User viscosity scoring method and device, vehicle and storage medium

Abstract

The embodiment of the invention provides a user viscosity scoring method and device, a vehicle and a storage medium, and relates to the technical field of vehicle software. A subjective judgment matrix corresponding to a to-be-scored user type is obtained, the subjective judgment matrix comprises influence factors of multiple dimensions on the user type and influence factors among the multiple dimensions, and the multiple dimensions are related to user stickiness of vehicle-mounted terminal software; according to the subjective judgment matrix, determining respective weights of a plurality of dimensions corresponding to the user type by adopting a graph convolutional network; and determining the user viscosity score of the user type according to the weights of the multiple dimensions, thereby solving the problem that the current vehicle machine user viscosity scoring method lacks quantitative specifications and objectivity, resulting in inaccurate vehicle machine user viscosity score.

Description

Technical Field

[0001] The present application relates to the technical field of in-vehicle software, and more specifically, to a method and device for user stickiness scoring, a vehicle, and a storage medium. Background Art

[0002] In the operation of in-vehicle software products, the in-vehicle user stickiness score refers to the degree of dependence and the degree of re-consumption expectation formed by users for in-vehicle software. The higher the score, the higher the degree of dependence and the degree of re-consumption expectation of users. The in-vehicle user stickiness score can be used to find high-quality users, enable users to voice their opinions for the iteration of in-vehicle software, put forward suggestions, ensure that the in-vehicle software meets the effective needs of users during iteration, and increase user stickiness.

[0003] Currently, there is no standard for quantifying the in-vehicle user stickiness score in the field of in-vehicle software product operation. Related technologies analyze the driving behavior characteristics of drivers, screen the characteristics and form a driving behavior feature matrix, construct a driver portrait according to the similarity edge connection rule and use a clustering algorithm, and further use a radar chart to display the driving behavior of drivers. The clustering algorithm is an unsupervised category learning process, so it is necessary to make manual judgments according to the eigenvalue of the actual clustering situation, which has subjectivity to a certain extent and lacks objectivity. In addition, only using a radar chart to display the driving behaviors of drivers does not form a specific quantification specification and cannot intuitively display the distribution of various data.

[0004] That is to say, the current in-vehicle user stickiness scoring method lacks a quantification specification and objectivity, resulting in inaccurate in-vehicle user stickiness scores. Summary of the Invention

[0005] Embodiments of the present application provide a method and device for user stickiness scoring, a vehicle, and a storage medium to solve the problem that the current in-vehicle user stickiness scoring method lacks a quantification specification and objectivity, resulting in inaccurate in-vehicle user stickiness scores.

[0006] In a first aspect, embodiments of the present application provide a method for user stickiness scoring, the method including: obtaining a subjective judgment matrix corresponding to the user type to be scored, the subjective judgment matrix including influence factors of each of multiple dimensions on the user type and influence factors between multiple dimensions, the multiple dimensions being related to the user stickiness of in-vehicle software; determining weights of each of the multiple dimensions corresponding to the user type by using a graph convolutional network according to the subjective judgment matrix; and determining a user stickiness score of the user type according to the weights of the multiple dimensions.

[0007] In a second aspect, an embodiment of the present application provides a user stickiness scoring device, which includes: a matrix acquisition module, configured to acquire a subjective judgment matrix corresponding to the user type to be scored, where the subjective judgment matrix includes influence factors of each of multiple dimensions on the user type and influence factors between the multiple dimensions, and the multiple dimensions are related to the user stickiness of the in-vehicle software; a weight calculation module, configured to determine the weights of each of the multiple dimensions corresponding to the user type by using a graph convolutional network according to the subjective judgment matrix; and a user stickiness scoring module, configured to determine the user stickiness score of the user type according to the weights of the multiple dimensions.

[0008] In a third aspect, an embodiment of the present application provides a vehicle, which includes: a memory and a processor, where an application program is stored in the memory, and the application program is configured to execute the method provided by the embodiment of the present application when called by the processor.

[0009] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which program code is stored, and the program code is configured to cause the processor to execute the method provided by the embodiment of the present application when called by the processor.

[0010] The user stickiness scoring method, device, vehicle, and storage medium provided by the embodiments of the present application divide dimensions related to the user stickiness of the in-vehicle software. According to the subjective judgment matrix corresponding to the user type to be scored, the user stickiness of the user type to be scored can be subjectively evaluated. In addition, according to the subjective judgment matrix corresponding to the user type to be scored, the weights of each of the multiple dimensions corresponding to the user type to be scored are determined by using a graph convolutional network, and the user stickiness score of the user type to be scored is determined according to the weights of the multiple dimensions, so that the user stickiness of the user type to be scored can be objectively evaluated. By mainly using subjective evaluation and supplemented by objective evaluation, a comprehensive score is given to the user stickiness score of the user type to be scored, providing a quantitative specification for scoring the in-vehicle user stickiness scores of different user types, which can improve the accuracy of the in-vehicle user stickiness score, thereby solving the problem that the current in-vehicle user stickiness scoring method lacks a quantitative specification and objectivity, resulting in inaccurate in-vehicle user stickiness scores. Description of the Drawings

