E-commerce information intelligent recommendation system and method based on data analysis

Abstract

The invention discloses an e-commerce information intelligent recommendation system and method based on data analysis, and relates to the technical field of information recommendation, and the recommendation method comprises the following steps: collecting each e-commerce information browsing process of a user after user authorization, and generating a corresponding information browsing record; performing feature extraction on user behavior information and browsing information generated in any information browsing record; determining the information category of each piece of browsing information, analyzing the preference condition of each information category, and carrying out information layering on each information category; performing association analysis on the user behavior and the browsing information in the information browsing record to obtain the information browsing density of the user, and presetting an expected recommendation density for the user; when the user browses the e-commerce information, corresponding information recommendation proportions are distributed to various information types, and real-time browsing density is obtained; and comparing with a preset expected recommendation density, and adjusting subsequent information recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of information recommendation, and in particular to an e-commerce information intelligent recommendation system and method based on data analysis. Background Art

[0002] With the rapid development of e-commerce, online shopping has become an integral part of consumers' daily lives. E-commerce platforms provide users with personalized shopping experiences through intelligent recommendation systems, helping them quickly find products that meet their needs among a vast array of products. However, as the volume and complexity of recommended information increases, users may experience "decision fatigue" when choosing products. Decision fatigue refers to the gradual decline in a user's decision-making ability and willingness when faced with too many choices, leading to difficulty in making choices, decreased satisfaction, and even purchase abandonment.

[0003] Currently, many e-commerce platforms use recommendation algorithms based on user behavior and preferences to display a large number of recommended products to users. These systems often fail to effectively control the density of recommended information, causing users to feel confused and exhausted when faced with too many choices. Existing recommendation systems usually lack the ability to dynamically adjust the density of recommended information and are unable to adapt to users' real-time feedback and behavioral changes, which also affects users' shopping experience and purchasing decisions. Summary of the Invention

[0004] The purpose of the present invention is to provide an e-commerce information intelligent recommendation system and method based on data analysis to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligently recommending e-commerce information based on data analysis, the recommendation method comprising the following steps:

[0006] Step S100: After the user's authorization, each e-commerce information browsing process of the user is collected to generate corresponding information browsing records; feature extraction is performed on the user behavior information and browsing information generated in any information browsing record;

[0007] Step S200: Obtaining browsing information of the user in any information browsing history, determining the information category of each browsing information, analyzing the preferences of each information category; and stratifying each information category based on the user's preferences;

[0008] Step S300: For any information browsing record, correlation analysis is performed on the user behavior and browsing information in the information browsing record to obtain the user's information browsing density; based on the information browsing density presented in different information browsing records, an expected recommendation density is preset for the user;

[0009] Step S400: Whenever a user browses e-commerce information, a corresponding information recommendation ratio is assigned to each information type according to the information stratification, and a real-time browsing density is obtained based on the user's information browsing situation; by comparing it with the preset expected recommendation density, subsequent information recommendations are adjusted.

[0010] Furthermore, step S100 includes the following steps:

[0011] Step S101: During the e-commerce information browsing process, whenever a user is on a browsing page, the e-commerce information on the browsing page is collected to obtain a browsing information set. Simultaneously, the user behavior information generated during the browsing process is collected to obtain a behavior information set. The browsing information set and the behavior information set are aggregated to generate an information browsing record for the user. The user behavior information includes the user's stay time on the browsing page, the number of clicks, the number of page entries, etc. The browsing information includes the text and image information of the product, such as the product name and product attributes.

[0012] Step S102: randomly selecting an information browsing record, reading a browsing information set in the information browsing record, randomly selecting a browsing information from the browsing information set, wherein the browsing information includes a plurality of text information and image information, performing keyword extraction on the plurality of text information and feature extraction on the plurality of image information, generating an information data set of the arbitrary browsing information, and obtaining a plurality of information data sets of the selected information browsing record;

[0013] Step S103: extracting a behavior information set from the selected information browsing record, and dividing each behavior information in the behavior information set into a behavior name or behavior data according to the data type, wherein the behavior data is a specific value or a value range; wherein, any behavior name corresponds to a behavior data, and any behavior name is merged with the corresponding behavior data to generate a corresponding behavior data set, and obtain several behavior data sets of the selected information browsing record; the behavior name includes the stay time, the number of clicks, and the number of page entries. If it is a stay time period, a time interval will be obtained, and if it is the number of clicks, a specific value will be obtained.

