A feature correlation analysis method and device in an insurance advertisement delivery scene and related products
By constructing a scale code from the relevance scores of the sliced and spliced feature sequences, and combining the relevance coefficient and intention feature information, the problem of inaccurate feature selection in traditional methods is solved, and more precise advertising strategy optimization is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING SHUISHOU INTELLIGENT TECHNOLOGY CO LTD
- Filing Date
- 2026-03-18
- Publication Date
- 2026-06-05
AI Technical Summary
In insurance advertising scenarios, traditional feature relevance analysis methods are difficult to accurately screen feature sequences that match specific intent information, resulting in insufficient accuracy in advertising placement.
By constructing slice sequences of selected feature sequences under different sorting strategies and calculating their relevance scores to the original sorted sequences, a scale code is generated. The candidate feature sequences are then filtered by combining their relevance coefficients and intended feature information.
It improves the accuracy and robustness of feature selection, optimizes advertising strategies, and can more accurately identify feature combinations that match specific intent information.
Smart Images
Figure CN122153487A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus and related products for feature correlation analysis in insurance advertising scenarios. Background Technology
[0002] In the field of insurance advertising, accurately targeting potential customers is key to improving conversion rates and reducing customer acquisition costs. To optimize advertising strategies, it is usually necessary to analyze massive amounts of user characteristics (such as age, region, browsing history, insurance preferences, etc.) to identify the combination of characteristics most relevant to users' purchase intentions.
[0003] Traditional methods typically use statistical indicators such as the Pearson correlation coefficient to measure the linear correlation between features. However, in complex insurance advertising scenarios, the relationships between features are often non-linear, and the correlation coefficient of a single feature cannot fully reflect its value in a specific feature combination. For example, a feature with low relevance to the target intent may produce a strong synergistic effect when combined with another feature. Therefore, relying solely on traditional correlation coefficients makes it difficult to accurately select feature sequences that truly match specific intent information from numerous candidate features, resulting in insufficient accuracy in advertising targeting. Summary of the Invention
[0004] In view of the above problems, this application is made to provide a method, apparatus, and related products for feature relevance analysis in insurance advertising scenarios that overcomes or at least partially solves the above problems. The technical solution is as follows: Firstly, a method for feature correlation analysis in insurance advertising scenarios is provided, the method comprising: Acquire insurance advertising placement data and determine the target feature sequence of the insurance advertising placement data as the selected feature sequence; The selected feature sequences are sorted according to the first sorting strategy to obtain the first sorted sequence corresponding to the selected feature sequences; the selected feature sequences are sorted according to the second sorting strategy to obtain the second sorted sequence corresponding to the selected feature sequences. The first sorted sequence is divided into equal parts according to a preset number to obtain a preset number of first-class slices; based on the preset number of first-class slices, multiple first-class splicing sequences are determined; the correlation between each first-class splicing sequence and the first sorted sequence is calculated to obtain the correlation score between each first-class splicing sequence and the first sorted sequence. The second sorting sequence is divided into equal parts according to a preset number to obtain a preset number of second-type slices; based on the preset number of second-type slices, multiple second-type splicing sequences are determined; the correlation between each second-type splicing sequence and the first sorting sequence is calculated to obtain the correlation score between each second-type splicing sequence and the first sorting sequence. Based on the correlation scores between each first-class spliced sequence and the first sorted sequence, and the correlation scores between each second-class spliced sequence and the first sorted sequence, the tick code of the selected feature sequence is determined. For multiple candidate feature sequences of insurance advertising data, calculate the correlation coefficient between each candidate feature sequence and the selected feature sequence. Based on the tick code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intention feature information of the selected feature sequence, the feature sequence that matches the intention feature information is selected from multiple candidate feature sequences, and is called the intention feature sequence.
[0005] In one possible implementation, based on a preset number of first-type slices, multiple first-type splicing sequences are determined, including: In a preset number of n first-class slices, the first k first-class slices are retained, the remaining nk first-class slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple first-class splicing sequences; where k takes any value from 1 to n.
[0006] In one possible implementation, based on a preset number of second-type slices, multiple second-type splicing sequences are determined, including: In a preset number of n second-type slices, the first m second-type slices are retained, the remaining nm second-type slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple second-type splicing sequences; where m takes any value from 1 to n.