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application, rather than all embodiments. Based on the embodiments of the present application, all other embodiments and drawings obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0012] Figure 1It is a flowchart of a user stickiness scoring method provided by an embodiment of the present application;

[0013] Figure 2 It is a schematic diagram of multiple user types provided by an exemplary embodiment of the present application;

[0014] Figure 3 It is a schematic diagram of multiple dimensions related to vehicle owner attributes provided by an exemplary embodiment of the present application;

[0015] Figure 4 It is a schematic diagram of at least two dimensions related to vehicle owner attributes provided by an exemplary embodiment of the present application;

[0016] Figure 5 It is a structural block diagram of a user stickiness scoring device provided by an embodiment of the present application;

[0017] Figure 6 It is a structural block diagram of a vehicle provided by an embodiment of the present application. Detailed implementation manners

[0018] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application.

[0019] The user stickiness scoring method provided by the embodiments of the present application can be applied to a user stickiness scoring device or a vehicle. The vehicle can be a gasoline vehicle or an electric vehicle. Electric vehicles can include, but are not limited to, pure electric vehicles, hybrid vehicles, or fuel cell vehicles, etc., and no specific limitations are made here.

[0020] See Figure 1 , Figure 1 It is a flowchart of a user stickiness scoring method provided by an embodiment of the present application. This method can include step S110 to step S130.

[0021] Step S110: Obtain a subjective judgment matrix corresponding to the user type to be scored. The subjective judgment matrix includes the influence factors of each of multiple dimensions on the user type to be scored and the influence factors between multiple dimensions. The multiple dimensions are related to the user stickiness of the in-vehicle software.

[0022] The user type to be scored is one of multiple user types pre-manually divided. Multiple users can be divided according to actual needs. Exemplarily, see Figure 2 , Figure 2It is a schematic diagram of multiple user types provided by an exemplary embodiment of the present application. Four user types can be divided from the user value dimension according to the activity and usage intensity of users using the in-vehicle software, namely, valuable users, retained users, developing users, and recovering users. For example, vehicle owners with high activity and high usage intensity can be classified as valuable users; vehicle owners with low activity and high usage intensity can be classified as retained users; vehicle owners with high activity and medium usage intensity can be classified as developing users; vehicle owners with low activity and medium usage intensity can be classified as recovering users. That is to say, the user type to be scored in the embodiments of the present application can be one of valuable users, retained users, developing users, and recovering users.

[0023] Multiple dimensions related to the user stickiness of the in-vehicle software can be pre-divided according to actual needs and stored in a specified location, and multiple dimensions related to the user stickiness of the in-vehicle software can be directly obtained from the specified location when needed. Among them, the specified location can include but is not limited to a storage area in the vehicle or a storage area in the cloud server.

[0024] In some embodiments, in order to conduct targeted and fine-grained division of the portrait feature dimensions of vehicle owners (such as drivers) during vehicle use, the dimensions related to the user stickiness of the in-vehicle software can be further divided into finer-grained dimensions from two basic dimensions: the vehicle owner's basic attribute dimension and the vehicle use dimension. For example, multiple dimensions related to the user stickiness of the in-vehicle software can be divided into multiple dimensions related to the vehicle owner's attributes and at least two dimensions related to vehicle use.

[0025] For example, see Figure 3 , Figure 3 It is a schematic diagram of multiple dimensions related to vehicle owner attributes provided by an exemplary embodiment of the present application. Based on the vehicle owner's basic attribute dimension, the vehicle owner's basic attribute dimension can be divided into eight finer-grained dimensions: the gender of the vehicle owner, the age of the vehicle owner, the permanent residence city, the marital and childbearing status of the vehicle owner, the income situation of the vehicle owner, the education level of the vehicle owner, the work industry of the vehicle owner, and the frequency of vehicle use by the vehicle owner. Each of the eight dimensions includes at least two labels.

[0026] See Figure 3 ,the gender dimension of the vehicle owner is divided into two labels, namely male and female.

[0027] See Figure 3 ,according to the age distribution range of the vehicle owner, the age dimension of the vehicle owner can be divided into four labels, namely, young, middle-aged, middle-aged and elderly, and elderly. For example, vehicle owners under 35 years old can be classified as young; vehicle owners aged 35-50 can be classified as middle-aged; vehicle owners aged 50-65 can be classified as middle-aged and elderly; vehicle owners over 65 years old can be classified as elderly.

[0028] See Figure 3 , according to the urban areas covered by the owner's activity range, the main activity city is determined as the owner's permanent residence city. The dimension of the owner's permanent residence city can be divided into four labels, namely, first-tier cities, new first-tier cities, second-tier cities, and third-tier cities.

[0029] See Figure 3 , the dimension of the owner's marital and childbearing status is divided into three labels, namely, unmarried, married but childless, and married with children.

