Big data driven smart marketing prediction method

Abstract

The invention relates to the technical field of marketing prediction, in particular to a big-data-driven smart marketing prediction method, which comprises the steps of determining a prediction mode adopted for a user based on a comprehensive evaluation value of acquired historical behavior data of a plurality of users; determining the appeal degree of the user based on the feature complexity characterization parameters of the keyword features; determining the classification of the required products based on the dimension number of the acquired required product information data; determining the qualification of the number of extracted required product parameters based on the attention of the demand degree user; determining the eligibility of the number of core function features of the extracted required product based on the feature discrete coefficient of the degree user; according to the method, real-time behavior data of a plurality of users are obtained, the qualification of the prediction mode is determined based on the prediction accuracy of the prediction mode, and increase of keyword feature extraction frequency or increase of demand product information data acquisition frequency is determined according to the ratio of the prediction accuracy to the prediction accuracy threshold, so that the accuracy of marketing prediction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of marketing forecasting, and in particular to a big data driven intelligent marketing forecasting method. Background Art

[0002] Consumer behavior is becoming increasingly complex, and market data is growing in massive amounts. Traditional methods mostly rely on a small amount of historical sales data, simple market research, and empirical judgments, which have significant limitations. Historical data cannot keep up with current consumer demand and new market trends, research samples are limited and subjective, and empirical judgments lack objective basis, making it difficult to cope with the complex and changing market landscape.

[0003] Chinese patent application publication number: CN118780845A discloses a method for generating precision marketing strategies based on big data, including the following steps: S1: collecting real-time data of target users, processing the real-time data of target users, and determining the characteristic data of target users; S2: constructing a precision generation model for marketing strategies, testing and optimizing the precision generation model for marketing strategies, and determining the best precision generation model for marketing strategies; S3: analyzing and predicting the characteristic data of target users based on the best precision generation model for marketing strategies, and generating precision marketing strategies. However, the prior art has the following problems: the prior art does not fully conduct in-depth analysis of user behavior data, resulting in the problem of low accuracy of marketing predictions. Summary of the invention

[0004] To this end, the present invention provides a big data-driven intelligent marketing prediction method to overcome the problem in the prior art that user behavior data is not fully analyzed in depth, resulting in low accuracy of marketing predictions.

[0005] To achieve the above objectives, the present invention provides a big data driven intelligent marketing prediction method, comprising:

[0006] Obtain historical behavior data of several users, and determine a prediction model for the users based on the comprehensive evaluation value of the historical behavior data;

[0007] Obtaining chat record data of the user, extracting keyword features related to user needs in the chat record data, and determining the user's demand level based on feature complexity representation parameters of the keyword features, including a low demand level or a high demand level;

[0008] Under the condition of obtaining the information data of the user's demand product, determine whether the demand product is a single demand product or a complex demand product based on the number of dimensions of the demand product information data;

[0009] Under the condition that the demand product is determined to be a single demand product, the demand product parameters are extracted, and the eligibility of the number of extracted demand product parameters is determined based on the attention of users with low or high demand levels;

[0010] Under the condition that the demand product is determined to be a complex demand product, the core functional features of the demand product are extracted, and the eligibility of the number of extracted core functional features of the demand product is determined based on the characteristic dispersion coefficient of users with low or high demand levels;

[0011] Real-time behavior data of several users are obtained, and the eligibility of the corresponding prediction model is determined based on the prediction accuracy of the user's corresponding prediction model, and the frequency of keyword feature extraction or demand product information data collection is increased according to the ratio of the prediction accuracy to the prediction accuracy threshold.

[0012] Furthermore, by capturing users' purchase decision data and marketing strategy data on demand products in big data, the correlation determined based on the purchase decision change rate of the purchase decision data and the marketing strategy change rate of the marketing strategy data is compared with the preset correlation, and it is determined to increase the number of dimensions of the keyword features by the first preset keyword dimension adjustment coefficient or the second preset keyword dimension adjustment coefficient.

[0013] Further, based on the comparison result that the comprehensive evaluation value of the historical behavior data is less than or equal to the preset comprehensive evaluation value, it is determined that the first prediction mode is adopted for the user.

[0014] Furthermore, based on the comparison result that the comprehensive evaluation value of the historical behavior data is greater than the preset comprehensive evaluation value, it is determined that the second prediction mode is adopted for the user.