[0007] In one possible implementation, the tick code of the selected feature sequence includes multiple tick values. Each tick value corresponds to the percentage of overlapping slices and the correlation score of a first-class spliced sequence and a first-sorted sequence. The positions of the multiple tick values are sorted in ascending order of their correlation scores.
[0008] In one possible implementation, based on the tick code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intentional feature information of the selected feature sequence, feature sequences matching the intentional feature information are selected from multiple candidate feature sequences, including: Based on the correlation coefficient between each candidate feature sequence and the selected feature sequence, match the scale value corresponding to the scale code of the selected feature sequence for the correlation coefficient between each candidate feature sequence and the selected feature sequence. Based on the correlation coefficient between each candidate feature sequence and the selected feature sequence, the scale value corresponding to the scale code of the selected feature sequence, and the intentional feature information of the selected feature sequence, feature sequences that match the intentional feature information are selected from multiple candidate feature sequences.
[0009] Secondly, a feature relevance analysis device for insurance advertising scenarios is provided, the device comprising: The first determining unit is used to acquire insurance advertising placement data and determine the target feature sequence of the insurance advertising placement data as the selected feature sequence. The sorting unit is used to sort the selected feature sequence according to a first sorting strategy to obtain a first sorted sequence corresponding to the selected feature sequence; and to sort the selected feature sequence according to a second sorting strategy to obtain a second sorted sequence corresponding to the selected feature sequence. The first calculation unit is used to divide the first sorted sequence into equal parts according to a preset number to obtain a preset number of first-type slices; based on the preset number of first-type slices, determine multiple first-type splicing sequences; calculate the correlation between each first-type splicing sequence and the first sorted sequence to obtain the correlation score between each first-type splicing sequence and the first sorted sequence. The second calculation unit is used to divide the second sorting sequence into equal parts according to a preset number to obtain a preset number of second-type slices; based on the preset number of second-type slices, determine multiple second-type splicing sequences; calculate the correlation between each second-type splicing sequence and the first sorting sequence to obtain the correlation score between each second-type splicing sequence and the first sorting sequence. The second determining unit is used to determine the tick code of the selected feature sequence based on the correlation scores of each first type of spliced sequence and the first sorted sequence, and the correlation scores of each second type of spliced sequence and the first sorted sequence. The analysis unit is used to calculate the correlation coefficient between each candidate feature sequence and the selected feature sequence for multiple candidate feature sequences of insurance advertising data; based on the scale code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intention feature information of the selected feature sequence, the unit selects the feature sequence that matches the intention feature information from multiple candidate feature sequences, which is called the intention feature sequence.
[0010] In one possible implementation, the first computing unit is further configured to: In a preset number of n first-class slices, the first k first-class slices are retained, the remaining nk first-class slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple first-class splicing sequences; where k takes any value from 1 to n.
[0011] In one possible implementation, the second computing unit is further used for: In a preset number of n second-type slices, the first m second-type slices are retained, the remaining nm second-type slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple second-type splicing sequences; where m takes any value from 1 to n.
[0012] Thirdly, an electronic device is provided, including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the feature relevance analysis method in the insurance advertising placement scenario described in any of the preceding claims.
[0013] Fourthly, a storage medium is provided, wherein a computer program is stored in the storage medium, and the computer program is configured to execute the feature relevance analysis method in the insurance advertising placement scenario described above when running.
[0014] By employing the aforementioned technical solution, this application provides a feature relevance analysis method, apparatus, and related products for insurance advertising scenarios. This method constructs slice sequences of selected feature sequences under different ranking strategies and calculates their relevance scores to the original ranked sequences, thereby generating a scale code that quantifies the intrinsic structure of the feature sequences. Then, by combining the relevance coefficients of candidate feature sequences with selected feature sequences and intentional feature information, the scale code is used for filtering, enabling more accurate identification of feature combinations matching specific intentional information. Compared to methods relying solely on a single relevance coefficient, this solution, based on the scale code, can uncover deeper feature interaction relationships, effectively improving the accuracy and robustness of feature selection in complex insurance advertising scenarios, thereby optimizing advertising strategies. Attached Figure Description
[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.
[0016] Figure 1 A flowchart of the feature relevance analysis method in the insurance advertising placement scenario provided in the embodiments of this application is shown; Figure 2This paper shows a structural diagram of the feature relevance analysis device in the insurance advertising scenario provided in an embodiment of this application; Figure 3 A structural diagram of an electronic device provided in an embodiment of this application is shown. Detailed Implementation
[0017] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.
