CN116894709A - Advertisement commodity recommendation method and system - Google Patents
Advertisement commodity recommendation method and system Download PDFInfo
- Publication number
- CN116894709A CN116894709A CN202310872267.9A CN202310872267A CN116894709A CN 116894709 A CN116894709 A CN 116894709A CN 202310872267 A CN202310872267 A CN 202310872267A CN 116894709 A CN116894709 A CN 116894709A
- Authority
- CN
- China
- Prior art keywords
- user
- information
- recommendation
- commodity
- characteristic information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000006399 behavior Effects 0.000 claims abstract description 143
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 79
- 238000010801 machine learning Methods 0.000 claims abstract description 16
- 238000013135 deep learning Methods 0.000 claims abstract description 11
- 238000001914 filtration Methods 0.000 claims description 43
- 238000005457 optimization Methods 0.000 claims description 37
- 239000013598 vector Substances 0.000 claims description 26
- 238000004458 analytical method Methods 0.000 claims description 18
- 238000012544 monitoring process Methods 0.000 claims description 17
- 238000011156 evaluation Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 16
- 238000004891 communication Methods 0.000 claims description 7
- 238000004140 cleaning Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 238000007637 random forest analysis Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 230000003993 interaction Effects 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000007405 data analysis Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 description 12
- 238000006243 chemical reaction Methods 0.000 description 9
- 230000009286 beneficial effect Effects 0.000 description 4
- 235000011888 snacks Nutrition 0.000 description 3
- 241001122767 Theaceae Species 0.000 description 2
- 235000021152 breakfast Nutrition 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 235000020965 cold beverage Nutrition 0.000 description 2
- 238000010921 in-depth analysis Methods 0.000 description 2
- 239000008267 milk Substances 0.000 description 2
- 210000004080 milk Anatomy 0.000 description 2
- 235000013336 milk Nutrition 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000002537 cosmetic Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000012482 interaction analysis Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000037072 sun protection Effects 0.000 description 1
- 230000000475 sunscreen effect Effects 0.000 description 1
- 239000000516 sunscreening agent Substances 0.000 description 1
- 230000009182 swimming Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
Abstract
The invention provides a method and a system for recommending advertisement commodities. The method comprises the following steps: obtaining operation logs and commodity characteristic information of a user from a server background; analyzing first behavior data of the user according to the acquired operation log of the user, and acquiring first characteristic information of the user; dividing an operation log of a user into different log time according to different time nodes; combining the log time with the first behavior data of the user to obtain second behavior data of the user with different time nodes; acquiring second characteristic information of the user according to the second behavior data; and establishing a user image through a machine learning algorithm according to the second characteristic information, extracting and encoding the commodity characteristic information through a deep learning algorithm, and constructing a commodity characteristic model, wherein the commodity characteristic model comprises semantic information and implicit characteristics of the commodity.
Description
Technical Field
The invention provides a method and a system for recommending advertisement commodities, and belongs to the technical field of advertisement recommendation.
Background
With the development and popularization of the internet, shopping modes of people have also changed greatly. Traditional off-line shopping is gradually shifted to on-line, and people rely more on internet platforms to acquire commodity information and purchase commodities. In this process, the role of the advertisement recommendation system becomes more and more important.
The advertisement commodity recommendation system is a system for recommending proper commodities to users through algorithms and technical means based on interest and behavior data of the users. According to personal preference and purchase history of the user, the method and the device can accurately recommend commodities interested by the user, and improve shopping experience and satisfaction of the user.
Disclosure of Invention
The invention provides an advertisement commodity recommending method and system, which are used for solving the problem that the existing advertisement commodity recommending method cannot accurately recommend:
the invention provides an advertisement commodity recommending method, which comprises the following steps:
s1: obtaining operation logs and commodity characteristic information of a user from a server background;
s2: analyzing first behavior data of the user according to the acquired operation log of the user, and acquiring first characteristic information of the user;
s3: dividing an operation log of a user into different log time according to different time nodes;
s4: combining the log time with the first behavior data of the user to obtain second behavior data of the user with different time nodes; acquiring second characteristic information of the user according to the second behavior data;
s5: establishing a user image through a machine learning algorithm according to the second characteristic information, extracting and encoding the commodity characteristic information through a deep learning algorithm, and constructing a commodity characteristic model, wherein the commodity characteristic model comprises semantic information and implicit characteristics of a commodity;
S6: based on the user portrait and the commodity feature model, recommending advertisement commodities to a user by adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to obtain first recommendation information;
s7: according to feedback of a user on the first recommendation information, first feedback information is obtained, the first feedback information is combined with the second characteristic information, user portraits are optimized, and a first optimization model is obtained;
s8: continuously adjusting the first recommendation information according to the first optimization model to obtain second recommendation information, monitoring feedback of the user on the second recommendation information according to the first monitoring module to obtain second feedback information, obtaining third behavior data of the user according to the second feedback information and the first optimization model, obtaining third characteristic information of the user according to the third behavior data of the user, and continuously adjusting recommendation strategies according to the third characteristic information of the user.