[0014] Furthermore, step S200 includes the following steps:

[0015] Step S201: pre-constructing an information category database, wherein the information category database includes several information categories and several information feature sets, wherein one information category is matched with one information feature set;

[0016] Step S202: arbitrarily select an information browsing record, extract any browsing information from the selected information browsing record, compare each keyword and each feature in the information data set of the browsing information with the information features in the arbitrary information feature set, and count the number of identical features between the information data set and the information feature set; the information feature set with the largest number of identical features is defined as the target information feature set; if there is only one target information feature set, the information category corresponding to the target information feature set is defined as the information category of the browsing information; if there is more than one target information feature set, obtain the similarity between each difference feature between the arbitrary target information feature set and the information data set, calculate the average value to obtain the average similarity, and select the information category corresponding to the target information feature set with the largest average similarity as the information category of the browsing information;

[0017] To determine the information category of the browsed information, each keyword and feature needs to be compared with the information features corresponding to the information category. The more common features exist, the higher the probability that it belongs to the information category. If there are the same number of common features in multiple information categories, the remaining features need to be compared and the information category with the highest average similarity is selected.

[0018] Step S203: Obtain the information category of each browsing information in the information browsing record, and count the number of information of each information category; obtain the record generation time of each information browsing record, and sort them according to the record generation time, and obtain the number of information of the jth information category in the last information browsing record (N j ) last , sort the various information categories in the last information browsing record in ascending order according to the number of information, and get the position of the jth information category as A j ; Set the number of information of the jth information category in the i-th information browsing record (N j ) i , according to the formula:

[0019]

[0020] Where i is a positive integer and i∈(1,m), m is the total number of information categories; the number of comprehensive views of the j-th information category is calculated as X j The calculation of the number of comprehensive browsing takes into account the average browsing history and the browsing history of the most recent browsing history. The historical browsing history shows the browsing habits of users over a long period of time, while the most recent browsing history shows the user's current needs. Only by combining the two can we obtain a more accurate understanding of user preferences.

[0021] Step S204: Extract the maximum number of messages N from all message browsing records max, preset an information quantity threshold N th , if N max >N th , then the information quantity range (N th ,N max ) as the user's first preference level, and the information quantity range (0,N th ) is evenly divided into b preference levels; if N max <N th , then the information quantity range (0,N max ) is evenly divided into b+1 preference levels, and the preference levels corresponding to various information categories are obtained; if X j ∈(N th ,N max ), the jth information category is set as the first preference level; the purpose of setting the information quantity threshold is to prevent users from browsing information categories relatively evenly. If preference stratification is forced, it will cause users to lose some types of e-commerce information, resulting in incomplete e-commerce information recommendations.

[0022] Furthermore, step S300 includes the following steps:

[0023] Step S301: arbitrarily select an information browsing record, extract several behavior data sets from the information browsing record, and pre-set several density evaluation indicators, wherein the several density evaluation indicators include browsing time. The density evaluation indicators mainly analyze user behavior, such as the number of clicks and the length of time a user stays on a browsing page. Each user behavior can directly reflect the user's browsing experience. The more clicks, the more likely the user is to understand the information, and the shorter the stay time, the less interested the user is.

[0024] Step S302: arbitrarily select an information browsing record, extract several behavior data sets from the information browsing record, perform semantic similarity comparison between the behavior name in any behavior data set and any density evaluation index, match the behavior name with the greatest semantic similarity with the density evaluation index, and obtain the behavior data of the density evaluation index in each behavior data set; if the behavior data corresponding to the density evaluation index is a numerical value, then accumulate the behavior data of the density evaluation index in each behavior data set to obtain the behavior characteristic value z of the density evaluation index; if the behavior data corresponding to the density evaluation index is a numerical range, then perform a union operation on the behavior data in each behavior data set to obtain the numerical range (min, max) of the density evaluation index, and obtain the corresponding behavior characteristic value z = max-min;

[0025] Step S303: extract the information data set of each browsing information from the selected information browsing record, and count the number of keywords contained in all information data sets as nkey and the number of features n aspect , the number of information features contained in the selected information browsing record is n behave =n key +n aspect ; According to the formula:

[0026]

[0027] Where k is a positive integer and k∈(1,d), d is the number of density evaluation indicators set, z k is the behavioral characteristic value of the kth density evaluation index, and t is the browsing time. The information browsing density P of the user in the selected information browsing record is calculated. The information characteristics included in all the information browsed by the user are removed from the characteristic values of the browsing time and density evaluation index. The obtained value is the average amount of information browsed by the user under the influence of different indicators per unit time, which reflects the amount of information received by the user per unit time.