[0018] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such use can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the term "comprising" and its variations should be interpreted as open-ended terms meaning "including but not limited to."
[0019] To address the aforementioned technical problems, embodiments of this application provide a feature relevance analysis method for insurance advertising scenarios, such as... Figure 1 As shown, the feature relevance analysis method in this insurance advertising placement scenario may include the following steps S101 to S106: Step S101: Obtain insurance advertising data and determine the target feature sequence of the insurance advertising data as the selected feature sequence.
[0020] Insurance advertising data includes user information and advertising behavior records that have been authorized for public release. The target feature sequence is a pre-determined set of features based on specific business needs (e.g., a high-conversion user group). These features are considered highly relevant to the current analysis objective and are therefore used as the benchmark for subsequent analysis.
[0021] Step S102: Sort the selected feature sequence according to the first sorting strategy to obtain the first sorted sequence corresponding to the selected feature sequence; sort the selected feature sequence according to the second sorting strategy to obtain the second sorted sequence corresponding to the selected feature sequence.
[0022] To observe the structure of the selected feature sequence from different perspectives, two different sorting strategies are employed to rearrange it. The first and second sorting strategies can be based on feature value magnitude, feature importance, random sorting, or other user-defined rules. The resulting first and second sorted sequences reflect the arrangement structure of the feature sequence from different viewpoints.
[0023] Step S103: Divide the first sorted sequence into equal parts according to a preset number to obtain a preset number of first-class slices; based on the preset number of first-class slices, determine multiple first-class splicing sequences; calculate the correlation between each first-class splicing sequence and the first sorted sequence to obtain the correlation score between each first-class splicing sequence and the first sorted sequence.
[0024] The first sorted sequence is uniformly divided into a predetermined number of n consecutive segments, each segment being called a first-type slice. The value of n can be set according to actual needs, and this embodiment does not impose any restrictions on it. To explore the impact of the local structure of the sequence on the overall correlation, multiple first-type spliced sequences are constructed based on these first-type slices. Each first-type spliced sequence is formed by retaining a portion of the ordered slices and randomly shuffling the order of the remaining slices before splicing them together. Then, the correlation score between each first-type spliced sequence and the original first sorted sequence is calculated. The higher the correlation score, the more original order information the spliced sequence retains.
[0025] Step S104: Divide the second sorting sequence into equal parts according to a preset number to obtain a preset number of second-type slices; based on the preset number of second-type slices, determine multiple second-type splicing sequences; calculate the correlation between each second-type splicing sequence and the first sorting sequence to obtain the correlation score between each second-type splicing sequence and the first sorting sequence.
[0026] Similarly, the second sorted sequence is also uniformly divided into a predetermined number of n second-type slices. Multiple second-type spliced sequences are then constructed based on these slices. Each second-type spliced sequence is obtained by retaining a portion of the ordered slices and randomly shuffling the remaining slices. Then, the correlation score between each second-type spliced sequence and the original first sorted sequence is calculated. This step aims to examine the correlation impact of local structural changes in the sequence relative to the baseline of the first sorted sequence under the second sorting strategy.
[0027] Step S105: Based on the correlation scores of each first-class spliced sequence and the first sorted sequence, and the correlation scores of each second-class spliced sequence and the first sorted sequence, determine the tick code of the selected feature sequence.
[0028] All the correlation scores obtained in the above steps are organized and mapped to form a tick code for the selected feature sequence. This tick code can be regarded as a fingerprint of the feature sequence, quantifying its stability and structural characteristics under different orderings and local perturbations.
[0029] Step S106: For multiple candidate feature sequences of insurance advertising data, calculate the correlation coefficient between each candidate feature sequence and the selected feature sequence; based on the scale code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intention feature information of the selected feature sequence, select the feature sequence that matches the intention feature information from multiple candidate feature sequences, which is called the intention feature sequence.
[0030] Finally, for other candidate feature sequences to be evaluated, their conventional correlation coefficients with the selected feature sequences are first calculated. Then, a comprehensive evaluation and screening are performed by combining the scale code of the selected feature sequences (as a reference benchmark), the intended feature information of the selected feature sequences (i.e., the target feature patterns to be matched), and the calculated correlation coefficients. The final selected intended feature sequences are feature combinations that highly match the intended feature information in both structure and content.