[0015] Further, based on the comparison result that the feature complexity representation parameter of the extracted keyword feature is less than or equal to the preset feature complexity representation parameter, it is determined that the user demand level is low.

[0016] Furthermore, based on the comparison result that the feature complexity representation parameter of the extracted keyword feature is greater than the preset feature complexity representation parameter, it is determined that the user demand degree is a high demand degree.

[0017] Furthermore, based on the comparison result that the number of dimensions of the demand product information data is less than or equal to the preset number of dimensions, the demand product is determined to be a single demand product, and based on the comparison result that the user's attention to the corresponding demand level is greater than the preset attention level, it is determined that the number of extracted demand product parameters is qualified.

[0018] Furthermore, based on the fact that the number of dimensions of the demand product information data is greater than the preset number of dimensions, the demand product is determined to be a complex demand product, and based on the comparison result that the feature discrete coefficient of users with corresponding demand degrees is greater than the preset feature discrete coefficient, it is determined that the number of core functional features of the demand product to be extracted is qualified.

[0019] Furthermore, based on the comparison result that the prediction accuracy of the corresponding prediction model is less than or equal to the prediction accuracy threshold, it is determined that the corresponding prediction model is unqualified, and based on the comparison result that the ratio of the prediction accuracy to the prediction accuracy threshold is less than or equal to the preset ratio, it is determined to increase the keyword feature extraction frequency by a preset extraction frequency adjustment coefficient.

[0020] Furthermore, according to the comparison result that the ratio of the prediction accuracy to the prediction accuracy threshold is greater than the preset ratio, it is determined that the frequency of collecting the demand product information data is increased by a preset collection frequency adjustment coefficient.

[0021] Compared with the prior art, the beneficial effect of the present invention lies in that the present invention determines the prediction model by obtaining user historical behavior data, extracts chat record keyword features to determine the degree of user demand, distinguishes product types according to the number of required product information data dimensions, and determines the eligibility of the number of extracted parameters or functional features for different product types in combination with user demands, and then verifies the prediction model based on real-time behavior data and determines the data processing frequency adjustment strategy, so as to accurately grasp user needs, make product information processing more targeted, improve the efficiency of marketing prediction, and improve the accuracy of marketing prediction judgment.

[0022] Furthermore, the present invention determines whether to adopt the first or second prediction mode through the comprehensive evaluation value of the user's historical behavior data, obtains the chat record data and extracts the keyword features under the determined prediction mode, and then determines the degree of user demand according to the feature complexity characterization parameters, accurately predicts according to the user's historical behavior, and improves the accuracy of the prediction.

[0023] Furthermore, the present invention determines the classification of products through the number of dimensions of demand product information data, and for a single demand product, determines whether the number of extracted product parameters is qualified based on the comparison result between user attention and preset attention, thereby efficiently performing preliminary classification of demand products and ensuring that the number of extracted product parameters meets user needs for a single demand product, thereby improving the accuracy of product recommendations.

[0024] Furthermore, the present invention determines that the demand product is a complex demand product, extracts core functional features, and judges whether the number of extracted core functional features is qualified by comparing the user feature discrete coefficient with the preset feature discrete coefficient. For complex demand products, the extracted core functional features both meet the user's basic cognition and are not too lengthy or lack key information, thereby improving the accuracy of product recommendations.

[0025] Furthermore, the present invention captures and analyzes the user's purchase decision data on the demand products and the enterprise's marketing strategy data, and adjusts the number of dimensions of the keyword features according to the correlation between the change rates of the two, so as to more effectively match user needs with marketing strategies and improve the accuracy of marketing predictions.