[0030] See Figure 3 , the dimension of the owner's income situation is divided into three labels, namely, low income, medium income, and high income. For example, the income of a car owner with a monthly income of less than 5,000 can be classified as low income; the income of a car owner with a monthly income of 5,000 - 20,000 can be classified as medium income; the income of a car owner with a monthly income of more than 20,000 can be classified as high income.

[0031] See Figure 3 , the dimension of the owner's education level is divided into four labels, namely, high school and below, junior college, undergraduate, and master and above.

[0032] See Figure 3 , the dimension of the owner's working industry is divided into five labels, namely, high-tech, government, finance, trade, real estate, commercial service industry, and others.

[0033] See Figure 3 , the dimension of the owner's frequency of using the car is divided into four labels, namely, operation, commuting, daily, and rarely.

[0034] For example, see Figure 4 , Figure 4 is a schematic diagram of at least two dimensions related to the owner's attributes provided by an exemplary embodiment of the present application. Based on the dimension of in-vehicle infotainment system usage, the dimension of in-vehicle infotainment system usage can be divided into two finer-grained dimensions: activity and usage intensity. Both the activity dimension and the usage intensity dimension include multiple labels.

[0035] See Figure 4 , the activity dimension can be divided into three labels according to the user's activity rate of using in-vehicle infotainment system software and the interval since the last activity, namely, high activity, low activity, and inactive, with the level decreasing in turn. For example, the activity with an activity rate greater than a certain percentage (set according to actual needs) and an interval since the last activity less than a certain threshold number of days (set according to actual needs) can be classified as high activity; the activity with an activity rate less than a certain percentage or an interval since the last activity greater than a certain threshold number of days can be classified as low activity; the activity with an activity rate less than a certain percentage and an interval since the last activity greater than a certain threshold number of days can be classified as inactive.

[0036] See Figure 4 , according to the usage duration and click times of the user using the in-vehicle software, the usage intensity dimension can be divided into three labels with decreasing levels: high usage intensity, medium usage intensity, and low usage intensity. For example, the usage intensity with a usage duration greater than a certain percentile (set according to actual needs) and a click times greater than a certain percentile (set according to actual needs) can be classified as high usage intensity; the usage intensity with a usage duration less than a certain percentile and a click times less than a certain percentile can be classified as low usage intensity; and the usage intensity in the remaining other cases can be classified as medium usage intensity.

[0037] To distinguish multiple dimensions related to the user stickiness of the in-vehicle software, multiple dimensions related to the owner attributes, and at least two dimensions related to the in-vehicle usage, in the following text, multiple dimensions related to the user stickiness of the in-vehicle software are referred to as X dimensions, multiple dimensions related to the owner attributes are referred to as Y dimensions, and at least two dimensions related to the in-vehicle usage are referred to as Z dimensions, where X = Y + Z, Y is a positive integer, and Z is a positive integer greater than 1.

[0038] After dividing multiple user types and the X dimensions related to the user stickiness of the in-vehicle software, based on the comprehensive big data statistical rules and expert knowledge, the influence factors of the X dimensions on each type of user and the influence factors between the X dimensions can be determined in advance, forming a subjective judgment matrix corresponding to each type of user. The subjective judgment matrix is stored in the above storage area so that the subjective judgment matrix corresponding to each type of user can be directly obtained from the storage area when needed. The influence factor of each dimension in the X dimensions on each type of user can be understood as the influence weight of each dimension on each type of user. The greater the influence factor, the higher the influence degree of each dimension on each type of user. The influence factor between the X dimensions can be understood as the interaction between the two dimensions corresponding to the influence factor. The greater the influence factor, the higher the degree of mutual influence between the two dimensions corresponding to the influence factor.

[0039] The subjective judgment matrix corresponding to each type of user can be stored in tabular form. The first row and the first column of the table of the subjective judgment matrix corresponding to each type of user are respectively the X dimensions. The values on the diagonal between the first row and the first column to the last row and the last column in the table represent the influence factors of the X dimensions on this type of user, and the other values outside the diagonal represent the influence factors between the X dimensions.

[0040] Table 1

[0041]

[0042] For example, taking the subjective judgment matrix corresponding to value users as an example, see Table 1, the value a in Table 1 11 -a 1010is the influence factor. Specifically, the values a on the diagonal between the first row and the first column and the last row and the last column in Table 1 11 , a 22 , a 33 , a 44 , a 55 , a 66 , a 77 , a 88 , a 99 , a 1010 respectively represent the influence factors of each dimension in the X dimensions on the valuable users. For example, a 11 represents the influence factor of the gender dimension on the valuable users. The other values outside the diagonal represent the influence factors between the X dimensions. For example, a 12 represents the influence factor between the age dimension and the gender dimension.

[0043] After obtaining the subjective judgment matrix corresponding to the user type to be scored (such as valuable users), the consistency verification can be performed on each influence factor in the subjective judgment matrix. After passing the consistency verification, based on the subjective judgment matrix corresponding to the user type to be scored (such as valuable users), determine the weights of each of the multiple dimensions corresponding to the user type to be scored (such as valuable users), that is, execute step S120, so as to ensure the accuracy of the weights and the accuracy of the user stickiness scoring result.

