory vector corresponding to the x-th interaction data unit is the time feature of the current purchase operation. When integrating, factors such as the chronological order and time interval of the two will be considered, and this information will be fused together. Through such integration operations on the transmitted interaction trajectory vector and the discrete interaction trajectory vector corresponding to the x-th interaction data unit in each dimension, the target interaction trajectory vector corresponding to the x-th interaction data unit is finally generated. This target interaction trajectory vector not only contains the information of the previous overall interaction trajectory but also incorporates the specific interaction information of the current x-th interaction data unit.

[0105] In a possible implementation manner, the financial product interaction data stream is extracted from the initial interaction big data. The method further includes:

[0106] Step A110: Remove the first y interaction data units from the initial interaction big data to generate updated interaction big data.

[0107] Step A120: Obtain the interest point prediction labels corresponding to each of the remaining interaction data units in the updated interaction big data.

[0108] Step A130: Based on the interest point prediction labels corresponding to each of the remaining interaction data units, divide the remaining interaction data units to generate multiple interaction data unit clusters, and each interaction data unit cluster corresponds to one of the interest point prediction labels.

[0109] Step A140: Extract one of the remaining interaction data units from the multiple interaction data unit clusters respectively multiple times to generate multiple combined unit blocks.

[0110] Step A150: For each interest point prediction process description model, obtain the influence weights of the interest point prediction process description model under the limitation of each of the combined unit blocks. The influence weight represents the proportion of the positive contribution of the first remaining interaction data unit in the combined unit block to the positive contribution of the combined unit block.

[0111] Step A160: Based on the influence weights of the interest point prediction process description model under the limitation of each of the combined unit blocks, obtain the target influence weight corresponding to the interest point prediction process description model.

[0112] Step A170: Based on the target influence weights corresponding to each of the interest point prediction process description models, determine the interest point prediction process description model that matches the initial interaction big data.

[0113] Among them, Step A150 includes:

[0114] Step A151: For each remaining interaction data unit in the combined unit block, obtain the cumulative value of the positive contribution between the remaining interaction data unit and the interest point prediction process description model. The cumulative value of the positive contribution is used to represent the sum of the positive contributions between each interaction transfer event in the remaining interaction data unit and the interest point prediction process description model under the limitation of the interest point prediction label corresponding to the remaining interaction data unit.

[0115] Step A152: Add up the cumulative values of the positive contributions between each remaining interaction data unit in the combined unit block and the interest point prediction process description model respectively to generate a first addition result.

[0116] Step A153: Divide the cumulative value of the positive contribution between the first remaining interaction data unit and the interest point prediction process description model by the first addition result to obtain the influence weight of the interest point prediction process description model under the limitation of the combined unit block.

[0117] First of all, the financial product interaction data stream is extracted from the initial interaction big data. In Mr. G's case, the initial interaction big data contains all the original data related to his financial product interactions on the "Investment Platform A". These data come from a wide range of sources and cover various operations after he logs in to the platform, such as browsing different financial product pages, querying interest rates, viewing product details, performing fund operations (depositing, withdrawing, transferring, etc.), and interacting with the platform customer service.

[0118] Next, remove the first y interaction data units from the initial interaction big data to generate updated interaction big data. Assuming y = 5, for Mr. G, these first 5 interaction data units may be some basic operations after he initially logged in to the platform, such as viewing the platform welcome page during the first login, checking basic account information, initially browsing the financial product categories in the platform navigation bar, and simply viewing some popular recommended financial products. After removing these 5 interaction data units, the remaining interaction data units constitute the updated interaction big data.