[0026] Furthermore, the present invention determines the eligibility of the prediction model by comparing the prediction accuracy of the prediction model with the prediction accuracy threshold. If it is unqualified, the prediction model is optimized by adjusting the keyword feature extraction frequency or the demand product information data collection frequency, and the performance of the prediction model is evaluated. The optimization strategy is adjusted according to the evaluation results, thereby improving the accuracy and adaptability of the prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flow chart of a big data driven smart marketing prediction method according to an embodiment of the present invention;

[0028] Figure 2 A flowchart for extracting the quantity eligibility of required product parameters according to an embodiment of the present invention;

[0029] Figure 3 A flowchart for extracting a qualified number of core functional features for an embodiment of the present invention;

[0030] Figure 4 The flowchart of the embodiment of the present invention is used to predict the eligibility of the model. DETAILED DESCRIPTION

[0031] In order to make the purpose and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be recognized that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0032] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be recognized by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0033] It should be pointed out that the data in this embodiment are obtained by comprehensive analysis and evaluation of the historical test data and the corresponding historical test results of the present invention in the three months before this test. It can be recognized by those skilled in the art that the present invention can determine the above parameters for a single item by selecting the highest proportion of values ​​as the preset standard parameters according to the data distribution, using weighted summation to use the obtained values ​​as the preset standard parameters, substituting each historical data into a specific formula and using the values ​​obtained by the formula as the preset standard parameters or other selection methods, as long as the present invention can clearly define different specific situations in the single determination process through the obtained values.

[0034] See also Figure 1-Figure 4 As shown, Figure 1 This is a flow chart of a big data driven smart marketing prediction method according to an embodiment of the present invention; Figure 2 A flowchart for extracting the quantity eligibility of required product parameters according to an embodiment of the present invention; Figure 3A flowchart for extracting a qualified number of core functional features for an embodiment of the present invention; Figure 4 The flowchart of the embodiment of the present invention is used to predict the eligibility of the model.

[0035] The big data driven intelligent marketing prediction method of the embodiment of the present invention includes:

[0036] Step S1, obtaining historical behavior data of several users, and determining whether to adopt a first prediction mode or a second prediction mode for the users based on a comprehensive evaluation value of the historical behavior data;

[0037] Step S2, obtaining the user's chat record data, extracting the keyword features related to the user's needs in the chat record data, and determining whether the user's demand level is low or high based on the feature complexity representation parameters of the keyword features;

[0038] Step S3, under the condition of obtaining the user's demand product information data, determine whether the demand product is a single demand product or a complex demand product based on the number of dimensions of the demand product information data;

[0039] Step S4, under the condition that the demand product is determined to be a single demand product, extract the demand product parameters, and determine the eligibility of the number of extracted demand product parameters based on the attention of users with low or high demand levels;

[0040] Step S5, under the condition that the demand product is determined to be a complex demand product, extract the core functional features of the demand product, and determine the eligibility of the number of extracted core functional features of the demand product based on the characteristic dispersion coefficient of users with low or high demand levels;

[0041] Step S6, obtaining real-time behavior data of several users, determining the eligibility of the second prediction model or the first prediction model based on the prediction accuracy of the second prediction model or the first prediction model, and determining to increase the frequency of keyword feature extraction or increase the frequency of demand product information data collection based on the ratio of the prediction accuracy to the prediction accuracy threshold.

[0042] In the embodiment of the present invention, the behavior data refers to various behavior record information data generated by business activities related to users and enterprises, including browsing time data, browsing frequency data, and purchase amount data.

[0043] In the embodiment of the present invention, the chat record data is text information record data generated when the user communicates and interacts with the relevant platform within an extraction period of one month.

[0044] In the embodiment of the present invention, the demand product information data is all-round descriptive data about the demand product within an extraction period of one month, including basic attribute data, function parameter data, and performance parameter data.

[0045] In the embodiment of the present invention, obtaining the user's behavior data and chat record data requires user authorization.

[0046] Specifically, the present invention determines the prediction model by obtaining historical user behavior data, extracts keyword features from chat records to determine the degree of user demands, distinguishes product types according to the number of dimensions of required product information data, and determines the eligibility of the number of extracted parameters or functional features for different product types in combination with user demands. The prediction model is then verified based on real-time behavior data and a data processing frequency adjustment strategy is determined to accurately grasp user needs, make product information processing more targeted, improve the efficiency of marketing predictions, and improve the accuracy of marketing prediction judgments.

[0047] Specifically, the embodiment of the present invention determines the prediction mode adopted for the user based on the comparison result of the comprehensive evaluation value of the historical behavior data and the preset comprehensive evaluation value of 0.85 under the condition of obtaining the historical behavior data of several users;

[0048] When the comprehensive evaluation value is less than or equal to the preset comprehensive evaluation value, determining to adopt the first prediction mode for the user;

[0049] When the comprehensive evaluation value is greater than the preset comprehensive evaluation value, it is determined to adopt the second prediction mode for the user.