[0044] Step S120: According to the subjective judgment matrix corresponding to the user type to be scored, use the graph convolutional network to determine the weights of each of the multiple dimensions related to the user stickiness of the user type to be scored for the in-vehicle software.

[0045] Since the degree matrix and the adjacency matrix will be used in the graph convolutional network, after obtaining the subjective judgment matrix of the user type to be scored, the degree matrix and the adjacency matrix of the graph convolutional network can be constructed first according to the subjective judgment matrix corresponding to the user type to be scored, and then according to the degree matrix and the adjacency matrix, through two-layer graph convolutional network, determine the weights of the X dimensions corresponding to the user type to be scored.

[0046] The arithmetic mean method in the analytic hierarchy process can be used to determine the initial weights of each of the X dimensions corresponding to the user type to be scored according to the influence factors in the subjective judgment matrix corresponding to the user type to be scored; construct the degree matrix of the graph convolutional network according to the initial weights of the X dimensions corresponding to the user type to be scored.

[0047] Specifically, calculating the initial weight of each of the X dimensions corresponding to the user type to be scored may include the following steps: Divide each influencing factor by the sum of the column where the influencing factor is located to normalize the influencing factor; Sum each normalized influencing factor by row; Divide the sum result by the number of dimensions to obtain the initial weight of each dimension.

[0048] Exemplarily, assume that the subjective judgment matrix M corresponding to the user type to be scored is:

[0049]

[0050] Then, the arithmetic mean method shown in the following expression can be used to calculate the initial weight of each of the X dimensions corresponding to the user type to be scored:

[0051]

[0052] where ω i represents the initial weight of dimension i corresponding to the user type to be scored; n represents the number of dimensions X of the X dimensions corresponding to the user type to be scored. For example, if the number of dimensions X of the X dimensions corresponding to the user type to be scored is X = 10, then n is 10; a ij represents the influencing factor between two dimensions i and j in the subjective judgment matrix.

[0053] After obtaining the initial weights of each of the X dimensions corresponding to the user type to be scored, the initial weights of the X dimensions corresponding to the user type to be scored can be determined as the degree matrix D of the graph convolutional network corresponding to the user type to be scored. The matrix dimension of each degree matrix D is 10 * 1.

[0054] The labels of two dimensions in the X dimensions can be combined to obtain multiple label combinations corresponding to the two dimensions. Each label combination includes two labels, and the two labels come from different dimensions among the X dimensions. Determine the first (label) quantity of each label combination in the labels corresponding to the user type to be scored; Determine the second (label) quantity of each label combination in the labels corresponding to all user types, where the user type to be scored is one of all user types; Determine the ratio of the first quantity to the second quantity corresponding to each label combination; Determine the sum of the ratios of the multiple label combinations as the target ratio of the two dimensions; Construct the adjacency matrix of the graph convolutional network according to the target ratio of the two dimensions.

[0055] Table 2

[0056]

[0057] Exemplarily, please refer to Figure 2 and Figure 3, assume there are four types of users, namely, valuable users, retained users, developing users, and win-back users. Assume the type of the user to be scored is a valuable user, and taking the gender dimension and age dimension in valuable users as examples, combine the labels "male" and "female" in the gender dimension and the labels "youth", "middle-aged", "middle and old-aged", and "old-aged" in the age dimension pairwise. As shown in Table 2, 8 label combinations corresponding to the gender dimension and age dimension can be obtained, as well as the first quantity in the labels corresponding to valuable users and the second quantity in the labels corresponding to the four types of users.

[0058] According to Table 2, the ratio of the number of labels of each label combination in valuable users to the number of labels of this label combination in the four types of users can be calculated. Then, add up the ratios corresponding to all label combinations to obtain the target ratio of the gender dimension and age dimension corresponding to valuable users. For example, the target ratio of the gender dimension and age dimension corresponding to valuable users can be expressed as follows:

[0059]

[0060] The pairwise dimensions with a target ratio greater than the ratio threshold can be marked as 1, and the pairwise dimensions with a target ratio less than the ratio threshold can be marked as 0 to construct the adjacency matrix of the graph convolutional network. Among them, the ratio threshold can be set according to actual needs and is not limited here. For example, the adjacency matrix A can be expressed as follows:

[0061]

[0062] After obtaining the degree matrix and adjacency matrix corresponding to the type of user to be scored, the weights of each of the X dimensions corresponding to the type of user to be scored can be determined through a two-layer graph convolutional network according to the degree matrix and adjacency matrix. Among them, when the number of layers of the graph convolutional network is 2 or 3, the effect is better.