[0028] Step S304: Select the rth information browsing record, and accumulate the number of comprehensive browsings corresponding to each preference level in the rth information browsing record, wherein the total number of comprehensive browsings of the first preference level in the rth information browsing record is set as X1 r , according to the formula:

[0029]

[0030] Where r is a positive integer and r∈(1,u), u is the total number of information browsing records; the characteristic proportion coefficient δ of the rth information browsing record is calculated r ;

[0031] Step S305: Set the information browsing density of the rth information browsing record to P r , according to the formula:

[0032]

[0033] Calculate the user's expected recommendation density P ave ; The types of information that have the greatest impact on the user's browsing density must be the most popular with users. Therefore, under the premise of ensuring that the information recommendation is in line with the user's preferences, the more browsing records the user browses in the first preference level, the closer the resulting information browsing density will be to the expected recommendation density.

[0034] Furthermore, step S400 includes the following steps:

[0035] Step S401: Get the information categories contained in each current preference level, and set the number of information categories contained in the vth preference level to Q v; Randomly select the jth information category and set the jth information category to be at the vth preference level, according to the formula:

[0036]

[0037] Where b+1 is the number of preference levels; the information recommendation ratio W of the jth information category is calculated j ;

[0038] Step S402: Display e-commerce information corresponding to various information types according to the information recommendation ratio. After the user browses the information, the browsing information set and behavior information set browsed by the user are obtained after a time T. The number of information features n included in the browsing information set is counted. T And the total value of the behavioral characteristics of each density evaluation index z T , the user's real-time browsing density is calculated as P now =n T / (T×z T );

[0039] Step S403: Set the user's expected recommendation density as P ave , if P now <P ave , then when making the next e-commerce information recommendation, increase the recommendation ratio of the first preference level information; if P now >P ave , then continue to push e-commerce information according to the current information recommendation ratio.

[0040] In order to better implement the above method, an e-commerce information intelligent recommendation system is also proposed. The recommendation system includes a historical information analysis module, a recommendation information division module, a behavior density analysis module, and a user willingness analysis module;

[0041] The historical information analysis module is used to collect data on each user's e-commerce information browsing process after user authorization and generate corresponding information browsing records; it also extracts features from user behavior information and browsing information generated in any information browsing record;

[0042] The recommended information classification module is used to obtain the browsing information of the user in any information browsing history, determine the information category of each browsing information, analyze the preferences of various information categories, and classify the information of various information categories based on the user's preferences;

[0043] The behavior density analysis module is used to analyze the correlation between user behavior and browsing information in any information browsing record to obtain the user's information browsing density; based on the information browsing density presented in different information browsing records, an expected recommendation density is preset for the user;

[0044] The user intention analysis module is used to assign corresponding information recommendation ratios to various information types according to the information stratification whenever a user browses e-commerce information, and obtain real-time browsing density based on the user's information browsing situation; by comparing with the preset expected recommendation density, subsequent information recommendations are adjusted.

[0045] Furthermore, the historical information analysis module includes a historical information collection unit and a behavior information extraction unit;

[0046] The historical information collection unit is used to collect each e-commerce information browsing process of the user after the user's authorization and generate corresponding information browsing records; the behavior information extraction unit is used to extract features of the user behavior information and browsing information generated in any information browsing record.

[0047] Furthermore, the recommended information division module includes an information browsing analysis unit and a recommended information stratification unit;

[0048] The information browsing analysis unit is used to obtain the user's browsing information in any information browsing record, determine the information category of each browsing information, and analyze the preferences of various information categories; the recommended information stratification unit is used to stratify various information categories based on the user's preferences.

[0049] Furthermore, the behavior density analysis module includes a user behavior analysis unit and an expected density setting unit;

[0050] The user behavior analysis unit is used to perform correlation analysis on the user behavior and browsing information in any information browsing record to obtain the user's information browsing density; the expected density setting unit is used to preset an expected recommendation density for the user based on the information browsing density presented in different information browsing records.