[0050] In the embodiment of the present invention, the preset comprehensive evaluation value is 0.85, but the above value is not limited thereto, and those skilled in the art may also adjust the value according to actual needs.

[0051] In the embodiment of the present invention, the first prediction mode is to analyze the user behavior data, and the second prediction mode is to analyze the user behavior data, the user's purchase decision data, and the marketing strategy data respectively.

[0052] Specifically, the embodiment of the present invention calculates the comprehensive evaluation value according to the following formula, setting:

[0053]

[0054] Among them, P represents the comprehensive evaluation value, T max is the maximum browsing time, T min is the minimum browsing time, μ T is the average browsing time, F max is the maximum browsing frequency value, F min is the minimum browsing frequency value, μ F is the average browsing frequency value, M max is the maximum purchase amount, M min is the minimum purchase amount, μ M is the average purchase amount.

[0055] Specifically, the embodiment of the present invention obtains the user's chat record data under the condition of determining different prediction modes for the user, and extracts keyword features about user needs in the chat record data at a preset keyword feature extraction frequency of 20 times / min. The preset keyword feature extraction frequency is obtained by taking the average of several historical keyword feature extraction frequencies.

[0056] In the embodiment of the present invention, the dimensions of the keyword feature are the frequency and quantity of occurrence of the keyword feature, and the number of dimensions of the keyword feature is 2.

[0057] Specifically, the embodiment of the present invention determines the user's appeal level based on the comparison result of the feature complexity characterization parameter of the keyword feature and the preset feature complexity characterization parameter 0.83.

[0058] When the feature complexity characterization parameter is less than or equal to the preset feature complexity characterization parameter, it is determined that the user demand level is low demand level;

[0059] When the feature complexity characterization parameter is greater than the preset feature complexity characterization parameter, it is determined that the user demand level is a high demand level.

[0060] In the embodiment of the present invention, the preset feature complexity characterization parameter value is 0.83, but the above value is not limited to this, and those skilled in the art can also adjust the value according to actual needs.

[0061] In the embodiment of the present invention, the keyword feature is a phrase that can reflect the user's purchase intention, such as words such as "like", "expect", or "intend".

[0062] In the embodiment of the present invention, the low appeal level refers to an appeal with a small number of keyword features, and the high appeal level refers to an appeal with a large number of keyword features.

[0063] Specifically, the embodiment of the present invention calculates the feature complexity characterization parameter according to the following formula, setting:

[0064]

[0065] Among them, Q represents the characteristic complexity representation parameter, f i is the frequency of occurrence of the ith keyword feature in the chat record data, n represents the total number of extracted keyword features, max(f i ) is the frequency of the keyword feature with the highest frequency in the chat record data, k is the number of types of keyword features in the chat record data, k 0 The maximum number of types of keyword features in the chat record data of several users.

[0066] Specifically, the present invention determines whether to adopt the first or second prediction mode through the comprehensive evaluation value of the user's historical behavior data, obtains chat record data and extracts keyword features under the determined prediction mode, and then determines the degree of user demand based on the feature complexity characterization parameters, accurately predicts based on the user's historical behavior, and improves the accuracy of the prediction.

[0067] Specifically, the embodiment of the present invention acquires user demand product information data at a demand product information data collection frequency of 5 times / s, and the preset demand product information data collection frequency is an average value of several historical demand product information data collection frequencies.

[0068] Specifically, the embodiment of the present invention determines the classification of the demand product according to the comparison result of the number of dimensions of the demand product information data and the preset number of dimensions 3;

[0069] When the number of dimensions is less than or equal to the preset number of dimensions, it is determined that the demand product is a single demand product;

[0070] When the number of dimensions is greater than the preset number of dimensions, it is determined that the demand product is a complex demand product.

[0071] In the embodiment of the present invention, the preset value of the number of dimensions is 3, but the above value is not limited to this, and those skilled in the art can also adjust the value according to actual needs.

[0072] In the embodiment of the present invention, the single demand product is a product that only includes basic attribute data, and the complex demand product is a product that includes basic attribute data, function parameter data and performance parameter data.

[0073] In the embodiment of the present invention, the number of dimensions is the number of attributes inherent in the required product information data, such as size and weight of basic attribute data, control mode and compatibility of functional parameter data, power and efficiency of performance parameter data, etc.