[0063] For example, the weights of the X dimensions corresponding to the type of user to be scored can be expressed by the following expression:

[0064]

[0065] Among them, H l+1 represents the weights of each of the X dimensions corresponding to the type of user to be scored after being updated by the graph convolutional network. The dimension is four groups of 10*1, and each group of 10*1 corresponds to the weights of the X dimensions corresponding to the type of user to be scored; the softmax function is a normalized exponential function, which is used to display the results of multi-classification in the form of probabilities; A is the adjacency matrix; I is the identity matrix; is 's degree matrix; H lThe weights of X dimensions of the user type to be scored for the current layer, W l is the weight matrix to be learned.

[0066] In the embodiments of the present application, first, based on the subjective judgment matrix, the preliminary weights of the X dimensions corresponding to the user type to be scored are determined, and then, based on the objective target ratio of the pairwise dimensions of the user type to be scored, the weights of each of the X dimensions corresponding to the user type to be scored are further updated, so as to realize fine-tuning of the weights of the X dimensions corresponding to the user type to be scored, obtain a weight result combining subjective and objective evaluations, and improve the accuracy of determining the user stickiness score in the subsequent process.

[0067] Step S130: Determine the user stickiness score of the user type to be scored according to the weights of multiple dimensions related to the user stickiness of the in-vehicle software.

[0068] The basic score of each dimension in the X dimensions corresponding to the user type to be scored can be determined according to the number of dimensions of the X dimensions corresponding to the user type to be scored. As mentioned above, the X dimensions include Y dimensions related to the basic attributes of the vehicle owner and Z dimensions related to the in-vehicle software usage. The basic score of each of the Y dimensions corresponding to the user type to be scored can be calculated according to the number of dimensions Y of the Y dimensions. The basic score of each of the Z dimensions corresponding to the user type to be scored can be calculated according to the number of dimensions Z of the Z dimensions.

[0069] In some embodiments, the preset score (for example, 100 points) can be divided by the number of dimensions Y of the Y dimensions to obtain the basic score of each dimension in the Y dimensions. By way of example, let the preset score be 100, then the basic score of each dimension in the Y dimensions can be expressed by the following expression:

[0070]

[0071] where Score driver represents the basic score of each dimension in the Y dimensions related to the basic attributes of the vehicle owner; Y represents the number of dimensions of the Y dimensions related to the basic attributes of the vehicle owner.

[0072] For example, referring to Figure 3 , assuming that the dimensions related to the vehicle owner attributes include eight dimensions: gender dimension, age dimension, permanent residence area dimension, marital and childbearing status dimension, income situation dimension, education level dimension, work industry dimension, and vehicle usage frequency dimension (Y = 8), then the basic score of each dimension related to the vehicle owner attributes is 100 / 8 = 12.5.

[0073] In some embodiments, each of the Z dimensions related to in-vehicle infotainment system usage includes multiple different levels of tags. One tag corresponding to the user type to be scored can be obtained from each of the Z dimensions related to in-vehicle infotainment system usage as the target tag, resulting in Z target tags. Based on the levels of the Z target tags and the number of dimensions of the Z dimensions related to in-vehicle infotainment system usage, the basic score for each dimension related to in-vehicle infotainment system usage corresponding to the user type to be scored is determined.

[0074] Among them, the correspondence between the target tag and the user type to be scored (for example, a user corresponds to a low activity tag and a high usage intensity tag) is preset and stored in the above-mentioned specified area, and can be directly obtained from the specified area when needed. For example, Figure 2 and Figure 4 As shown, the Z dimensions related to in-vehicle infotainment system usage include activity and usage intensity, and there are four types of user types. Among them, the two target tags corresponding to valuable users are high activity tags and high usage intensity tags; the two target tags corresponding to retained users are low activity tags and high usage intensity tags; the two target tags corresponding to developing users are high activity tags and medium usage intensity tags; the two target tags corresponding to win-back users are low activity tags and medium usage intensity tags.

[0075] For example, assume that the two dimensions related to in-vehicle infotainment system usage are activity and usage intensity respectively, and assume that the user type to be scored is a retained user. Refer to Figure 2 , a retained user corresponds to a low activity tag and a high usage intensity tag. Then the low activity tag and the high usage intensity tag are the target tags corresponding to the retained user. Based on the level of the low activity tag and the number of dimensions of the Z dimensions, the basic score of the activity dimension corresponding to the retained user can be determined. Based on the level of the high usage intensity tag and the number of dimensions of the Z dimensions, the basic score of the usage intensity dimension corresponding to the retained user can be determined.

[0076] In some embodiments, the preset score (for example, 100 points) can be divided by the number of dimensions Z of the Z dimensions related to in-vehicle infotainment system usage to obtain an intermediate value of 100 / Z. The level of the target tag is multiplied by the intermediate value to obtain the basic score of the dimension related to in-vehicle infotainment system usage corresponding to the target tag. For example, let the preset score be 100, then the basic score of each of the Z dimensions related to in-vehicle infotainment system usage corresponding to the user type to be scored can be expressed by the following expression:

[0077]

[0078] Among them, Score vehicleRepresents the basic score for each dimension related to in-vehicle infotainment system usage (e.g., the activity dimension) corresponding to the user type to be scored (e.g., retained user); level is the level of the target label (low activity label) corresponding to each dimension related to in-vehicle infotainment system usage (e.g., the activity dimension) for the user type to be scored. For example, as Figure 4 shown, the activity dimension includes high activity label, low activity label, and inactive label with gradually decreasing levels. The levels of the high activity label, low activity label, and inactive label are 3, 2, and 1 respectively.