[0051] Furthermore, the user willingness analysis module includes a user willingness evaluation unit and a recommendation density adjustment unit;

[0052] The user willingness evaluation unit is used to assign corresponding information recommendation ratios to various information types according to the information stratification whenever the user browses e-commerce information, and obtain the real-time browsing density based on the user's information browsing situation; the recommendation density adjustment unit is used to adjust subsequent information recommendations by comparing with the preset expected recommendation density.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. This application dynamically adjusts the density of information recommendations on e-commerce platforms and makes adaptive adjustments based on real-time user feedback, helping users maintain a good browsing experience while receiving e-commerce information recommendations, and avoiding the situation where excessive recommendations hinder the user's browsing process.

[0055] 2. The present invention divides the e-commerce information browsed by users and stratifies different types of e-commerce information to accurately understand users' browsing preferences. This facilitates the subsequent recommendation of e-commerce information that maintains information density while also recommending more suitable e-commerce information for users.

[0056] 3. The present invention captures user behaviors generated during the user browsing process, and identifies the user's preferences for various types of e-commerce information and the information density under normal circumstances through user behaviors, which helps to subsequently set the expected information density and effectively improve the user's browsing experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic diagram of the steps of an intelligent recommendation method for e-commerce information based on data analysis;

[0058] Figure 2 This is a structural diagram of an e-commerce information intelligent recommendation system based on data analysis. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] Example: Figures 1 to 2 As shown, the present invention provides an e-commerce information intelligent recommendation method based on data analysis, and the recommendation method includes the following steps:

[0061] Step S100: After the user's authorization, each e-commerce information browsing process of the user is collected to generate corresponding information browsing records; feature extraction is performed on the user behavior information and browsing information generated in any information browsing record;

[0062] Wherein, step S100 includes the following steps:

[0063] Step S101: During the e-commerce information browsing process, whenever a user is on a browsing page, the e-commerce information on the browsing page is collected to obtain a browsing information set. Simultaneously, the user behavior information generated during the browsing process is collected to obtain a behavior information set. The browsing information set and the behavior information set are aggregated to generate an information browsing record for the user.

[0064] Step S102: randomly selecting an information browsing record, reading a browsing information set in the information browsing record, randomly selecting a browsing information from the browsing information set, wherein the browsing information includes a plurality of text information and image information, performing keyword extraction on the plurality of text information and feature extraction on the plurality of image information, generating an information data set of the arbitrary browsing information, and obtaining a plurality of information data sets of the selected information browsing record;

[0065] Step S103: extracting a behavior information set from the selected information browsing record, and dividing each behavior information in the behavior information set into a behavior name or behavior data according to the data type, wherein the behavior data is a specific value or a value range; wherein any behavior name corresponds to one behavior data, and any behavior name is merged with the corresponding behavior data to generate a corresponding behavior data set, and obtain several behavior data sets of the selected information browsing record.

[0066] Step S200: Obtaining browsing information of the user in any information browsing history, determining the information category of each browsing information, analyzing the preferences of each information category; and stratifying each information category based on the user's preferences;

[0067] Step S200 includes the following steps:

[0068] Step S201: pre-constructing an information category database, wherein the information category database includes several information categories and several information feature sets, wherein one information category is matched with one information feature set;

[0069] Step S202: arbitrarily select an information browsing record, extract any browsing information from the selected information browsing record, compare each keyword and each feature in the information data set of the browsing information with the information features in the arbitrary information feature set, and count the number of identical features between the information data set and the information feature set; the information feature set with the largest number of identical features is defined as the target information feature set; if there is only one target information feature set, the information category corresponding to the target information feature set is defined as the information category of the browsing information; if there is more than one target information feature set, obtain the similarity between each difference feature between the arbitrary target information feature set and the information data set, calculate the average value to obtain the average similarity, and select the information category corresponding to the target information feature set with the largest average similarity as the information category of the browsing information;

[0070] Step S203: Obtain the information category of each browsing information in the information browsing record, and count the number of information of each information category; obtain the record generation time of each information browsing record, and sort them according to the record generation time, and obtain the number of information of the jth information category in the last information browsing record (N j ) last , sort the various information categories in the last information browsing record in ascending order according to the number of information, and get the position of the jth information category as A j ; Set the number of information of the jth information category in the i-th information browsing record (N j ) i , according to the formula:

[0071]

[0072] Where i is a positive integer and i∈(1,m), m is the total number of information categories; the number of comprehensive views of the j-th information category is calculated as X j ;