[0074] Specifically, the embodiment of the present invention determines the number of dimensions according to the number of attributes inherent in the listed demand product information data.

[0075] Specifically, under the condition that the demand product is determined to be a single demand product, the embodiment of the present invention extracts the demand product parameters, and determines the eligibility of the number of extracted demand product parameters according to the comparison result of the attention degree of the user of the corresponding demand degree and the preset attention degree of 0.79;

[0076] When the attention level is less than or equal to the preset attention level, it is determined that the number of extracted demand product parameters is qualified;

[0077] When the attention level is greater than the preset attention level, it is determined that the number of extracted demand product parameters is unqualified.

[0078] In the embodiment of the present invention, the preset attention value is 0.79, and the preset attention value is obtained by taking the average of the attention values ​​of several historical qualified demand product parameters, but the above value is not limited to this, and technical personnel in this field can also adjust the value according to actual needs.

[0079] In the embodiment of the present invention, the required product parameters are parameters included in the basic attribute data.

[0080] Specifically, the attention degree is the difference between the maximum attention frequency of the parameter and the minimum attention frequency of the parameter in the basic attribute data divided by the average frequency of the parameter.

[0081] Specifically, the present invention determines the classification of products through the number of dimensions of demand product information data, and for a single demand product, determines whether the number of extracted product parameters is qualified based on the comparison result between user attention and preset attention, thereby efficiently performing preliminary classification of demand products and ensuring that the number of extracted product parameters meets user needs for a single demand product, thereby improving the accuracy of product recommendations.

[0082] Specifically, under the condition that the demand product is determined to be a complex demand product, the embodiment of the present invention extracts the core functional features, and determines the eligibility of the number of extracted core functional features based on the comparison result of the feature dispersion coefficient of the user with the corresponding demand degree and the preset feature dispersion coefficient of 0.81;

[0083] When the feature dispersion coefficient is less than or equal to the preset feature dispersion coefficient, it is determined that the number of extracted core functional features is qualified;

[0084] When the feature dispersion coefficient is greater than the preset feature dispersion coefficient, it is determined that the number of extracted core functional features is unqualified.

[0085] In the embodiment of the present invention, the preset feature dispersion coefficient is 0.81, and the preset feature dispersion coefficient is obtained by taking the average value of the feature dispersion coefficients of several qualified core functional features in history, but the above value is not limited to this, and technical personnel in this field can also adjust the value according to actual needs.

[0086] In the embodiment of the present invention, the core functional features are features included in the functional parameter data and the performance parameter data.

[0087] Specifically, the embodiment of the present invention calculates the characteristic dispersion coefficient according to the following formula, setting:

[0088]

[0089] Among them, D represents the characteristic dispersion coefficient, A max is the maximum frequency of the parameter in the functional parameter data, Amin is the minimum frequency of attention of the parameter in the functional parameter data, is the average frequency of attention of parameters in the functional parameter data, B max is the maximum frequency of the parameter in the performance parameter data, B min is the minimum frequency of attention of the parameter in the performance parameter data, is the average attention frequency of the parameters in the performance parameter data.

[0090] Specifically, the present invention determines that the demand product is a complex demand product, extracts core functional features, and judges whether the number of extracted core functional features is qualified by comparing the user feature discrete coefficient with the preset feature discrete coefficient. For complex demand products, the extracted core functional features both meet the user's basic cognition and are not too lengthy or lack key information, thereby improving the accuracy of product recommendations.

[0091] Specifically, the present invention captures users' purchase decision data and marketing strategy data on demand products from big data. The purchase decision data is data reflecting the decision-making process of users purchasing demand products, and the marketing strategy data is data for enterprises to formulate and implement marketing strategies. The marketing strategies include strategies such as advertising, promotional activities and product recommendations.

[0092] Specifically, under the condition of determining the classification of demand products, the embodiment of the present invention determines the marketing strategy change rate according to the user's purchase decision change rate for the demand products to analyze the change rate of the marketing strategy, the purchase decision change rate is the ratio of the number of purchase decision changes to the total number of purchase decision changes, the purchase decision change is the process from the user determining a number of demand products to the payment, due to the influence of other factors, the purchase decision is changed, and the demand products that are not selected are purchased, the total number of purchase decisions is the number of times the user purchases the demand products; the marketing strategy change rate is the ratio of the number of marketing strategy changes to the total number of marketing strategy formulations, the number of marketing strategy changes is the changes in advertising strategies, promotional activities, product recommendations, etc. during the marketing process, and the total number of marketing strategies is the total number of times the marketing strategies are executed.