[0079] For example, referring to Figure 2 , assume that the user type to be scored is a retained user, and the retained user corresponds to the low activity label (level = 2) and the high usage intensity label (level = 3). Then the basic score of the retained user corresponding to the activity dimension is 2 * 100 / 2 = 100, and the basic score of the retained user corresponding to the usage intensity dimension is 3 * 100 / 2 = 150.

[0080] The user stickiness score of the user type to be scored can be determined based on the basic score of each of the X dimensions corresponding to the user type to be scored and the weights of the X dimensions corresponding to the user type to be scored. Specifically, the product of the weight and the basic score of each of the X dimensions corresponding to the user type to be scored can be calculated; the sum of the products of the X dimensions corresponding to the user type to be scored is determined as the user stickiness score of the user type to be scored. For example, the following expression can be used to calculate the user stickiness score of the user type to be scored:

[0081]

[0082] where Score is the user stickiness score of the user type to be scored; Score driver,i represents the basic score of each dimension i (e.g., gender dimension) related to vehicle owner attributes corresponding to the user type to be scored, α i is the weight of dimension i; Score vehicle,j represents the basic score of each dimension j (e.g., activity) related to in-vehicle infotainment system usage corresponding to the user type to be scored, α j represents the weight of dimension j.

[0083] For example, referring to Figure 3 , assume that the dimensions related to vehicle owner attributes include the gender dimension (weight is α j1 ), the young dimension (weight is α j2 ), the permanent residence area dimension (weight is α j3 ), the marital and fertility status dimension (weight is α j4 ), and the income status dimension (weight is α j5) Degree dimension (weight is α j6 ) Industry dimension of work (weight is α j7 ) And vehicle usage frequency dimension (weight is α j8 ) A total of eight dimensions (Y = 8), then the basic score for each dimension related to the vehicle owner attributes is 100 / 8 = 12.5. See Figure 4 , assuming that the dimensions related to in-vehicle infotainment system usage include the activity dimension (weight is α i1 ) And usage intensity dimension (weight is α i2 ) A total of two dimensions (Z = 2). See Figure 2 , assuming that the user type to be scored is a retained user, and the retained user corresponds to a low-activity label (level = 2) and a high-activity label (level = 3), then the basic score for the activity dimension corresponding to the retained user is 2*100 / 2 = 100, and the basic score for the usage intensity dimension corresponding to the retained user is 3*100 / 2 = 150. The user stickiness score score of the retained user is:

[0084] score = 12.5*α j1 + 12.5*α j2 + 12.5*α j3 + 12.5*α j4 + 12.5*α j5 + 12.5*α j6 + 12.5*α j7 + 12.5*α j8 + 100*α i1 + 150*α i2 = 12.5*(α j1 + α j2 + α j3 + α j4 + α j5 + θ j6 + α j7 + α j8 ) + 100*α i1 + 150*α i2 .

[0085] The user stickiness scoring method provided by the embodiments of the present application divides the dimensions related to the user stickiness of in-vehicle software, and according to the subjective judgment matrix corresponding to the user type to be scored, the user stickiness of the user type to be scored can be subjectively evaluated. In addition, according to the subjective judgment matrix corresponding to the user type to be scored, a graph convolutional network is used to determine the weights of each of the multiple dimensions corresponding to the user type to be scored, and according to the weights of the multiple dimensions, the user stickiness score of the user type to be scored is determined, and the user stickiness of the user type to be scored can be objectively evaluated. By mainly using subjective evaluation and supplemented by objective evaluation, a comprehensive score is given to the user stickiness score of the user type to be scored, providing a quantitative specification for scoring the in-vehicle user stickiness scores of different user types, which can improve the accuracy of the in-vehicle user stickiness score, thereby solving the problem that the current in-vehicle user stickiness scoring method lacks a quantitative specification and objectivity, resulting in inaccurate in-vehicle user stickiness scores.

[0086] It should be noted that the multiple dimensions, multiple user types, subjective judgment matrix, and influencing factors related to the user stickiness of in-vehicle software mentioned in the embodiments of the present application are all statistically counted when the vehicle of the vehicle owner is a new vehicle. If the vehicle of the vehicle owner is a used vehicle, in order to ensure the accuracy of the in-vehicle user stickiness score, it is necessary to re-statistically count the above data for the vehicle owner.

[0087] See Figure 5 , Figure 5 FIG. is the structural block diagram of a user stickiness scoring device provided by an embodiment of the present application. The user stickiness scoring device 100 can be applied to a vehicle. The user stickiness scoring device 100 may include a matrix acquisition module 110, a weight calculation module 120, and a user stickiness scoring module 130.