[0119] Then, obtain the interest point prediction labels corresponding to each of the remaining interaction data units in the updated interaction big data. Taking one of Mr. G's remaining interaction data units as an example, assume this interaction data unit is his operation of conducting a detailed study on a specific high-risk fund product, including viewing detailed information such as the historical return trend, investment portfolio, and management team of the fund, and continuously following the dynamics of the fund for some time after that. Based on these operations, the interest point prediction label may be determined as "having an in-depth research interest in high-risk fund products and paying attention to long-term return potential". For other remaining interaction data units, such as Mr. G's operation of re-evaluating the interest rate adjustment of time deposit products, the corresponding interest point prediction label may be "paying attention to the interest rate changes and stability of time deposits", etc.

[0120] Based on the interest point prediction labels corresponding to each of the remaining interaction data units, divide the remaining interaction data units to generate multiple interaction data unit clusters, with each interaction data unit cluster corresponding to one interest point prediction label. Continuing with Mr. G's example, all the remaining interaction data units marked as "having an in-depth research interest in high-risk fund products and paying attention to long-term return potential" will be divided into one interaction data unit cluster, and those marked as "paying attention to the interest rate changes and stability of time deposits" will be divided into another interaction data unit cluster. In this way, the remaining interaction data units in the updated interaction big data are classified according to the interest point prediction labels in this manner.

[0121] After that, extract one remaining interaction data unit from multiple interaction data unit clusters separately multiple times to generate multiple combined unit blocks. For example, extract one interaction data unit from the interaction data unit cluster of "having an in-depth research interest in high-risk fund products and paying attention to long-term return potential", and then extract one interaction data unit from the interaction data unit cluster of "paying attention to the interest rate changes and stability of time deposits", and combine these two interaction data units together to form a combined unit block. By operating in this way multiple times, multiple different combined unit blocks can be generated.

[0122] For each point of interest prediction process description model, obtain the influence weights of the point of interest prediction process description model under the limitations of each combined unit block. Here, the point of interest prediction process description model is a model structure used to predict points of interest for interactive data units, which contains a series of algorithms and rules to analyze various features in the interactive data units to determine the point of interest prediction labels. Taking a specific point of interest prediction process description model as an example, assume that this model will focus on factors such as the frequency of operations, the amount involved in the operations, and the risk level of the operation objects when analyzing interactive data units. For a combined unit block, it contains interactive data units extracted from different interactive data unit clusters.

[0123] For each remaining interactive data unit in the combined unit block, obtain the cumulative value of the positive contribution degrees between the remaining interactive data unit and the point of interest prediction process description model. Taking an interactive data unit from the "interested in in-depth research on high-risk fund products and concerned about long-term income potential" interactive data unit cluster in the combined unit block as an example, the interactive transfer events in this interactive data unit include operations such as viewing the historical income trend and portfolio of high-risk funds. Under the limitation of the point of interest prediction label "interested in in-depth research on high-risk fund products and concerned about long-term income potential" corresponding to this interactive data unit, the calculation of the positive contribution degree between it and the point of interest prediction process description model is as follows: For the interactive transfer event of viewing the historical income trend, since it is an important basis for analyzing the long-term income potential of high-risk fund products, according to the evaluation criteria for the importance of operations in the point of interest prediction process description model, it may be given a relatively high positive contribution degree, such as 0.3; The operation of viewing the portfolio is also crucial for evaluating the risk and income of the fund, and may be given a positive contribution degree of 0.25. Add up the positive contribution degrees of each interactive transfer event in this interactive data unit to obtain the cumulative value of the positive contribution degree between this remaining interactive data unit and the point of interest prediction process description model, assumed to be 0.55.

[0124] Add up the cumulative values of the positive contribution degrees between each remaining interactive data unit in the combined unit block and the point of interest prediction process description model respectively to generate the first addition result. Assume that there is another interactive data unit from the "concerned about the change and stability of the fixed deposit interest rate" interactive data unit cluster in the combined unit block, and its cumulative value of the positive contribution degree is 0.4. Then the first addition result is 0.55 + 0.4 = 0.95.