[0073] Example 1: Assume that the number of information of the jth information category in the user's historical information browsing records is 15, 20, 25, and 20 respectively, where the number of information of the jth information category in the last information browsing record is 20, and the order of the number in the information browsing record is 2, then the comprehensive browsing number X of the jth information category is obtained. j =(15+20+25+20) / 4+20 / 2=30;

[0074] Step S204: Extract the maximum number of messages N from all message browsing records max , preset an information quantity threshold N th , if N max >N th , then the information quantity range (Nth ,N max ) as the user's first preference level, and the information quantity range (0,N th ) is evenly divided into b preference levels; if N max <N th , then the information quantity range (0,N max ) is evenly divided into b+1 preference levels, and the preference levels corresponding to various information categories are obtained; if X j ∈(N th ,N max ), then the jth information category is set as the first preference level.

[0075] Step S300: For any information browsing record, correlation analysis is performed on the user behavior and browsing information in the information browsing record to obtain the user's information browsing density; based on the information browsing density presented in different information browsing records, an expected recommendation density is preset for the user;

[0076] Wherein, step S300 includes the following steps:

[0077] Step S301: randomly selecting an information browsing record, extracting several behavior data sets from the information browsing record, and presetting several density evaluation indicators, wherein the several density evaluation indicators include browsing time;

[0078] Step S302: arbitrarily select an information browsing record, extract several behavior data sets from the information browsing record, perform semantic similarity comparison between the behavior name in any behavior data set and any density evaluation index, match the behavior name with the greatest semantic similarity with the density evaluation index, and obtain the behavior data of the density evaluation index in each behavior data set; if the behavior data corresponding to the density evaluation index is a numerical value, then accumulate the behavior data of the density evaluation index in each behavior data set to obtain the behavior characteristic value z of the density evaluation index; if the behavior data corresponding to the density evaluation index is a numerical range, then perform a union operation on the behavior data in each behavior data set to obtain the numerical range (min, max) of the density evaluation index, and obtain the corresponding behavior characteristic value z = max-min;

[0079] Step S303: extract the information data set of each browsing information from the selected information browsing record, and count the number of keywords contained in all information data sets as n key and the number of features n aspect , the number of information features contained in the selected information browsing record is n behave =n key +n aspect ; According to the formula:

[0080]

[0081] Where k is a positive integer and k∈(1,d), d is the number of density evaluation indicators set, z k is the behavioral characteristic value of the kth density evaluation index, t is the browsing time; the information browsing density P of the user in the selected information browsing record is calculated;

[0082] Example 2: Assume that the number of information features contained in an information browsing record is 30, and the browsing duration is 10 seconds. There are two density evaluation indicators, namely the number of clicks and the number of page entries, which are 10 and 5 respectively. The information browsing density P is calculated to be 30 / (10×15)=0.2;

[0083] Step S304: Select the rth information browsing record, and accumulate the number of comprehensive browsings corresponding to each preference level in the rth information browsing record, wherein the total number of comprehensive browsings of the first preference level in the rth information browsing record is set as X1 r , according to the formula:

[0084]

[0085] Where r is a positive integer and r∈(1,u), u is the total number of information browsing records; the characteristic proportion coefficient δ of the rth information browsing record is calculated r ;

[0086] Step S305: Set the information browsing density of the rth information browsing record to P r , according to the formula:

[0087]

[0088] Calculate the user's expected recommendation density P ave ;

[0089] Example 3: Assume that there are 3 information browsing records, and the total number of comprehensive browsings at the first preference level in each information browsing record is 20, 10, and 20 respectively. Then, the characteristic proportion coefficients of the 3 information browsing records are 40%, 20%, and 40% respectively. The information browsing densities of the 3 information browsing records are 0.2, 0.4, and 0.3 respectively, and the user's expected recommendation density P is calculated. ave =(0.2×40%+0.4×20%+0.3×40%) / 3=(0.08+0.08+0.12) / 3=0.093.