[0093] Specifically, the embodiment of the present invention determines the number of dimensions for adjusting keyword features according to the comparison result of the correlation between the marketing strategy change rate and the purchase decision change rate and the preset correlation;

[0094] When the correlation is less than or equal to the preset correlation, it is determined to increase the number of dimensions of the keyword feature to a corresponding value by a first preset keyword dimension adjustment coefficient of 2;

[0095] When the correlation is greater than a preset correlation, it is determined to increase the number of dimensions of the keyword feature to a corresponding value using a second preset keyword dimension adjustment coefficient of 4.

[0096] In the embodiment of the present invention, the preset correlation value is 0.52, but the above value is not limited thereto, and those skilled in the art may also adjust the value according to actual needs.

[0097] In an embodiment of the present invention, the number of dimensions of the increased keyword feature is the product of the number of dimensions and the lth preset keyword dimension adjustment coefficient, where l is 1 or 2, U1 is the first preset keyword dimension adjustment coefficient 2, and U2 is the second preset keyword dimension adjustment coefficient 4.

[0098] Specifically, the embodiment of the present invention calculates the correlation degree according to the following formula, setting:

[0099]

[0100] in, Indicates the degree of association, is the purchase decision change rate at time t, is the marketing strategy change rate at time t, is the mean of the purchase decision change rate, is the mean rate of change of marketing strategies.

[0101] Specifically, the present invention captures and analyzes the user's purchase decision data on the demand products and the enterprise's marketing strategy data, adjusts the number of dimensions of keyword features according to the correlation between the change rates of the two, and more effectively matches user needs with marketing strategies, thereby improving the accuracy of marketing predictions.

[0102] Specifically, the embodiment of the present invention determines the eligibility of the prediction model based on the comparison result of the prediction accuracy of the prediction model with the prediction accuracy threshold of 95% under the condition of obtaining the real-time behavior data of several users;

[0103] When the prediction accuracy is less than or equal to the prediction accuracy threshold, it is determined that the prediction model is unqualified;

[0104] When the prediction accuracy is greater than the prediction accuracy threshold, it is determined that the prediction model is qualified.

[0105] In the embodiment of the present invention, the prediction accuracy threshold is set to 95%, but the above value is not limited thereto, and those skilled in the art may also adjust the value according to actual needs.

[0106] Specifically, the prediction accuracy is the ratio of the result of the prediction model on the real-time behavior data to the actual result of the real-time behavior data.

[0107] Specifically, under the condition that the corresponding prediction model is determined to be unqualified, the embodiment of the present invention determines the adjustment method of the prediction model according to the comparison result of the ratio of the prediction accuracy rate of the prediction model to the prediction accuracy rate threshold and the preset ratio 0.55;

[0108] When the ratio is less than or equal to the preset ratio, it is determined to increase the keyword feature extraction frequency to a corresponding value by a preset extraction frequency adjustment coefficient of 1.04;

[0109] When the ratio is greater than the preset ratio, it is determined to increase the frequency of collecting demand product information data to a corresponding value by using a preset collection frequency adjustment coefficient of 1.07;

[0110] The ratio is the ratio of the prediction accuracy to the prediction accuracy threshold.

[0111] In the embodiment of the present invention, the preset ratio is 0.55, but the above value is not limited thereto, and those skilled in the art may also adjust the value according to actual needs.

[0112] In an embodiment of the present invention, the increased keyword feature extraction frequency is the product of the preset keyword feature extraction frequency and the preset extraction frequency adjustment coefficient 1.04, and the increased demand product information data collection frequency is the product of the preset demand product information data collection frequency and the preset collection frequency adjustment coefficient 1.07.

[0113] Specifically, the present invention determines the eligibility of the prediction model by comparing the prediction accuracy of the prediction model with the prediction accuracy threshold. If it is unqualified, the prediction model is optimized by adjusting the keyword feature extraction frequency or the demand product information data collection frequency, and the performance of the prediction model is evaluated. The optimization strategy is adjusted according to the evaluation results, thereby improving the accuracy and adaptability of the prediction model.