[0031] This application embodiment constructs sliced and spliced sequences of feature sequences under different sorting strategies and calculates their relevance scores to generate a scale code that can characterize the internal structure of the feature sequence. Thus, when screening features, it not only considers the single relevance coefficient but also refers to the structured information based on the scale code, which can more accurately identify feature sequences that match specific intention feature information from candidate features, thereby improving the accuracy and robustness of feature screening.
[0032] This application embodiment provides a possible implementation method in which step S103, based on a preset number of first-type slices, determines multiple first-type splicing sequences, which may specifically include: In a preset number of n first-class slices, the first k first-class slices are retained, the remaining nk first-class slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple first-class splicing sequences; where k takes any value from 1 to n.
[0033] For example, when n=5, we can keep the first 1 slice and shuffle the last 4; keep the first 2 slices and shuffle the last 3; and so on, until we keep the first 5 slices (i.e., keep all and shuffle 0). By iterating through the values of k, we can obtain n different first-type splicing sequences, each of which retains different degrees of the original order information.
[0034] This application embodiment provides a possible implementation method in which step S104, based on a preset number of second-type slices, determines multiple second-type splicing sequences, which may specifically include: In a preset number of n second-type slices, the first m second-type slices are retained, the remaining nm second-type slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple second-type splicing sequences; where m takes any value from 1 to n.
[0035] Consistent with the construction logic of the first type of spliced sequences, n different second-type spliced sequences can be obtained by changing the number of retained slices m. This ensures that the evaluation of local structural perturbations of the sequence is symmetric and comparable under the two sorting strategies.
[0036] This application provides a possible implementation method in which the tick code of the selected feature sequence can include multiple tick values. Each tick value corresponds to the overlap ratio and correlation score of a first type of spliced sequence and a first sorted sequence. The positions of the multiple tick values are sorted in ascending order of correlation score.
[0037] The overlapping slice percentage here refers to the proportion of slices retained when constructing the spliced sequence out of the total number of slices, such as k / n or m / n. The tick code can be a list or array, where each element is a tuple (overlapping slice percentage, relevance score), and these tuples are sorted according to their relevance score. For example, the tick value with the highest relevance score corresponds to the spliced sequence that retains the most original order information, while the tick value with the lowest relevance score corresponds to the spliced sequence whose order is most severely disrupted. This tick code format provides a clear quantitative benchmark for subsequent filtering.
[0038] This application embodiment provides a possible implementation method. In step S106, based on the tick code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intention feature information of the selected feature sequence, feature sequences matching the intention feature information are selected from multiple candidate feature sequences. Specifically, this may include: First, based on the correlation coefficients between each candidate feature sequence and the selected feature sequence, the scale values corresponding to the correlation coefficients of each candidate feature sequence and the selected feature sequence in the scale code of the selected feature sequence are matched. For example, the correlation coefficients of the candidate feature sequences can be compared with the correlation scores corresponding to each scale value in the scale code to find the closest score, thereby determining the structural information such as the "overlapping slice ratio" corresponding to the candidate feature sequence.
[0039] Then, based on the correlation coefficients between the matched candidate feature sequences and the selected feature sequences at the scale values corresponding to the scale codes of the selected feature sequences, and the intentional feature information of the selected feature sequences, feature sequences that match the intentional feature information are selected from multiple candidate feature sequences. For example, if the intentional feature information requires the feature sequences to have high stability, then candidate feature sequences with high correlation coefficients and high matching scale values (i.e., the proportion of overlapping slices) can be selected.
[0040] The above introduces Figure 1 The embodiments shown have various implementation methods for each stage. The following will further illustrate the feature correlation analysis method in the insurance advertising placement scenario of this application through a specific application scenario.
[0041] Suppose that in insurance advertising, a feature sequence A, such as annual income range, has been selected that is highly correlated with "high-net-worth life insurance potential users." The next step is to find other feature sequences B from the candidate feature pool that are structurally similar to A and can supplement and enhance its feature expression, such as: {recent life insurance browsing frequency, education level, investment preferences, asset value range, etc.}.
[0042] 1. Preparation stage: The feature sequence A is sorted in ascending order according to the size of the feature values to obtain the first sorted sequence A1.
[0043] The feature sequence A is sorted in descending order according to the size of the feature value to obtain the second sorted sequence A2.
[0044] Set the preset quantity n=5, and divide A1 and A2 into 5 equal slices respectively.