[0125] Divide the cumulative value of the positive contribution degree between the first remaining interactive data unit and the point-of-interest prediction process description model by the first addition result to obtain the influence weight of the point-of-interest prediction process description model under the combined unit block limitation. In the above example, the first remaining interactive data unit is an interactive data unit from the interactive data unit cluster of "having an in-depth research interest in high-risk fund products and paying attention to long-term return potential", and its cumulative positive contribution value is 0.55. Then, the influence weight of the point-of-interest prediction process description model under this combined unit block limitation is 0.55 / 0.95 ≈ 0.579.

[0126] Based on the influence weights of the point-of-interest prediction process description model under the limitations of each combined unit block respectively, obtain the target influence weight corresponding to the point-of-interest prediction process description model. By performing the above calculations on multiple combined unit blocks, obtain the influence weights of the point-of-interest prediction process description model under different combined unit blocks, and then synthesize these influence weights according to a certain algorithm (such as weighted average, etc.) to obtain the target influence weight.

[0127] Finally, based on the target influence weights corresponding to each point-of-interest prediction process description model respectively, determine the point-of-interest prediction process description model that matches the initial interactive big data. Assume that there are multiple point-of-interest prediction process description models, and each model has its corresponding target influence weight. By comparing these target influence weights, select the point-of-interest prediction process description model with the highest target influence weight. This model is the point-of-interest prediction process description model that best matches the initial interactive big data, and it can most accurately perform point-of-interest prediction based on the interactive data units in the initial interactive big data, thereby providing a better basis for user interest analysis and service decision-making for the platform.

[0128] In a possible implementation manner, the method further includes:

[0129] Step S101, obtain the to-be-learned sample data, where the to-be-learned sample data includes x sample interactive data units that are updated stage by stage in time series, and each sample interactive data unit carries a labeled point-of-interest prediction label, and x is a positive integer.

[0130] In this embodiment, during the learning process of constructing the financial product interest point prediction network, it is necessary to obtain sufficient sample data to be learned. For the "Investment Platform A", these sample data to be learned are derived from the financial product interaction records of numerous users (including Mr. G) on the platform. Suppose the sample data to be learned here includes x sample interaction data units that are updated stage by stage in time series, where x is a positive integer. For example, for Mr. G's sample interaction data unit, it may contain his operation records of different financial products within a certain period of time. Each sample interaction data unit carries an annotated interest point prediction label, and these labels are determined in advance according to the actual behavior of the user and clear business logic.

[0131] Taking a sample interaction data unit of Mr. G as an example, assume that this sample interaction data unit is related to his purchase operation of a certain financial product. This sample interaction data unit contains detailed operation information, such as operation time, operation type (purchase), name of the financial product involved, purchase amount, etc. The corresponding annotated interest point prediction label may be "having a purchase tendency for stable income financial products". This label is obtained by analyzing Mr. G's historical behavior, his other relevant operations on the platform (such as browsing similar products before, attention to income stability, etc.), and his account basic information (such as the risk tolerance assessment result, etc.).

[0132] Step S102: Use the financial product interest point prediction network to obtain the sample discrete interaction trajectory vectors corresponding to each of the x sample interaction data units in the sample data to be learned.

[0133] Next, the financial product interest point prediction network is a network model specifically used to analyze financial product interaction data and predict interest points. For each sample interaction data unit of Mr. G, the network will generate a sample discrete interaction trajectory vector according to the various operation information contained therein.

[0134] Still taking the sample interaction data unit of Mr. G regarding the purchase operation of the financial product as an example, the network will analyze factors such as the sequence relationship between the operation time and other operations, the association between the operation type and his previous operation types (such as whether there was a viewing operation of similar financial products before), and the proportional relationship between the purchase amount and his account fund scale and historical investment amount. After complex calculations and feature extractions, these factors are integrated into a sample discrete interaction trajectory vector. This vector comprehensively describes the characteristics of this sample interaction data unit in the entire financial product interaction process in a specific mathematical representation form.