Further, the first behavior data comprises commodity browsing, searching, shopping cart adding, ordering, payment, evaluation and communication information with merchants; the first characteristic information includes shopping preferences, interests, and purchasing behavior of the user.
Further, the specific step of analyzing the first behavior data of the user according to the obtained operation log of the user to obtain the first feature information of the user includes:
cleaning an operation log of a user, removing invalid or error records, and processing missing values and abnormal values;
classifying the behavior types of the user according to the first behavior data of the user in the operation log;
extracting first characteristic information according to the classified behavior types;
based on the extracted first characteristic information, carrying out shopping preference analysis on the user; the shopping preference analysis comprises counting commodity categories, brands and price ranges of user preferences;
deducing the interests and hobbies of the user through the search keywords of the user and the browsed commodity information;
carrying out shopping behavior analysis through the purchasing behavior data of the user; the shopping behavior analysis comprises statistics of purchase frequency, guest price and purchase time period of the user;
if the operation log contains communication information with the merchant, the evaluation, complaint and consultation condition of the user on the goods and/or services are obtained by analyzing the interaction behavior of the user and the merchant.
Further, the log time includes a first log time, a second log time, and a third log time; the first log time is divided into a first season time, a second season time, a third season time and a fourth season time according to seasons; the second log time is divided into working day time and rest day time; the third log time is divided into a first stage time, a second stage time, a third stage time and a fourth stage time;
Further, the method comprises the steps of,
the second behavior data includes first behavior data of a user at the different log times;
the second characteristic information includes second behavior data of the user at the different log times.
Further, the specific step of recommending advertisement goods to the user based on the user portrait and the commodity feature model by adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm, and the specific step of obtaining the first recommendation information comprises the following steps:
the second characteristic information of the user is subjected to numerical representation to form a first user characteristic vector;
carrying out numerical processing on the commodity characteristic information to form characteristic vectors of commodities;
matching the first user feature vector with the feature vector of the commodity through a content recommendation algorithm, and calculating the matching degree between the user and the commodity;
sorting according to the matching degree, and recommending advertisement commodities related to the user interests;
carrying out commodity recommendation on the user through a collaborative filtering algorithm;
the collaborative filtering algorithm comprises collaborative filtering based on users and collaborative filtering based on commodities;
the collaborative filtering based on the users comprises the steps of calculating the similarity between the users, finding other users similar to the current user, obtaining advertisement commodities liked by the similar users, and taking the advertisement commodities as recommendation options;
The collaborative filtering based on the commodities comprises the steps of calculating the similarity between advertisement commodities, finding out other commodities similar to the commodities historically liked by the user, and taking the other commodities as recommendation candidates;
comprehensively sequencing recommendation results of the collaborative filtering algorithm and the content recommendation algorithm through recommendation accuracy and individuation degree;
obtaining first recommendation information according to the sorting result, determining a final recommendation list according to the first recommendation information, and filtering out commodities which do not accord with user preference or recommendation strategies;
the ranked and filtered list of recommendations is presented to the user.
Further, according to feedback of the user on the first recommendation information, first feedback information is obtained, the first feedback information is combined with the second characteristic information, user portraits are optimized, and a first optimization model is obtained; comprising the following steps:
when the user interacts and/or gives feedback to the first recommendation information, recording feedback information of the user, wherein the feedback information comprises preference, satisfaction, clicking behavior and scoring of the user;
extracting information related to the recommended commodity or the recommended result from the user feedback information;
carrying out data fusion on the first feedback information and second characteristic information of the user to obtain a second user characteristic vector;
Optimizing the user image according to a machine learning algorithm through the second user feature vector, and training a model according to a random forest algorithm;
and acquiring a first optimization model according to the optimized portrait and the trained model.