[0090] Step S400: Whenever a user browses e-commerce information, a corresponding information recommendation ratio is assigned to each information type according to the information stratification, and a real-time browsing density is obtained based on the user's information browsing behavior. By comparing this with the preset expected recommendation density, subsequent information recommendations are adjusted;

[0091] Step S400 includes the following steps:

[0092] Step S401: Get the information categories contained in each current preference level, and set the number of information categories contained in the vth preference level to Q v ; Randomly select the jth information category and set the jth information category to be at the vth preference level, according to the formula:

[0093]

[0094] Where b+1 is the number of preference levels; the information recommendation ratio W of the jth information category is calculated j ;

[0095] Example 4: Assume that the jth information category is in the second preference level, and the second preference level contains three information categories; obtain that there are three levels of preference levels, and calculate the information recommendation ratio W of the jth information category j =(2 / 6)×1 / 3=1 / 9;

[0096] Step S402: Display e-commerce information corresponding to various information types according to the information recommendation ratio. After the user browses the information, the browsing information set and behavior information set browsed by the user are obtained after a time T. The number of information features n included in the browsing information set is counted. T And the total value of the behavioral characteristics of each density evaluation index z T , the user's real-time browsing density is calculated as P now =n T / (T×z T );

[0097] Step S403: Set the user's expected recommendation density as P ave , if P now <P ave , then when making the next e-commerce information recommendation, increase the recommendation ratio of the first preference level information; if P now >P ave , then continue to push e-commerce information according to the current information recommendation ratio.

[0098] An e-commerce information intelligent recommendation system, the recommendation system includes a historical information analysis module, a recommendation information classification module, a behavior density analysis module and a user willingness analysis module;

[0099] The historical information analysis module is used to collect data on each user's e-commerce information browsing process after user authorization and generate corresponding information browsing records; it also extracts features from user behavior information and browsing information generated in any information browsing record;

[0100] The recommended information classification module is used to obtain the browsing information of the user in any information browsing history, determine the information category of each browsing information, analyze the preferences of various information categories, and classify the information of various information categories based on the user's preferences;

[0101] The behavior density analysis module is used to analyze the correlation between user behavior and browsing information in any information browsing record to obtain the user's information browsing density; based on the information browsing density presented in different information browsing records, an expected recommendation density is preset for the user;

[0102] The user intention analysis module is used to assign corresponding information recommendation ratios to various information types according to the information stratification whenever a user browses e-commerce information, and obtain real-time browsing density based on the user's information browsing situation; by comparing with the preset expected recommendation density, subsequent information recommendations are adjusted.

[0103] Among them, the historical information analysis module includes a historical information collection unit and a behavior information extraction unit;

[0104] The historical information collection unit is used to collect each e-commerce information browsing process of the user after the user's authorization and generate corresponding information browsing records; the behavior information extraction unit is used to extract features of the user behavior information and browsing information generated in any information browsing record.

[0105] Among them, the recommended information division module includes an information browsing analysis unit and a recommended information stratification unit;

[0106] The information browsing analysis unit is used to obtain the user's browsing information in any information browsing record, determine the information category of each browsing information, and analyze the preferences of various information categories; the recommended information stratification unit is used to stratify various information categories based on the user's preferences.

[0107] Among them, the behavior density analysis module includes a user behavior analysis unit and an expected density setting unit;

[0108] The user behavior analysis unit is used to perform correlation analysis on the user behavior and browsing information in any information browsing record to obtain the user's information browsing density; the expected density setting unit is used to preset an expected recommendation density for the user based on the information browsing density presented in different information browsing records.

[0109] Among them, the user willingness analysis module includes a user willingness evaluation unit and a recommendation density adjustment unit;

[0110] The user willingness evaluation unit is used to assign corresponding information recommendation ratios to various information types according to the information stratification whenever the user browses e-commerce information, and obtain the real-time browsing density based on the user's information browsing situation; the recommendation density adjustment unit is used to adjust subsequent information recommendations by comparing with the preset expected recommendation density.

[0111] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for intelligently recommending e-commerce information based on data analysis, characterized by: The recommended method comprises the following steps: Step S100: After the user's authorization, each e-commerce information browsing process of the user is collected to generate corresponding information browsing records; feature extraction is performed on the user behavior information and browsing information generated in any information browsing record; Step S200: Obtaining browsing information of the user in any information browsing history, determining the information category of each browsing information, analyzing the preferences of each information category; and stratifying each information category based on the user's preferences; Step S300: For any information browsing record, correlation analysis is performed on the user behavior and browsing information in the information browsing record to obtain the user's information browsing density; based on the information browsing density presented in different information browsing records, an expected recommendation density is preset for the user; Step S400: Whenever a user browses e-commerce information, a corresponding information recommendation ratio is assigned to each information type according to the information stratification, and a real-time browsing density is obtained based on the user's information browsing situation; by comparing it with the preset expected recommendation density, subsequent information recommendations are adjusted.