[0135] Step S103, using the financial product interest point prediction network to generate a sample transfer interaction trajectory vector based on the sample discrete interaction trajectory vectors corresponding to the first y sample interaction data units in the sample data to be learned.

[0136] Then, assume that in Mr. G's sample data to be learned, y = 5. For the first five sample interaction data units, each unit has its corresponding sample discrete interaction trajectory vector. The financial product interest point prediction network will comprehensively consider the information of these five vectors to generate the sample transfer interaction trajectory vector.

[0137] For example, the first sample interaction data unit may be Mr. G's basic browsing operation after logging into the platform for the first time. Its sample discrete interaction trajectory vector contains information such as the sections he browsed and the duration of his stay; the second sample interaction data unit is his initial viewing operation of a certain type of fund product, including the type of fund viewed, viewing duration and other information. The network will process these sample discrete interaction trajectory vectors according to the preset algorithm. This algorithm may take into account factors such as the temporal sequence of each sample interaction data unit, the importance weight of different operations (such as operations that have a more direct impact on purchase decisions have a higher weight). By performing feature fusion, weighted calculation and other operations on these five sample discrete interaction trajectory vectors, a sample transfer interaction trajectory vector is finally generated. This sample transfer interaction trajectory vector reflects the overall characteristics of the interaction stage covered by the first five sample interaction data units from a more macro perspective.

[0138] Step S104: generating a target sample interaction trajectory vector corresponding to the x-th sample interaction data unit based on the sample transfer interaction trajectory vector and the sample discrete interaction trajectory vector corresponding to the x-th sample interaction data unit.

[0139] Assume that x = 10. For the 10th sample interaction data unit (for example, the sample interaction data unit related to Mr. G's large investment operation in a high-risk financial product), the corresponding sample discrete interaction trajectory vector contains unique information about this investment operation, such as a large investment amount and a high-risk investment product.

[0140] The financial product interest point prediction network will integrate the previously generated sample transfer interaction trajectory vector (which synthesizes the information of the first 5 sample interaction data units) and the sample discrete interaction trajectory vector corresponding to the 10th sample interaction data unit. This integration process may involve operations such as vector addition, multiplication, etc., as well as weighted processing according to the importance of information in different dimensions. For example, for the dimension representing the operation type, if the sample transfer interaction trajectory vector reflects the comprehensive trend of various previous operation types in this dimension, while the sample discrete interaction trajectory vector corresponding to the 10th sample interaction data unit highlights the operation type of this high-risk investment in this dimension, the network will fuse these two pieces of information according to the preset rules. For the dimension representing the operation amount, a similar approach will be taken to consider the amount values of both and their relative importance in the entire interaction process for fusion. Through such operations, the target sample interaction trajectory vector corresponding to the 10th sample interaction data unit is finally generated.

[0141] Step S105: Using the financial product interest point prediction network, based on the target sample interaction trajectory vector corresponding to the xth sample interaction data unit, perform interest point prediction on the xth sample interaction data unit to generate a training interest point prediction label corresponding to the xth interaction data unit.

[0142] For the 10th sample interaction data unit (Mr. G's large-scale investment operation in high-risk financial products), the financial product interest point prediction network will perform interest point prediction based on its corresponding target sample interaction trajectory vector.

[0143] The financial product interest point prediction network will analyze various information in the target sample interaction trajectory vector. For example, a large operation amount may indicate that Mr. G has high expectations for high returns, and the type of high-risk product invested shows that he may have a high risk tolerance and is interested in high-risk and high-return financial products, etc. Based on these analysis results, combined with the interest point prediction logic pre-constructed inside the network (this logic is constructed based on a large amount of historical data and financial business knowledge), a training interest point prediction label corresponding to the 10th sample interaction data unit is generated. Suppose this training interest point prediction label is "has a strong investment interest in high-risk and high-return financial products and has a relatively high risk tolerance".