The invention provides an advertisement commodity recommendation system, which comprises:
an information acquisition module: obtaining operation logs and commodity characteristic information of a user from a server background;
and a data analysis module: analyzing first behavior data of the user according to the acquired operation log of the user, and acquiring first characteristic information of the user;
and a log classification module: dividing an operation log of a user into different log time according to different time nodes;
and a data combining module: combining the log time with the first behavior data of the user to obtain second behavior data of the user with different time nodes; acquiring second characteristic information of the user according to the second behavior data;
model construction module: establishing a user image through a machine learning algorithm according to the second characteristic information, extracting and encoding the commodity characteristic information through a deep learning algorithm, and constructing a commodity characteristic model, wherein the commodity characteristic model comprises semantic information and implicit characteristics of a commodity;
And a commodity recommendation module: based on the user portrait and the commodity feature model, recommending advertisement commodities to a user by adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to obtain first recommendation information;
and an optimization module: according to feedback of a user on the first recommendation information, first feedback information is obtained, the first feedback information is combined with the second characteristic information, user portraits are optimized, and a first optimization model is obtained;
and a recommendation adjustment module: continuously adjusting the first recommendation information according to the first optimization model to obtain second recommendation information, monitoring feedback of the user on the second recommendation information according to the first monitoring module to obtain second feedback information, obtaining third behavior data of the user according to the second feedback information and the first optimization model, obtaining third characteristic information of the user according to the third behavior data of the user, and continuously adjusting recommendation strategies according to the third characteristic information of the user.
The invention provides an electronic device, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the memory, wherein the processor executes the program to realize any one of the advertisement commodity recommending methods.
The invention provides a non-transitory computer readable storage medium having stored thereon a computer program for execution by a processor to implement any of the above-described advertisement article recommendation methods.
The invention has the beneficial effects that: according to the invention, personalized commodities can be recommended to the user according to the interest and behavior data of the user; the shopping experience of the user can be improved, and the user can find interesting commodities more easily, so that the satisfaction degree of the user is improved; through accurate commodity recommendation, the method and the device can increase clicking and purchasing will of the user on the commodity, so that purchasing conversion rate is improved. When the user sees the commodity recommendation of interest, the user is more easily attracted and purchases; according to the invention, more accurate advertisement delivery can be provided for merchants according to the interest and behavior data of the users; the click rate and conversion rate of advertisements can be improved, and the advertising effect and return on investment of merchants are improved; according to the invention, more accurate advertisement delivery can be provided for merchants according to the interest and behavior data of the users; the advertising waste can be reduced, the advertising effect can be improved, and the advertising cost can be saved. Meanwhile, the method and the system can analyze the information such as browsing records of different time periods in different seasons of the user by acquiring the log records of different time periods of the user, can more accurately know the browsing preference data of different time nodes of the user, and can make recommendation more accurate according to the data.
Drawings
FIG. 1 is a step diagram of an advertisement recommendation method according to the present application.
Detailed Description
In order that the above-recited objects, features and advantages of the present application will be more clearly understood, a more particular description of the application will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. It should be noted that, without conflict, the embodiments of the present application and features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application, and the described embodiments are merely some, rather than all, embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the description of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.
In one embodiment of the application, a method for recommending advertisement goods comprises the following steps:
S1: obtaining operation logs and commodity characteristic information of a user from a server background;
s2: analyzing first behavior data of the user according to the acquired operation log of the user, and acquiring first characteristic information of the user;
s3: dividing an operation log of a user into different log time according to different time nodes;
s4: combining the log time with the first behavior data of the user to obtain second behavior data of the user with different time nodes; acquiring second characteristic information of the user according to the second behavior data;
s5: establishing a user image through a machine learning algorithm according to the second characteristic information, extracting and encoding the commodity characteristic information through a deep learning algorithm, and constructing a commodity characteristic model, wherein the commodity characteristic model comprises semantic information and implicit characteristics of a commodity; the semantic information and implicit features of the commodity feature model comprise: the method comprises the following steps of (1) carrying out text description of the commodity, category labels of the commodity, picture characteristics of the commodity, statistical information such as price and sales volume of the commodity, user evaluation and comment of the commodity, historical purchasing record of the commodity and related information of the commodity;
s6: based on the user portrait and the commodity feature model, recommending advertisement commodities to a user by adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to obtain first recommendation information;
S7: according to feedback of a user on the first recommendation information, first feedback information is obtained, the first feedback information is combined with the second characteristic information, user portraits are optimized, and a first optimization model is obtained;
s8: continuously adjusting the first recommendation information according to the first optimization model to obtain second recommendation information, monitoring feedback of the user on the second recommendation information according to the first monitoring module to obtain second feedback information, obtaining third behavior data of the user according to the second feedback information and the first optimization model, obtaining third characteristic information of the user according to the third behavior data of the user, and continuously adjusting recommendation strategies according to the third characteristic information of the user.