[0045] 2. Calculate the scale marks: For A1, construct five first-class concatenated sequences (k=1 to 5), and calculate their relevance scores to A1. For example, we might get: k=5, all results are retained, correlation score 0.99; k=4, keep the first 4, correlation score 0.85; k=3, keep the top 3, correlation score 0.60; k=2, keep the top 2, correlation score 0.30; k=1, keep the first 1, correlation score 0.10.
[0046] For A2, five second-class spliced sequences (m=1 to 5) are constructed, and their correlation scores with A1 are calculated.
[0047] The two sets of scores are merged and processed to form the tick code of the feature sequence A. This tick code reveals the internal structure of A: only by retaining most of the original order can the correlation be high, indicating that the arrangement of its feature values has a strong regularity.
[0048] 3. Filtering intentional feature sequences: For candidate feature sequence B1 (recent browsing of life insurance products), calculate its correlation coefficient with feature sequence A, assuming it to be 0.33.
[0049] For candidate feature sequence B2 (education level), calculate its correlation coefficient with feature sequence A, assuming it to be 0.55.
[0050] For candidate feature sequence B3 (investment preference), calculate its correlation coefficient with feature sequence A, assuming it to be 0.65.
[0051] For candidate feature sequence B4 (asset value range), calculate its correlation coefficient with feature sequence A, assuming it to be 0.8.
[0052] Matching 0.33, 0.55, 0.65, and 0.8 with the scores in the scale code of A yields the corresponding scale values of 0.30, 0.60, 0.60, and 0.85, respectively.
[0053] Suppose that the intended feature information of the selected feature sequence A is "robust and highly structured" (i.e., the relevance score decreases sharply as the proportion of retained slices decreases).
[0054] The analysis unit will make a comprehensive judgment. The state matched by B4 on the scale code (k=4) indicates that the association with A depends on most of the structural information of A; and will then select the intentional feature sequence of B4.
[0055] It should be noted that the sequence numbers of the steps in the above embodiments do not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. In practical applications, all the above possible implementation methods can be arbitrarily combined in a combined manner to form possible embodiments of this application, which will not be described in detail here.
[0056] Based on the feature correlation analysis method for insurance advertising placement scenarios provided in the above embodiments, and based on the same inventive concept, this application also provides a feature correlation analysis device for insurance advertising placement scenarios.
[0057] Figure 2 This is a structural diagram of the feature relevance analysis device in the insurance advertising placement scenario provided in this application embodiment. For example... Figure 2As shown, the feature relevance analysis device in the insurance advertising placement scenario may specifically include a first determining unit 210, a sorting unit 220, a first calculation unit 230, a second calculation unit 240, a second determining unit 250, and an analysis unit 260.
[0058] The first determining unit 210 is used to acquire insurance advertising placement data and determine the target feature sequence of the insurance advertising placement data as the selected feature sequence. The sorting unit 220 is used to sort the selected feature sequence according to a first sorting strategy to obtain a first sorted sequence corresponding to the selected feature sequence; and to sort the selected feature sequence according to a second sorting strategy to obtain a second sorted sequence corresponding to the selected feature sequence. The first calculation unit 230 is used to divide the first sorting sequence into equal parts according to a preset number to obtain a preset number of first-type slices; based on the preset number of first-type slices, determine multiple first-type splicing sequences; calculate the correlation between each first-type splicing sequence and the first sorting sequence to obtain the correlation score between each first-type splicing sequence and the first sorting sequence. The second calculation unit 240 is used to divide the second sorting sequence into equal parts according to a preset number to obtain a preset number of second-type slices; based on the preset number of second-type slices, determine multiple second-type splicing sequences; calculate the correlation between each second-type splicing sequence and the first sorting sequence to obtain the correlation score between each second-type splicing sequence and the first sorting sequence. The second determining unit 250 is used to determine the scale code of the selected feature sequence based on the correlation scores of each first type of spliced sequence and the first sorted sequence, and the correlation scores of each second type of spliced sequence and the first sorted sequence. Analysis unit 260 is used to calculate the correlation coefficient between each candidate feature sequence and the selected feature sequence for multiple candidate feature sequences of insurance advertising data; based on the scale code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intention feature information of the selected feature sequence, it selects the feature sequence that matches the intention feature information from multiple candidate feature sequences, which is called the intention feature sequence.
[0059] This application embodiment provides a possible implementation, wherein the first computing unit 230 is further configured to: In a preset number of n first-class slices, the first k first-class slices are retained, the remaining nk first-class slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple first-class splicing sequences; where k takes any value from 1 to n.