[0144] Step S106: Based on the training interest point prediction label and the labeled interest point prediction label corresponding to the xth sample interaction data unit, optimize the parameters of the financial product interest point prediction network to generate an optimized financial product interest point prediction network.

[0145] For the 10th sample interaction data unit, its labeled interest point prediction label is "has a strong investment interest in high-risk and high-return financial products and has a relatively high risk tolerance" (hypothesized previously), and the training interest point prediction label generated by the financial product interest point prediction network is also "has a strong investment interest in high-risk and high-return financial products and has a relatively high risk tolerance".

[0146] If the two are exactly the same, it means that the network's prediction for this sample interaction data unit is accurate. However, if there are differences, for example, the labeled interest point prediction label is "has a certain interest in high-risk and high-return financial products but has a medium risk tolerance", while the training interest point prediction label is "has a strong investment interest in high-risk and high-return financial products and has a relatively high risk tolerance", it means that there is a deviation in the network's prediction. At this time, the parameters of the financial product interest point prediction network will be adjusted according to a preset optimization algorithm. This optimization algorithm may be based on principles such as error backpropagation. According to the difference between the labeled interest point prediction label and the training interest point prediction label, it calculates the influence degree of each parameter on the prediction result, and then adjusts the parameters in the network accordingly. By performing such operations on a large number of sample interaction data units, the parameters of the financial product interest point prediction network are continuously optimized, and finally an optimized financial product interest point prediction network is generated. This optimized network can more accurately predict interest points based on the characteristics of the sample interaction data unit, thereby improving the prediction accuracy in practical applications.

[0147] Figure 2 FIG. shows the hardware structure diagram of the big data push processing system 100 based on Internet finance provided by the embodiments of the present invention for implementing the above-mentioned big data push processing method based on Internet finance, as Figure 2 shown, the big data push processing system 100 based on Internet finance may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

[0148] The machine-readable storage medium 120 can store data and / or instructions. In some embodiments, the machine-readable storage medium 120 can store data obtained from external terminals. In some embodiments, the machine-readable storage medium 120 can store the data and / or instructions used by the big data push processing system 100 based on Internet finance to execute or use to complete the exemplary methods described in the present invention.

[0149] In the specific implementation process, one or more processors 110 execute computer-executable instructions stored in the machine-readable storage medium 120, enabling the processors 110 to execute the big data push processing method based on Internet finance in the above method embodiments. The processors 110, the machine-readable storage medium 120, and the communication unit 140 are connected through the bus 130, and the processors 110 can be used to control the transceiver actions of the communication unit 140.

[0150] For the specific implementation process of the processors 110, reference can be made to the respective method embodiments executed by the above-mentioned big data push processing system 100 based on Internet finance. Their implementation principles and technical effects are similar, and will not be elaborated here in this embodiment.

[0151] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the big data push processing method based on Internet finance as described above is implemented.

[0152] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing, or description thereof.