The first row of data comprises commodity browsing, searching, shopping cart adding, ordering, payment, evaluation and communication information with merchants; the first characteristic information includes shopping preferences, interests, and purchasing behavior of the user.
The working principle of the technical scheme is as follows: and acquiring the operation log and commodity characteristic information of the user from the server background. The operation log of the user comprises behavior operations of the user on the platform, such as commodity browsing, searching, shopping cart adding, ordering, payment, evaluation and the like, and commodity characteristic information comprises text description, category labels, picture characteristics, prices, sales statistics, user evaluation and comment, historical purchase records, associated information and the like of the commodity; and analyzing the operation log of the user, extracting first behavior data of the user, and analyzing and processing the first behavior data to obtain first characteristic information of the user. Such characteristic information may include shopping preferences, interests, purchasing behavior, etc. of the user; and dividing the operation log of the user according to different time nodes to form different log time periods. Thus, the behavior of the user can be better analyzed and speculated, and more accurate recommendation is performed; combining the user behavior data of different time nodes with the first characteristic information to obtain second behavior data of the user, analyzing and processing according to the second behavior data, extracting second characteristic information of the user, and further optimizing the user portrait; for example, the user likes to search for living goods, breakfast food materials and the like in the morning, the user likes to search for some movie tickets and the like in the noon, and perfectly likes to search for cosmetics and the like, so that shopping preferences, interest and purchase behaviors and the like of the user in the morning and evening can be deduced, for example, the user likes to search for some play places in the rest days, the user likes to search for some content related to work, such as office supplies and the like in the working days, so that shopping preferences, interest and purchase behaviors and the like of the user in the rest days and the working days can be deduced, for example, the user likes to search for sun protection products in the summer and likes to search for skin protection products in the winter, so that shopping preferences, interest and purchase behaviors and the like of the user in different seasons can be deduced. And establishing a user portrait by using a machine learning algorithm, and integrating and analyzing the characteristic information of the user to form a user portrait model. And meanwhile, extracting and encoding commodity characteristic information by using a deep learning algorithm to construct a commodity characteristic model. The commodity feature model comprises semantic information and implicit features of commodities, such as text description, category labels, picture features, prices, sales statistics, user evaluation and comment, historical purchase records, associated information and the like; based on the user portrait and commodity feature model, adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to recommend advertisement commodity to the user. Recommending the most relevant and attractive advertisement goods according to the interests and preferences of the user, and generating first recommendation information, for example, the user prefers to search clothes in the evening, and the user is preferentially pushed relevant advertisements every morning; according to the feedback of the user on the first recommendation information, the first feedback information is obtained, for example, the user is recommended that the clothing is related at night, but the user does not see the clothing at any point and stays on the page, so that the user is inferred to be uninteresting about the clothing temporarily, but the user can search for skin care products, and the like, and the user can be inferred to like searching for skin care products at night. And combining the first feedback information with the second characteristic information, optimizing the user portrait, and generating a first optimization model, wherein the first optimization model is to replace the commodity which is not interested by the user with the commodity which is interested by the user for recommendation. Continuously adjusting the first recommendation information by using the first optimization model to generate second recommendation information, wherein the second recommendation information is a commodity which is obtained after monitoring the user behavior and is of the latest interest to the user; and monitoring feedback of the user on the second recommendation information, and obtaining the second feedback information, for example, recommending skin care products to the user according to the user behavior, clicking the user to browse, indicating that the user is interested, and giving the feedback information is accurate in recommendation. According to the second feedback information and the first optimization model, third behavior data of the user are obtained, third characteristic information of the user is extracted according to the third behavior data, the third behavior data are changed according to the interests, the shopping behaviors and the like of different users in time, and the second characteristic information of the user is updated continuously to obtain the third characteristic information by continuously adjusting according to the user behaviors. And further optimizing a recommendation strategy according to the third characteristic information of the user, and continuously improving the accuracy and individuation degree of advertisement commodity recommendation.