[0060] This application embodiment provides a possible implementation, wherein the second computing unit 240 is further configured to: In a preset number of n second-type slices, the first m second-type slices are retained, the remaining nm second-type slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple second-type splicing sequences; where m takes any value from 1 to n.
[0061] This application provides a possible implementation method in which the tick code of the selected feature sequence includes multiple tick values. Each tick value corresponds to the overlap ratio and correlation score of a first type of spliced sequence and a first sorted sequence. The positions of the multiple tick values are sorted in ascending order of correlation score.
[0062] This application embodiment provides a possible implementation, wherein the analysis unit 260 is further configured to: Based on the correlation coefficient between each candidate feature sequence and the selected feature sequence, match the scale value corresponding to the scale code of the selected feature sequence for the correlation coefficient between each candidate feature sequence and the selected feature sequence. Based on the correlation coefficient between each candidate feature sequence and the selected feature sequence, the scale value corresponding to the scale code of the selected feature sequence, and the intentional feature information of the selected feature sequence, feature sequences that match the intentional feature information are selected from multiple candidate feature sequences.
[0063] Based on the same inventive concept, this application also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the feature relevance analysis method in the insurance advertising placement scenario of any of the above embodiments.
[0064] In an exemplary embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3 The illustrated electronic device 300 includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may also include a transceiver 304. It should be noted that in practical applications, the transceiver 304 is not limited to one type, and the structure of this electronic device 300 does not constitute a limitation on the embodiments of this application.
[0065] Processor 301 may be a CPU (Central Processing Unit), GPU (Graphics Processing Unit), DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.
[0066] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.
[0067] The memory 303 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.
[0068] The memory 303 stores computer program code that executes the scheme of this application, and its execution is controlled by the processor 301. The processor 301 executes the computer program code stored in the memory 303 to implement the content shown in the foregoing method embodiments.
[0069] Among them, electronic devices include, but are not limited to: mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 3 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.
[0070] Based on the same inventive concept, this application also provides a storage medium storing a computer program, wherein the computer program is configured to execute the feature relevance analysis method in the insurance advertising placement scenario of any of the above embodiments when running.
[0071] Those skilled in the art will clearly understand that the specific working process of the systems, devices, and modules described above can be referred to the corresponding process in the foregoing method embodiments. For the sake of brevity, it will not be repeated here.
[0072] Those skilled in the art will understand that the technical solution of this application, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several program instructions to cause an electronic device (e.g., a personal computer, server, or network device) to execute all or part of the steps of the methods described in the embodiments of this application when running the program instructions. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.
[0073] Alternatively, all or part of the steps of the foregoing method embodiments can be implemented by hardware (such as electronic devices like personal computers, servers, or network devices) associated with program instructions. The program instructions can be stored in a computer-readable storage medium. When the program instructions are executed by the processor of the electronic device, the electronic device executes all or part of the steps of the methods described in the embodiments of this application.
[0074] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that within the spirit and principles of this application, modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the corresponding technical solutions to leave the protection scope of this application.
Claims
1. A method for feature correlation analysis in insurance advertising scenarios, characterized in that, The method includes: Acquire insurance advertising placement data and determine the target feature sequence of the insurance advertising placement data as the selected feature sequence; The selected feature sequences are sorted according to the first sorting strategy to obtain the first sorted sequence corresponding to the selected feature sequences; the selected feature sequences are sorted according to the second sorting strategy to obtain the second sorted sequence corresponding to the selected feature sequences. The first sorted sequence is divided into equal parts according to a preset number to obtain a preset number of first-class slices; based on the preset number of first-class slices, multiple first-class splicing sequences are determined; the correlation between each first-class splicing sequence and the first sorted sequence is calculated to obtain the correlation score between each first-class splicing sequence and the first sorted sequence. The second sorting sequence is divided into equal parts according to a preset number to obtain a preset number of second-type slices; based on the preset number of second-type slices, multiple second-type splicing sequences are determined; the correlation between each second-type splicing sequence and the first sorting sequence is calculated to obtain the correlation score between each second-type splicing sequence and the first sorting sequence. Based on the correlation scores between each first-class spliced sequence and the first sorted sequence, and the correlation scores between each second-class spliced sequence and the first sorted sequence, the tick code of the selected feature sequence is determined. For multiple candidate feature sequences of insurance advertising data, calculate the correlation coefficient between each candidate feature sequence and the selected feature sequence. Based on the tick code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intention feature information of the selected feature sequence, the feature sequence that matches the intention feature information is selected from multiple candidate feature sequences, and is called the intention feature sequence.