2. The method for intelligently recommending e-commerce information based on data analysis according to claim 1, characterized in that: The step S100 includes the following steps: Step S101: During the e-commerce information browsing process, whenever a user is on a browsing page, the e-commerce information on the browsing page is collected to obtain a browsing information set. Simultaneously, the user behavior information generated during the browsing process is collected to obtain a behavior information set. The browsing information set and the behavior information set are aggregated to generate an information browsing record for the user. Step S102: randomly selecting an information browsing record, reading a browsing information set in the information browsing record, randomly selecting a browsing information from the browsing information set, wherein the browsing information includes a plurality of text information and image information, performing keyword extraction on the plurality of text information and feature extraction on the plurality of image information, generating an information data set of the arbitrary browsing information, and obtaining a plurality of information data sets of the selected information browsing record; Step S103: extracting a behavior information set from the selected information browsing record, and dividing each behavior information in the behavior information set into a behavior name or behavior data according to the data type, wherein the behavior data is a specific value or a value range; wherein any behavior name corresponds to one behavior data, and any behavior name is merged with the corresponding behavior data to generate a corresponding behavior data set, and obtain several behavior data sets of the selected information browsing record.

3. The method for intelligent e-commerce information recommendation based on data analysis according to claim 2, characterized in that: The step S200 includes the following steps: Step S201: pre-constructing an information category database, wherein the information category database includes several information categories and several information feature sets, wherein one information category is matched with one information feature set; Step S202: arbitrarily select an information browsing record, extract any browsing information from the selected information browsing record, compare each keyword and each feature in the information data set of the browsing information with the information features in the arbitrary information feature set, and count the number of identical features between the information data set and the information feature set; the information feature set with the largest number of identical features is defined as the target information feature set; if there is only one target information feature set, the information category corresponding to the target information feature set is defined as the information category of the browsing information; if there is more than one target information feature set, obtain the similarity between each difference feature between the arbitrary target information feature set and the information data set, calculate the average value to obtain the average similarity, and select the information category corresponding to the target information feature set with the largest average similarity as the information category of the browsing information; Step S203: Obtain the information category of each browsing information in the information browsing record, and count the number of information of each information category; obtain the record generation time of each information browsing record, and sort them according to the record generation time, and obtain the number of information of the jth information category in the last information browsing record (N j ) last , sort the various information categories in the last information browsing record in ascending order according to the number of information, and get the position of the jth information category as A j ; Set the number of information of the jth information category in the i-th information browsing record (N j ) i , according to the formula: Where i is a positive integer and i∈(1,m), m is the total number of information categories; the number of comprehensive views of the j-th information category is calculated as X j ; Step S204: Extract the maximum number of messages N from all message browsing records max , preset an information quantity threshold N th , if N max >N th , then the information quantity range (N th ,N max ) as the user's first preference level, and the information quantity range (0,N th ) is evenly divided into b preference levels; if N max <N th , then the information quantity range (0,N max ) is evenly divided into b+1 preference levels, and the preference levels corresponding to various information categories are obtained; if X j ∈(N th ,N max ), then the jth information category is set as the first preference level.

4. The method for intelligently recommending e-commerce information based on data analysis according to claim 3, characterized in that: The step S300 includes the following steps: Step S301: randomly selecting an information browsing record, extracting several behavior data sets from the information browsing record, and presetting several density evaluation indicators, wherein the several density evaluation indicators include browsing time; Step S302: arbitrarily select an information browsing record, extract several behavior data sets from the information browsing record, perform semantic similarity comparison between the behavior name in any behavior data set and any density evaluation index, match the behavior name with the greatest semantic similarity with the density evaluation index, and obtain the behavior data of the density evaluation index in each behavior data set; if the behavior data corresponding to the density evaluation index is a numerical value, then accumulate the behavior data of the density evaluation index in each behavior data set to obtain the behavior characteristic value z of the density evaluation index; if the behavior data corresponding to the density evaluation index is a numerical range, then perform a union operation on the behavior data in each behavior data set to obtain the numerical range (min, max) of the density evaluation index, and obtain the corresponding behavior characteristic value z = max-min; Step S303: extract the information data set of each browsing information from the selected information browsing record, and count the number of keywords contained in all information data sets as n key and the number of features n aspect , the number of information features contained in the selected information browsing record is n behave =n key +n aspect ; According to the formula: Where k is a positive integer and k∈(1,d), d is the number of density evaluation indicators set, z k is the behavioral characteristic value of the kth density evaluation index, t is the browsing time; the information browsing density P of the user in the selected information browsing record is calculated; Step S304: Select the rth information browsing record, and accumulate the number of comprehensive browsings corresponding to each preference level in the rth information browsing record, wherein the total number of comprehensive browsings of the first preference level in the rth information browsing record is set as X1 r , according to the formula: Where r is a positive integer and r∈(1,u), u is the total number of information browsing records; the characteristic proportion coefficient δ of the rth information browsing record is calculated r ; Step S305: Set the information browsing density of the rth information browsing record to P r , according to the formula: Calculate the user's expected recommendation density P ave .