The technical scheme has the effects that: and (3) establishing a user portrait and commodity characteristic model by analyzing and processing the operation log and commodity characteristic information of the user. Based on the user portrait and commodity feature model, a collaborative filtering algorithm and a content recommendation algorithm are combined, so that accurate advertisement commodity recommendation can be realized, and the most relevant and attractive advertisement commodity is provided for users; personalized recommendation strategies are customized by in-depth analysis of user interests, preferences and purchasing behavior. The user can obtain advertisement commodity recommendation meeting the interests and requirements of the user, so that the user experience and participation are improved, and the click rate and conversion rate of the user on advertisements are increased; accurate advertisement commodity recommendation can improve advertisement click rate and conversion rate, improves brand's awareness and sales volume. Meanwhile, the advertisement effect and the commercial value can be further improved by optimizing the recommendation strategy and continuously adjusting; and continuously monitoring and optimizing the recommendation strategy by combining the feedback of the user on the recommendation information. According to feedback information of the user and the first optimization model, recommended content can be dynamically adjusted, and timeliness and accuracy of advertisement commodity recommendation are ensured; the machine learning algorithm and the deep learning algorithm are used for processing and analyzing the user and commodity data, so that the characteristic information can be automatically extracted, and the user portrait and commodity characteristic model can be established. Thus, the utilization efficiency of resources can be greatly improved, and the manual intervention and the cost are reduced. Through dividing the operation log of the user into different log time according to different time nodes, the content such as browsing preference of the user in different time can be obtained, for example, the user likes browsing cold drink in noon in summer and browsing milk tea in noon in winter, and more accurate recommendation can be realized.
In one embodiment of the present invention, the specific step of analyzing the first row data of the user according to the obtained operation log of the user to obtain the first feature information of the user includes:
cleaning an operation log of a user, removing invalid or error records, and processing missing values and abnormal values;
classifying the behavior types of the user according to the first behavior data of the user in the operation log;
extracting first characteristic information according to the classified behavior types; for example, for merchandise browsing behavior, the merchandise category, brand preference, browsing duration, etc. browsed by the user may be extracted; for search behavior, search keywords, search time, etc. of the user can be extracted; for the order, the commodity type, the order quantity, the order time and the like of the order of the user can be extracted.
Based on the extracted first characteristic information, carrying out shopping preference analysis on the user; the shopping preference analysis comprises counting commodity categories, brands and price ranges of user preferences;
deducing the interests and hobbies of the user through the search keywords of the user and the browsed commodity information; for example, if a user frequently searches for content related to exercise equipment, it may indicate that the user is interested in exercise and sports; if the user frequently searches or browses books, the user is shown to be interested in reading the learning comparison.
Carrying out shopping behavior analysis through the purchasing behavior data of the user; the shopping behavior analysis comprises statistics of purchase frequency, guest price and purchase time period of the user;
if the operation log contains communication information with the merchant, the evaluation, complaint and consultation condition of the user on the goods and/or services are obtained by analyzing the interaction behavior of the user and the merchant.
The working principle of the technical scheme is as follows: cleaning the user operation log through a data preprocessing technology and a cleaning rule, removing invalid or error records, and processing missing values and abnormal values; and classifying the behavior types of the user according to the first behavior data in the operation log. For example, different behavior types may be divided according to whether the first behavior of the user is browsing, searching or placing a list; and extracting corresponding first characteristic information according to the classified behavior types. For example, for merchandise browsing behavior, merchandise categories, brand preferences, browsing durations, etc. that the user browses may be extracted; for the search behavior, a search keyword, a search time, etc. of the user may be extracted. This can be achieved through feature engineering techniques and model training; and carrying out shopping preference analysis on the user based on the extracted first characteristic information. This includes counting the categories of goods, brands, and price ranges that the user prefers, etc.; and deducing the interests and hobbies of the user according to the search keywords of the user and the browsed commodity information. For example, by analyzing user preferences to search for and browse fitness equipment related content, it may be inferred that the user is interested in fitness and sports; if the user searches or browses books frequently, the user is interested in reading and learning comparison; and carrying out shopping behavior analysis through the shopping behavior data of the user. This includes counting the purchase frequency of the user, the price of the customer, the time period of purchase, etc. By analyzing the shopping behaviors of the user, the consumption habit and the purchasing preference of the user can be known; if the operation log contains communication information with the merchant, the evaluation, complaint and consultation condition of the user on the goods and/or services can be obtained by analyzing the interaction behavior of the user and the merchant. This may help merchants improve products and services and provide a better user experience.