[0088] The matrix acquisition module 110 is configured to acquire a subjective judgment matrix corresponding to the user type to be scored, where the subjective judgment matrix includes the influencing factors of each of the multiple dimensions on the user type and the influencing factors between the multiple dimensions, and the multiple dimensions are related to the user stickiness of in-vehicle software. Among them, for the specific description of the matrix acquisition module 110, please refer to step S110.

[0089] The weight calculation module 120 is configured to determine the weights of each of the multiple dimensions corresponding to the user type by using a graph convolutional network according to the subjective judgment matrix. Among them, for the specific description of the weight calculation module 120, please refer to step S120.

[0090] The user stickiness scoring module 130 is configured to determine the user stickiness score of the user type according to the weights of the multiple dimensions. Among them, for the specific description of the user stickiness scoring module 130, please refer to step S130.

[0091] Those skilled in the art can clearly understand that the above device provided by the embodiments of the present application can implement the method provided by the embodiments of the present application. For the specific working processes of the above-described device and modules, reference may be made to the corresponding processes of the methods in the embodiments of the present application, which will not be elaborated herein.

[0092] In the embodiments provided by the present application, the coupling, direct coupling, or communication connection between the modules shown or discussed with each other may be indirect coupling or communication coupling through some interfaces, devices, or modules, and may be in electrical, mechanical, or other forms. The embodiments of the present application do not make specific limitations thereto.

[0093] In addition, in the embodiments of the present application, each functional module may be integrated into a processing module, or each module may exist physically alone, or two or more modules may be integrated into one module. The above integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0094] See Figure 6 , Figure 6 is a structural block diagram of a vehicle provided by an embodiment of the present application. The vehicle 200 may include a memory 210 and a processor 220. An application program is stored in the memory 210, and the application program is configured to execute the method provided by the embodiments of the present application when called by the processor 220.

[0095] The processor 220 may include one or more processing cores. The processor 220 connects various parts within the entire vehicle 200 using various interfaces and lines, and is used to run or execute instructions, programs, code sets, or instruction sets stored in the memory 210, as well as to call and run or execute data stored in the memory 210, and to execute various functions of the vehicle 200 and process data.

[0096] The processor 220 can be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). The processor 220 can integrate one or a combination of several of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the display content; the modem is used to process wireless communication. It can be understood that the above-mentioned modem may not be integrated in the processor 220 and can be implemented separately through a communication chip.

[0097] The memory 210 can include random access memory (RAM) and can also include read-only memory (ROM). The memory 210 can be used to store instructions, programs, codes, code sets, or instruction sets. The memory 210 can include a program storage area and a data storage area. Among them, the program storage area can store instructions for implementing the operating system, instructions for implementing at least one function, instructions for implementing the above method embodiments, etc. The data storage area can store data created during the use of the vehicle 200.

[0098] The embodiments of the present application also provide a computer-readable storage medium, on which program code is stored, and the program code is configured to execute the method provided by the embodiments of the present application when called by the processor.

[0099] The computer-readable storage medium can be an electronic memory such as flash memory, electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), a hard disk, or ROM.

[0100] In some embodiments, the computer-readable storage medium includes a non-transitory computer-readable medium (Non-Transitory Computer-Readable Storage Medium, hereinafter referred to as Non-TCRSM). The computer-readable storage medium has a storage space for program code that executes any of the method steps in the above methods. These program codes can be read from or written into one or more computer program products. The program code can be compressed in a suitable form.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for user stickiness scoring, characterized in that, it includes: Obtain the subjective judgment matrix corresponding to the user type to be scored, where the subjective judgment matrix includes the influence factors of each of multiple dimensions on the user type and the influence factors between multiple dimensions, and the multiple dimensions are related to the user stickiness of the in-vehicle software; According to the subjective judgment matrix, use a graph convolutional network to determine the weights of each of the multiple dimensions corresponding to the user type; According to the weights of the multiple dimensions, determine the user stickiness score of the user type.

2. The method according to claim 1, characterized in that, The step of using a graph convolutional network to determine the weights of each of the multiple dimensions corresponding to the user type according to the subjective judgment matrix includes: According to the subjective judgment matrix, construct the degree matrix and adjacency matrix of the graph convolutional network; According to the degree matrix and adjacency matrix, through a two-layer graph convolutional network, determine the weights of each of the multiple dimensions.

3. The method according to claim 2, characterized in that, The step of constructing the degree matrix of the graph convolutional network according to the subjective judgment matrix includes: Using the arithmetic mean method in the analytic hierarchy process, according to the influence factors in the subjective judgment matrix, determine the preliminary weights of each of the multiple dimensions; According to the preliminary weights of the multiple dimensions, construct the degree matrix of the graph convolutional network.