5. The method for intelligently recommending e-commerce information based on data analysis according to claim 4, characterized in that: The step S400 includes the following steps: Step S401: Get the information categories contained in each current preference level, and set the number of information categories contained in the vth preference level to Q v ; Randomly select the jth information category and set the jth information category to be at the vth preference level, according to the formula: Where b+1 is the number of preference levels; the information recommendation ratio W of the jth information category is calculated j ; Step S402: Display e-commerce information corresponding to various information types according to the information recommendation ratio. After the user browses the information, the browsing information set and behavior information set browsed by the user are obtained after a time T. The number of information features n included in the browsing information set is counted. T And the total value of the behavioral characteristics of each density evaluation index z T , the user's real-time browsing density is calculated as P now =n T / (T×z T ); Step S403: Set the user's expected recommendation density as P ave , if P now <P ave , then when making the next e-commerce information recommendation, increase the recommendation ratio of the first preference level information; if P now >P ave , then continue to push e-commerce information according to the current information recommendation ratio.

6. An e-commerce information intelligent recommendation system, configured to execute the e-commerce information intelligent recommendation method based on data analysis according to any one of claims 1 to 5, characterized in that: The recommendation system includes a historical information analysis module, a recommendation information division module, a behavior density analysis module and a user willingness analysis module; The historical information analysis module is used to collect each e-commerce information browsing process of the user after authorization by the user, generate corresponding information browsing records; and extract features of user behavior information and browsing information generated in any information browsing record; The recommended information classification module is used to obtain the browsing information of the user in any information browsing history, determine the information category of each browsing information, analyze the preferences of various information categories, and classify the various information categories based on the user's preferences; The behavior density analysis module is used to perform correlation analysis on user behavior and browsing information in any information browsing record to obtain the user's information browsing density; based on the information browsing density presented in different information browsing records, an expected recommendation density is preset for the user; The user intention analysis module is used to assign corresponding information recommendation ratios to various information types according to the information stratification whenever a user browses e-commerce information, and obtain real-time browsing density based on the user's information browsing situation; by comparing with the preset expected recommendation density, subsequent information recommendations are adjusted.

7. The e-commerce information intelligent recommendation system according to claim 6, characterized in that: The historical information analysis module includes a historical information collection unit and a behavior information extraction unit; The historical information collection unit is used to collect each e-commerce information browsing process of the user after the user's authorization and generate corresponding information browsing records; the behavior information extraction unit is used to extract features of the user behavior information and browsing information generated in any information browsing record.

8. The e-commerce information intelligent recommendation system according to claim 6, characterized in that: The recommended information division module includes an information browsing analysis unit and a recommended information stratification unit; The information browsing analysis unit is used to obtain the user's browsing information in any information browsing record, determine the information category of each browsing information, and analyze the preferences of various information categories; the recommended information stratification unit is used to stratify various information categories based on the user's preferences.

9. The e-commerce information intelligent recommendation system according to claim 6, characterized in that: The behavior density analysis module includes a user behavior analysis unit and an expected density setting unit; The user behavior analysis unit is used to perform correlation analysis on the user behavior and browsing information in any information browsing record to obtain the user's information browsing density; the expected density setting unit is used to preset an expected recommendation density for the user based on the information browsing density presented in different information browsing records.

10. The e-commerce information intelligent recommendation system according to claim 6, characterized in that: The user willingness analysis module includes a user willingness evaluation unit and a recommendation density adjustment unit; The user willingness evaluation unit is used to assign corresponding information recommendation ratios to various information types according to the information stratification situation whenever the user browses e-commerce information, and obtain real-time browsing density based on the user's information browsing situation; the recommendation density adjustment unit is used to adjust subsequent information recommendations by comparing with the preset expected recommendation density.

Images