The technical scheme has the effects that: invalid or erroneous records can be removed through data cleaning and processing, so that the interference to subsequent analysis is reduced, and the data quality is improved; through classification and feature extraction of user behaviors, shopping preference and interest of the user can be known more accurately, and basic data support is provided for personalized recommendation and marketing; the shopping preference analysis can help merchants to better know the preference of users on commodity categories, brands and price ranges, so that the product positioning, promotion strategies and inventory management are optimized; the interest inference can deeply gain insight into the interest and hobbies of the user, provides basis for accurate advertisement putting and personalized recommendation, and improves the click rate and conversion rate of the advertisements; shopping behavior analysis can reveal information such as buying habits, buying power and buying time period of users, and is beneficial to formulating more effective sales promotion and marketing strategies; through user interaction analysis, the evaluation and feedback of the user on goods and services can be known, the merchant is helped to improve the product quality and the service quality, and the satisfaction degree and the loyalty degree of the user are improved.
In one embodiment of the present invention, the log time includes a first log time, a second log time, and a third log time; the first log time is divided into a first season time, a second season time, a third season time and a fourth season time according to seasons; the second log time is divided into working day time and rest day time; the third log time is divided into a first stage time, a second stage time, a third stage time and a fourth stage time; the first season time is 3-5 months, the second season time is 6-8 months, the third season time is 9-11 months, and the fourth season time is 12-2 months; the first stage time is 6-12 points, the second stage time is 12-18 points, the third stage time is 18-24 points, and the third stage time is 24-6 points.
The working principle and the effect of the technical scheme are as follows: the user's behavior and interest changes in different seasons can be better understood by classifying according to the time of different seasons. For example, summer users may be interested in sunscreens and swimming articles, and winter users may be interested in thermal garments and ski equipment. By classifying according to seasons, commodities and advertisements suitable for users can be recommended more accurately; the user's behavior and buying habits in different time periods can be better understood by classifying according to working day and rest day time. For example, a workday user may be more concerned with work and learning related merchandise, while a workday user may be more concerned with recreational and home life related merchandise. By classifying according to workdays and rest days, commodities and advertisements suitable for users can be more accurately recommended; the user's behavior and demand changes in different time periods can be better understood by classifying according to different time periods. For example, a morning user may be more concerned with breakfast and health related merchandise, a afternoon user may be more concerned with lunch and work related merchandise, and a evening user may be more concerned with dinner and leisure entertainment related merchandise. By categorizing according to different time periods, goods and advertisements suitable for the user can be more accurately recommended.
In one embodiment of the present invention,
the second behavior data includes first behavior data of a user at the different log times;
the second characteristic information includes second behavior data of the user at the different log times.
The working principle and the effect of the technical scheme are as follows: by acquiring the first behavior data of the user at different log times, the behavior characteristics of the user can be more comprehensively known. The method is beneficial to deep analysis of the behavior preference and shopping habit of the user in different time periods by merchants, and provides more accurate user portraits and market insight; by extracting the second behavior data of the user at different log times, the user's hobbies can be inferred more accurately. The method is beneficial to personalized recommendation and customized marketing of merchants, provides products and services of interest to users, and improves user satisfaction and purchase conversion rate. By comprehensively analyzing the behavior data and the characteristic information of the user in different seasons and different time periods, the enterprise can make more refined decisions. For example, target user groups are precisely positioned and targeted marketing is performed according to shopping preferences and interests of users; optimizing time schedule of goods loading and goods returning according to time preference of users; by deeply mining user behaviors and interests, enterprises can provide products and services which more meet user requirements, and user experience and satisfaction are improved. This helps to enhance the user's loyalty to the enterprise, facilitating the user's repurchase and public praise dissemination.