4. The method according to claim 2, characterized in that, Each dimension in the multiple dimensions includes at least two labels. The step of constructing the adjacency matrix of the graph convolutional network according to the subjective judgment matrix includes: Combine the labels of two-by-two dimensions among the multiple dimensions to obtain multiple label combinations corresponding to two-by-two dimensions, and each label combination includes two labels, and the two labels come from different dimensions; Determine the first quantity of each label combination in the labels corresponding to the user type to be scored; Determine the second quantity of each label combination in the labels corresponding to all user types, where the user type to be scored is one of all user types; Determine the ratio of the first quantity to the second quantity corresponding to each label combination; Determine the sum of the ratios of the multiple label combinations as the target ratio of the two-by-two dimensions; According to the target ratio of the two-by-two dimensions, construct the adjacency matrix of the graph convolutional network.

5. The method according to claim 4, characterized in that, The step of constructing the adjacency matrix of the graph convolutional network according to the target ratio of the two-by-two dimensions includes: Mark the two-by-two dimensions with a target ratio greater than the ratio threshold as 1, and mark the two-by-two dimensions with a target ratio less than the ratio threshold as 0, and construct the adjacency matrix of the graph convolutional network.

6. The method according to claim 1, characterized in that, The step of determining the user stickiness score of the user type according to the weights of the multiple dimensions includes: According to the number of dimensions of the multiple dimensions, determine the basic score of each dimension in the multiple dimensions; According to the basic score of each dimension and the weights of the multiple dimensions, determine the user stickiness score of the user type.

7. The method according to claim 6, characterized in that, Determining the user stickiness score of the user type according to the basic score of each dimension and the weights of the multiple dimensions includes: Determining the product of the weight and the basic score of each dimension; Determining the sum of the products of the multiple dimensions as the user stickiness score of the user type.

8. The method according to claim 6, wherein, the multiple dimensions include multiple dimensions related to vehicle owner attributes and at least two dimensions related to in-vehicle infotainment system usage. Determining the basic score of each dimension in the multiple dimensions according to the number of dimensions of the multiple dimensions includes: Determining the basic score of each dimension related to vehicle owner attributes according to the number of dimensions of the multiple dimensions related to vehicle owner attributes; Determining the basic score of each dimension related to in-vehicle infotainment system usage according to the number of dimensions of at least two dimensions related to in-vehicle infotainment system usage.

9. The method according to claim 8, wherein, each dimension in the at least two dimensions related to in-vehicle infotainment system usage includes multiple different levels of tags. Determining the basic score of each dimension related to in-vehicle infotainment system usage according to the number of dimensions of the at least two dimensions related to in-vehicle infotainment system usage includes: Respectively obtaining a tag corresponding to the user type to be scored as a target tag from each of the at least two dimensions related to in-vehicle infotainment system usage, obtaining at least two target tags; Determining the basic score of each dimension related to in-vehicle infotainment system usage according to the levels of the at least two target tags and the number of dimensions of the at least two dimensions related to in-vehicle infotainment system usage.

10. The method according to claim 9, wherein, Determining the basic score of each dimension related to in-vehicle infotainment system usage according to the levels of the at least two target tags and the number of dimensions of the at least two dimensions related to in-vehicle infotainment system usage includes: Dividing a preset score by the number of dimensions of the at least two dimensions related to in-vehicle infotainment system usage to obtain an intermediate value; Multiplying the level of the target tag by the intermediate value to obtain the basic score of the dimension related to in-vehicle infotainment system usage corresponding to the target tag.

11. The method according to claim 8, wherein, Determining the basic score of each dimension related to vehicle owner attributes according to the number of dimensions of the multiple dimensions related to vehicle owner attributes includes: Dividing a preset score by the number of dimensions of the multiple dimensions related to vehicle owner attributes to obtain the basic score of each dimension related to vehicle owner attributes.

12. The method according to any one of claims 8-11, wherein, the multiple dimensions related to vehicle owner attributes include: the gender, age, permanent residence city, marital and childbearing status, income status, education level, work industry, and vehicle usage frequency of the vehicle owner; the at least two dimensions related to in-vehicle infotainment system usage include: the usage intensity and activity of in-vehicle infotainment system software.

13. A user stickiness scoring device, wherein, comprising: a matrix acquisition module, configured to acquire a subjective judgment matrix corresponding to the user type to be scored, the subjective judgment matrix including the influence factors of each of the multiple dimensions on the user type and the influence factors between the multiple dimensions, and the multiple dimensions being related to the user stickiness of in-vehicle infotainment system software; A weight calculation module, configured to determine the weights of respective multiple dimensions corresponding to the user type by using a graph convolutional network according to the subjective judgment matrix; A user stickiness scoring module, configured to determine the user stickiness score of the user type according to the weights of the multiple dimensions.

14. A vehicle, characterized in that, comprising: a memory and a processor, wherein an application program is stored on the memory, and the application program is configured to execute the method according to any one of claims 1-12 when called by the processor.

15. A computer-readable storage medium, characterized in that, program code is stored on the computer-readable storage medium, and the program code is configured to execute the method according to any one of claims 1-12 when called by a processor.

Images