[0114] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to recognize that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0115] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A big data driven intelligent marketing prediction method, characterized in that: include: Obtain historical behavior data of several users, and determine a prediction model for the users based on the comprehensive evaluation value of the historical behavior data; Obtaining chat record data of the user, extracting keyword features related to user needs in the chat record data, and determining the user's demand level based on feature complexity representation parameters of the keyword features, including a low demand level or a high demand level; Under the condition of obtaining the information data of the user's demand product, determine whether the demand product is a single demand product or a complex demand product based on the number of dimensions of the demand product information data; Under the condition that the demand product is determined to be a single demand product, the demand product parameters are extracted, and the eligibility of the number of extracted demand product parameters is determined based on the attention of users with low or high demand levels; Under the condition that the demand product is determined to be a complex demand product, the core functional features of the demand product are extracted, and the eligibility of the number of extracted core functional features of the demand product is determined based on the characteristic dispersion coefficient of users with low or high demand levels; Real-time behavior data of several users are obtained, and the eligibility of the corresponding prediction model is determined based on the prediction accuracy of the user's corresponding prediction model, and the frequency of keyword feature extraction or demand product information data collection is increased according to the ratio of the prediction accuracy to the prediction accuracy threshold.

2. The big data driven intelligent marketing prediction method according to claim 1, characterized in that: By capturing the purchase decision data and marketing strategy data of users on the demand products in the big data, and comparing the correlation determined by the purchase decision change rate of the purchase decision data and the marketing strategy change rate of the marketing strategy data with the preset correlation, it is determined to increase the number of dimensions of the keyword features by the first preset keyword dimension adjustment coefficient or the second preset keyword dimension adjustment coefficient.

3. The big data driven intelligent marketing prediction method according to claim 1, characterized in that: Based on the comparison result that the comprehensive evaluation value of the historical behavior data is less than or equal to the preset comprehensive evaluation value, it is determined that the first prediction mode is adopted for the user.

4. The big data driven intelligent marketing prediction method according to claim 1, characterized in that: Based on the comparison result that the comprehensive evaluation value of the historical behavior data is greater than the preset comprehensive evaluation value, it is determined that the second prediction mode is adopted for the user.

5. The big data driven intelligent marketing prediction method according to claim 1, characterized in that: Based on the comparison result that the feature complexity representation parameter of the extracted keyword feature is less than or equal to the preset feature complexity representation parameter, it is determined that the user demand level is low.

6. The big data driven intelligent marketing prediction method according to claim 1, characterized in that: The user's demand degree is determined to be high based on the comparison result that the feature complexity representation parameter of the extracted keyword feature is greater than the preset feature complexity representation parameter.

7. The big data driven intelligent marketing prediction method according to claim 5 or 6, characterized in that: Based on the comparison result that the number of dimensions of the demand product information data is less than or equal to the preset number of dimensions, the demand product is determined to be a single demand product, and based on the comparison result that the user's attention to the corresponding demand level is greater than the preset attention level, it is determined that the number of extracted demand product parameters is qualified.

8. The big data driven intelligent marketing prediction method according to claim 5 or 6, characterized in that: Based on the fact that the number of dimensions of the demand product information data is greater than the preset number of dimensions, the demand product is determined to be a complex demand product, and based on the comparison result that the feature discrete coefficient of users with corresponding demand levels is greater than the preset feature discrete coefficient, it is determined that the number of core functional features of the demand product to be extracted is qualified.

9. The big data driven intelligent marketing prediction method according to claim 1, characterized in that: Based on the comparison result that the prediction accuracy of the corresponding prediction model is less than or equal to the prediction accuracy threshold, the corresponding prediction model is determined to be unqualified, and based on the comparison result that the ratio of the prediction accuracy to the prediction accuracy threshold is less than or equal to the preset ratio, it is determined to increase the keyword feature extraction frequency by a preset extraction frequency adjustment coefficient.

10. The big data driven intelligent marketing prediction method according to claim 9, characterized in that: According to the comparison result that the ratio of the prediction accuracy to the prediction accuracy threshold is greater than the preset ratio, it is determined to increase the demand product information data collection frequency by a preset collection frequency adjustment coefficient.

Images