In one embodiment of the present invention, based on the user portrait and the commodity feature model, the specific steps of obtaining the first recommendation information include:
the second characteristic information of the user is subjected to numerical representation to form a first user characteristic vector;
carrying out numerical processing on the commodity characteristic information to form characteristic vectors of commodities;
matching the first user feature vector with the feature vector of the commodity through a content recommendation algorithm, and calculating the matching degree between the user and the commodity;
sorting according to the matching degree, and recommending advertisement commodities related to the user interests;
carrying out commodity recommendation on the user through a collaborative filtering algorithm;
the collaborative filtering algorithm comprises collaborative filtering based on users and collaborative filtering based on commodities;
the collaborative filtering based on the users comprises the steps of calculating the similarity between the users, finding other users similar to the current user, obtaining advertisement commodities liked by the similar users, and taking the advertisement commodities as recommendation options;
the collaborative filtering based on the commodities comprises the steps of calculating the similarity between advertisement commodities, finding out other commodities similar to the commodities historically liked by the user, and taking the other commodities as recommendation candidates;
Wherein, calculate the similarity between the commercial products by following;
wherein Sim (α, β) represents the similarity of commodity α and commodity β; f (F) α,β Indicating the use of both the evaluated commodity alpha and the evaluated commodity betaUser set, Fα, F β The user sets are respectively evaluated for commodity alpha and commodity beta,P fβ representing the score of user f on goods alpha, beta, respectively,/->The average value of scores of commodity alpha and beta is expressed, and the average value of scores is the ratio of the sum of all scores of the commodity to the number of users scoring the commodity.
Comprehensively sequencing recommendation results of the collaborative filtering algorithm and the content recommendation algorithm through recommendation accuracy and individuation degree;
obtaining first recommendation information according to the sorting result, determining a final recommendation list according to the first recommendation information, and filtering out commodities which do not accord with user preference or recommendation strategies;
the ranked and filtered list of recommendations is presented to the user.
The working principle of the technical scheme is as follows: converting the second characteristic information of the user into a numerical value representation to form a characteristic vector of the user; carrying out numerical treatment on characteristic information of the commodity to form a characteristic vector of the commodity; calculating the matching degree between the user feature vector and the commodity feature vector by using a content recommendation algorithm, and measuring the interest degree of the user on each commodity; sorting the commodities according to the matching degree, and recommending the advertisement commodity with the highest correlation degree with the user interest; and recommending commodities by using a collaborative filtering algorithm. And calculating the similarity between the users based on collaborative filtering of the users, finding other users similar to the current user, and acquiring advertisement commodities liked by the similar users as recommendation options. Calculating the similarity between advertisement commodities based on collaborative filtering of commodities, and finding out other commodities similar to the commodities historically liked by the user as recommendation candidates; and integrating recommendation results of the collaborative filtering algorithm and the content recommendation algorithm, and sequencing according to recommendation accuracy and individuation degree. Then, determining first recommendation information according to the sorting result, and filtering out non-satisfactory commodities according to the preference and recommendation policy of the user; the ranked and filtered list of recommendations is presented to the user for selection and purchase by the user. The above formula, for example, a college likes electronic products, books and snacks, B college likes electronic products, books and snacks, and the types of the electronic products, books and snacks liked by a college and B college are similar, so that shopping preference, interest and purchase behavior of the a college and B college can be inferred, and at the same time, a college likes to be served, B college may like to be served, and B college may be recommended to be served, and if B college is clicked, B college is inferred to be served.
The technical scheme has the effects that: through numerical representation and matching of the user characteristics and the commodity characteristics, personalized recommendation aiming at each user can be realized, and shopping satisfaction and experience of the user are improved; by combining a content recommendation algorithm and a collaborative filtering algorithm, interests of users and preference of similar users and similar commodities are comprehensively considered, and recommendation accuracy and hit rate can be improved; recommending according to the preference and the historical behavior of the user, increasing the viscosity and loyalty of the user to brands and platforms, and promoting continuous consumption; through accurate personalized recommendation, the purchasing desire and purchasing willingness of a user can be increased, the sales and conversion rate is improved, and the income of merchants is increased; filtering and sorting are carried out according to the preference of the user, advertisement commodities which are most relevant and meet the demands of the user are recommended to the user, the user is helped to reduce information overload from huge commodity selection, and shopping efficiency is improved; through continuous data accumulation and analysis, the recommendation algorithm can be optimized and improved, the recommendation effect and precision are improved, and the method is suitable for the change of user interests and the evolution of demands. Meanwhile, calculating the similarity between the advertised goods through the above formula can help to determine the relevance and correlation between the goods. When a user evaluates the commodity alpha, the commodity beta with higher similarity with the commodity alpha can be found according to the similarity calculation. In this way, the merchandise β may be used as an alternative to the user's interests in making merchandise recommendations to the user, thereby providing more personalized recommendations. By comparing the similarity between the commodity evaluated by the user and other commodities, the commodity which meets the taste of the user can be recommended. Therefore, the user can feel that the recommendation result is more fit with the requirements of the user, and the user satisfaction and the use experience are improved; by calculating the similarity between commodities, the advertising platform can know the association degree between commodities. Therefore, merchants can select commodities with higher similarity with target commodities as display objects of advertisements when putting advertisements, the exposure opportunity of the advertisements is increased, and the attention of potential users is better attracted; when the user evaluates the commodity alpha, the system can recommend other commodities related to the commodity alpha to the user according to the similarity calculation result. In this way, the user may be interested in purchasing other related goods, thereby increasing sales and customer value.