Claims

1. A big data push processing method based on Internet finance, characterized in that The method includes: Obtain the financial product interaction data stream of the target user on the Internet financial platform, where the financial product interaction data stream includes x interaction data units that are updated stage by stage in time series, and x is a positive integer; Obtain the discrete interaction trajectory vectors corresponding to each of the x interaction data units in the financial product interaction data stream, where the discrete interaction trajectory vector is the independent interaction trajectory vector of each interaction data unit; Based on the discrete interaction trajectory vectors corresponding to the first y interaction data units in the financial product interaction data stream, generate a transfer interaction trajectory vector, where the transfer interaction trajectory vector is the global interaction trajectory vector of the financial product interaction data stream, and y is a positive integer less than x; Based on the transfer interaction trajectory vector and the discrete interaction trajectory vector corresponding to the xth interaction data unit, generate the target interaction trajectory vector corresponding to the xth interaction data unit; Based on the target interaction trajectory vector corresponding to the xth interaction data unit, perform interest point prediction on the xth interaction data unit, generate an interest point prediction label corresponding to the xth interaction data unit, and perform information push on the target user based on the interest point prediction label corresponding to the xth interaction data unit; The generating a transfer interaction trajectory vector based on the discrete interaction trajectory vectors corresponding to the first y interaction data units in the financial product interaction data stream includes: Based on the discrete interaction trajectory vectors corresponding to the first y interaction data units, obtain the gain coefficients corresponding to the first y interaction data units; Perform feature space transformation on the discrete interaction trajectory vectors corresponding to the first y interaction data units to generate the mapped interaction trajectory vectors corresponding to the first y interaction data units; Based on the gain coefficients corresponding to the first y interaction data units, fuse the mapped interaction trajectory vectors corresponding to the first y interaction data units to generate the transfer interaction trajectory vector; The generating the target interaction trajectory vector corresponding to the xth interaction data unit based on the transfer interaction trajectory vector and the discrete interaction trajectory vector corresponding to the xth interaction data unit includes: Integrate the transfer interaction trajectory vector and the discrete interaction trajectory vector corresponding to the xth interaction data unit to generate the target interaction trajectory vector corresponding to the xth interaction data unit.

2. The big data push processing method based on Internet finance according to claim 1, wherein The obtaining the discrete interaction trajectory vectors corresponding to each of the x interaction data units in the financial product interaction data stream includes: Perform interference feature cleaning on the financial product interaction data stream to generate a financial product interaction data stream after interference feature cleaning, and the interference feature data is removed from the financial product interaction data stream after interference feature cleaning; Based on the financial product interaction data stream after interference feature cleaning, obtain the interaction state transition features corresponding to each of the x interaction data units; Based on the interaction state transition features corresponding to each of the x interaction data units, obtain the discrete interaction trajectory vectors corresponding to each of the x interaction data units.

3. The big data push processing method based on Internet finance according to claim 2, characterized in that, The interaction data unit includes interaction transfer events corresponding to multiple transferred interaction transfer nodes; Obtaining the interaction state transition features corresponding to each of the x interaction data units based on the financial product interaction data stream after cleaning based on the interference features, includes: Based on the financial product interaction data stream after cleaning based on the interference features, obtaining the target interaction transfer nodes corresponding to each of the x interaction data units, where the target interaction transfer node is used to represent the interaction transfer node with the largest interaction transfer event among the multiple interaction transfer nodes corresponding to the interaction data unit; For each of the interaction data units, taking the target interaction transfer node corresponding to the interaction data unit as the reference transfer node, obtaining A interaction transfer nodes in the forward direction and B interaction transfer nodes in the backward direction, and generating an interaction transfer node set corresponding to the interaction data unit, where A and B are preset positive integers; Based on the interaction transfer events corresponding to each interaction transfer node in the interaction transfer node set corresponding to the interaction data unit, obtaining the interaction state transition features corresponding to the interaction data unit.

4. The big data push processing method based on Internet finance according to claim 3, characterized in that The obtaining the interaction state transition features corresponding to the interaction data unit based on the interaction transfer events corresponding to each interaction transfer node in the interaction transfer node set corresponding to the interaction data unit, includes: Decomposing the interaction transfer events corresponding to the interaction transfer node set to generate multiple interaction data sub-units, where the interaction data sub-units include the interaction transfer events corresponding to C interaction transfer nodes; For each of the interaction data sub-units, based on the interaction transfer events corresponding to the interaction data sub-unit, obtaining the interaction state transition features corresponding to the interaction data sub-unit; Based on the interaction state transition features corresponding to each of the multiple interaction data sub-units, obtaining the interaction state transition features corresponding to the interaction data unit.