2. The method according to claim 1, characterized in that, Based on a preset number of first-type slices, multiple first-type splicing sequences are determined, including: In a preset number of n first-class slices, the first k first-class slices are retained, the remaining nk first-class slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple first-class splicing sequences; where k takes any value from 1 to n.
3. The method according to claim 1, characterized in that, Based on a preset number of second-type slices, multiple second-type splicing sequences are determined, including: In a preset number of n second-type slices, the first m second-type slices are retained, the remaining nm second-type slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple second-type splicing sequences; where m takes any value from 1 to n.
4. The method according to claim 1, characterized in that, The selected feature sequence's tick code includes multiple tick values. Each tick value corresponds to the percentage of overlapping slices and the correlation score of a first-class spliced sequence and a first-sorted sequence. The positions of the multiple tick values are sorted in ascending order of their correlation scores.
5. The method according to claim 4, characterized in that, Based on the tick codes of the selected feature sequences, the correlation coefficients between each candidate feature sequence and the selected feature sequence, and the intentional feature information of the selected feature sequences, feature sequences that match the intentional feature information are selected from multiple candidate feature sequences, including: Based on the correlation coefficient between each candidate feature sequence and the selected feature sequence, match the scale value corresponding to the scale code of the selected feature sequence for the correlation coefficient between each candidate feature sequence and the selected feature sequence. Based on the correlation coefficient between each candidate feature sequence and the selected feature sequence, the scale value corresponding to the scale code of the selected feature sequence, and the intentional feature information of the selected feature sequence, feature sequences that match the intentional feature information are selected from multiple candidate feature sequences.
6. A feature relevance analysis device for insurance advertising placement scenarios, characterized in that, The device includes: The first determining unit is used to acquire insurance advertising placement data and determine the target feature sequence of the insurance advertising placement data as the selected feature sequence. The sorting unit is used to sort the selected feature sequence according to a first sorting strategy to obtain a first sorted sequence corresponding to the selected feature sequence; and to sort the selected feature sequence according to a second sorting strategy to obtain a second sorted sequence corresponding to the selected feature sequence. The first calculation unit is used to divide the first sorted sequence into equal parts according to a preset number to obtain a preset number of first-type slices; based on the preset number of first-type slices, determine multiple first-type splicing sequences; calculate the correlation between each first-type splicing sequence and the first sorted sequence to obtain the correlation score between each first-type splicing sequence and the first sorted sequence. The second calculation unit is used to divide the second sorting sequence into equal parts according to a preset number to obtain a preset number of second-type slices; based on the preset number of second-type slices, determine multiple second-type splicing sequences; calculate the correlation between each second-type splicing sequence and the first sorting sequence to obtain the correlation score between each second-type splicing sequence and the first sorting sequence. The second determining unit is used to determine the tick code of the selected feature sequence based on the correlation scores of each first type of spliced sequence and the first sorted sequence, and the correlation scores of each second type of spliced sequence and the first sorted sequence. The analysis unit is used to calculate the correlation coefficient between each candidate feature sequence and the selected feature sequence for multiple candidate feature sequences of insurance advertising data; based on the scale code of the selected feature sequence, the correlation coefficient between each candidate feature sequence and the selected feature sequence, and the intention feature information of the selected feature sequence, the unit selects the feature sequence that matches the intention feature information from multiple candidate feature sequences, which is called the intention feature sequence.
7. The apparatus according to claim 6, characterized in that, The first computing unit is also used for: In a preset number of n first-class slices, the first k first-class slices are retained, the remaining nk first-class slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple first-class splicing sequences; where k takes any value from 1 to n.
8. The apparatus according to claim 6, characterized in that, The second computing unit is also used for: In a preset number of n second-type slices, the first m second-type slices are retained, the remaining nm second-type slices are randomly shuffled, and the retained slices and the randomly shuffled slices are spliced together to obtain multiple second-type splicing sequences; where m takes any value from 1 to n.
9. An electronic device, characterized in that, The device includes a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the feature relevance analysis method in the insurance advertising placement scenario according to any one of claims 1 to 5.
10. A storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the feature relevance analysis method in the insurance advertising placement scenario according to any one of claims 1 to 5 when it runs.