5. The big data push processing method based on Internet finance according to claim 4, characterized in that, The obtaining the discrete interaction trajectory vectors corresponding to each of the x interaction data units based on the interaction state transition features corresponding to each of the x interaction data units, includes: For each of the interaction data units, obtaining the gain coefficients corresponding to each of the interaction data sub-units in the interaction data unit; Based on the gain coefficients corresponding to each of the interaction data sub-units in the interaction data unit, performing weighted calculation on the interaction state transition features corresponding to each of the interaction data sub-units in the interaction data unit to generate a first weighted feature; For each of the interaction data sub-units in the interaction data unit, obtaining the time memory flow features between the interaction data sub-unit and each of the target interaction data sub-units corresponding to the interaction data sub-unit; where the target interaction data sub-unit is used to represent the interaction data sub-unit after the interaction data sub-unit in the interaction data unit; Fusing the time memory flow features between the interaction data sub-unit and each of the target interaction data sub-units corresponding to the interaction data sub-unit to generate a second weighted feature corresponding to the interaction data sub-unit; Fusing the second weighted features corresponding to each of the interaction data sub-units in the interaction data unit to generate a third weighted feature; Fuse the first weighted feature and the third weighted feature to generate a discrete interaction trajectory vector corresponding to the interaction data unit; Among them, the step of obtaining the time memory flow feature between the interaction data subunit and each target interaction data subunit corresponding to the interaction data subunit includes: Determine the first boundary information of the interaction data subunit, where the first boundary information includes the start and end ranges of the interaction transfer events included in the interaction data subunit, and the definition information of the relevant attributes involved in the interaction transfer events; Determine the second boundary information of the target interaction data subunit, where the second boundary information includes the interaction transfer event range and the definition information of the relevant attributes of the target interaction data subunit, and the relative relationship information with the interaction data subunit; Based on the first boundary information and the second boundary information, determine the associated range information between the interaction data subunit and each target interaction data subunit corresponding to the interaction data subunit. The associated range information covers various data connection information existing from the interaction data subunit to the target interaction data subunit. The data connection information includes direct interaction transfer event association information and indirect connection information constructed by relevant attributes; According to the associated range information, analyze the interaction transfer events in the interaction data subunit, and identify the position and importance of each interaction transfer event in the entire interaction data subunit; According to the logical relationship between interaction transfer events, construct a first event memory structure within the interaction data subunit. The first event memory structure is presented in the form of a logical hierarchy diagram, where nodes represent interaction transfer events and edges represent the logical connections between interaction transfer events; For each interaction transfer event, determine the position of the interaction transfer event on the time axis, and analyze the time interval between different interaction transfer events to obtain the structure information of the first event memory structure; And, according to the associated range information, identify the position and importance of each interaction transfer event in the target interaction data subunit, construct a second event memory structure by analyzing the logical relationship between each interaction transfer event, determine the position of each interaction transfer event on the time axis, analyze the time interval between interaction transfer events, and the time sequence relationship with the events in the interaction data subunit, and generate the structure information of the second event memory structure; Based on the structure information of the first event memory structure, the structure information of the second event memory structure, and the associated range information, find the nodes and edges connected to the event memory structure of the target interaction data subunit to obtain the basic connection part, identify the memory transfer path along the basic connection part, and sort and classify the identified memory transfer paths according to the length, complexity, total time span, and time interval distribution of the memory transfer path, and divide the memory transfer path into different categories; For each memory transfer path, determine the key nodes on the memory transfer path. For each key node, calculate the time delay of the key node during the memory transfer. The time delay refers to the time difference experienced from a starting interaction transfer event in the interaction data subunit to the key node. Analyze the time interval distribution between different key nodes on the memory transfer path, and determine the time interval distribution characteristics of the memory transfer time intervals between different key nodes. Calculate the overall time span of the memory transfer path. The overall time span is the total time experienced from the starting interaction transfer event of the interaction data subunit to the final interaction transfer event of the target interaction data subunit. The overall time span reflects the overall time cost of the memory from the interaction data subunit to the target interaction data subunit. Quantitatively represent the calculated time delay, time interval distribution characteristics, and overall time span to form quantified time memory flow parameters. Based on the quantified time memory flow parameters, determine the basic structure of the target model. The basic structure is a model based on a vector space, where each dimension represents a defined time memory flow parameter. Perform normalization processing on each time memory flow parameter, and fill the normalized time memory flow parameters into the corresponding dimensions of the target model to construct a time memory flow feature model. Extract time memory flow features from the constructed time memory flow feature model according to predefined feature extraction rules.

6. The big data push processing method based on Internet finance according to claim 1, characterized in that The financial product interaction data stream is extracted from the initial interaction big data. The method further includes: Remove the first y interaction data units from the initial interaction big data to generate updated interaction big data. Obtain the interest point prediction labels corresponding to each of the remaining interaction data units in the updated interaction big data. Based on the interest point prediction labels corresponding to each of the remaining interaction data units, divide the remaining interaction data units to generate multiple interaction data unit clusters, and each interaction data unit cluster corresponds to one of the interest point prediction labels. Extract one of the remaining interaction data units from the multiple interaction data unit clusters multiple times and combine them to generate multiple combined unit blocks. For each interest point prediction process description model, obtain the influence weights of the interest point prediction process description model under the limitation of each of the combined unit blocks. The influence weight represents the proportion of the positive contribution degree of the first remaining interaction data unit in the combined unit block to the positive contribution degree of the combined unit block. Based on the influence weights of the interest point prediction process description model under the limitation of each of the combined unit blocks, obtain the target influence weight corresponding to the interest point prediction process description model. Based on the target influence weights corresponding to each of the interest point prediction process description models, determine the interest point prediction process description model that matches the initial interaction big data. Among them, obtaining the influence weights of the interest point prediction process description model under the limitation of each of the combined unit blocks includes: For each remaining interaction data unit in the combined unit block, obtain the cumulative value of the positive contribution degree between the remaining interaction data unit and the interest point prediction process description model, where the cumulative value of the positive contribution degree is used to represent the sum of the positive contribution degrees between each interaction transfer event in the remaining interaction data unit and the interest point prediction process description model under the limitation of the interest point prediction label corresponding to the remaining interaction data unit; Add up the cumulative values of the positive contribution degrees between each remaining interaction data unit in the combined unit block and the interest point prediction process description model respectively to generate a first addition result; Divide the cumulative value of the positive contribution degree between the first remaining interaction data unit and the interest point prediction process description model by the first addition result to obtain the influence weight of the interest point prediction process description model under the limitation of the combined unit block.

7. The big data push processing method based on Internet finance according to claim 1, characterized in that The method further includes: Obtain the sample data to be learned, where the sample data to be learned includes x sample interaction data units that are updated stage by stage in time series, each of the sample interaction data units carries a labeled interest point prediction label, and x is a positive integer; Use the financial product interest point prediction network to obtain the sample discrete interaction trajectory vectors corresponding to the x sample interaction data units in the sample data to be learned; Use the financial product interest point prediction network to generate a sample transfer interaction trajectory vector based on the sample discrete interaction trajectory vectors corresponding to the first y sample interaction data units in the sample data to be learned; Generate a target sample interaction trajectory vector corresponding to the x-th sample interaction data unit based on the sample transfer interaction trajectory vector and the sample discrete interaction trajectory vector corresponding to the x-th sample interaction data unit; Use the financial product interest point prediction network to perform interest point prediction on the x-th sample interaction data unit based on the target sample interaction trajectory vector corresponding to the x-th sample interaction data unit, and generate a training interest point prediction label corresponding to the x-th interaction data unit; Optimize the parameters of the financial product interest point prediction network based on the training interest point prediction label corresponding to the x-th sample interaction data unit and the labeled interest point prediction label, and generate an optimized financial product interest point prediction network.

8. A big data push processing system based on Internet finance, characterized in that, The big data push processing system based on Internet finance includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the big data push processing method based on Internet finance according to any one of claims 1-7 above.

Images