According to one embodiment of the invention, first feedback information is obtained according to feedback of a user on first recommendation information, the first feedback information is combined with the second characteristic information, and a user portrait is optimized to obtain a first optimization model; comprising the following steps:
when the user interacts and/or gives feedback to the first recommendation information, recording feedback information of the user, wherein the feedback information comprises preference, satisfaction, clicking behavior and scoring of the user;
extracting information related to the recommended commodity or the recommended result from the user feedback information; for example, if a user purchases a certain recommended commodity, this purchase behavior may be regarded as first feedback information; if the user gives scores or comments to the recommended items, these scores and comments may be used as the first feedback information.
Carrying out data fusion on the first feedback information and second characteristic information of the user to obtain a second user characteristic vector;
optimizing the user image according to a machine learning algorithm through the second user feature vector, and training a model according to a random forest algorithm;
and acquiring a first optimization model according to the optimized portrait and the trained model.
The working principle of the technical scheme is as follows: when the user interacts with the first recommendation information, such as clicking, purchasing, scoring, commenting and other actions, the system records feedback information of the user; the system extracts information related to recommended commodities or results from the feedback information of the user, such as which recommended commodities the user purchases, scores and comments given by the user and the like; and carrying out data fusion on the second characteristic information and the feedback information of the user so as to obtain a more comprehensive user characteristic vector. Thus, the static characteristics and dynamic feedback of the user can be comprehensively considered, and the interests and the preferences of the user can be more accurately described; the user representation is optimized by a machine learning algorithm, such as a random forest algorithm. The second user feature vector is used as input data, a model is trained to predict interests and behaviors of a user, and user portraits are further improved; and obtaining a first optimization model according to the optimized user portrait and the model obtained by training. The model can be used for predicting the preference and the demand of the user more accurately, and further improving the accuracy and the individuation degree of recommendation.
The technical scheme has the effects that: by recording the feedback information of the user, the system can know the preference, satisfaction, clicking behavior, score and the like of the user, so that the interests and demands of the user can be known more accurately. The feedback information and the second characteristic information of the user are subjected to data fusion, so that more comprehensive and detailed user characteristic vectors can be obtained, and the recommendation accuracy is improved; by optimizing user portrayal and training the model, the system can better understand the user's preferences and behavior patterns. By means of a machine learning algorithm such as a random forest algorithm, interests and demands of the user can be predicted according to the second user feature vector, and personalized recommendation results can be provided. In this way, the user can obtain recommended content that better matches his personal preferences and interests; personalized recommendation can better meet the demands of users and provide more valuable recommendation results. By recording satisfaction feedback of the user, the system can continuously optimize the recommendation strategy and the model, and further improve the user satisfaction. When users are satisfied and have a good interactive experience, they are more likely to continue using the product or service; by analyzing the feedback information of the user, the system can know the evaluation and purchasing behavior of the user on the recommendation result. These data can be used to better understand user needs and market trends, and thus to refine the operation and promotion strategy. The system can adjust the recommendation strategy according to the characteristics of different user groups, and provides more personalized and accurate recommendation content.
In one embodiment of the present invention, an advertising merchandise recommendation system, the system comprising:
an information acquisition module: obtaining operation logs and commodity characteristic information of a user from a server background;
and a data analysis module: analyzing first behavior data of the user according to the acquired operation log of the user, and acquiring first characteristic information of the user;
and a log classification module: dividing an operation log of a user into different log time according to different time nodes;
and a data combining module: combining the log time with the first behavior data of the user to obtain second behavior data of the user with different time nodes; acquiring second characteristic information of the user according to the second behavior data;
model construction module: establishing a user image through a machine learning algorithm according to the second characteristic information, extracting and encoding the commodity characteristic information through a deep learning algorithm, and constructing a commodity characteristic model, wherein the commodity characteristic model comprises semantic information and implicit characteristics of a commodity;
and a commodity recommendation module: based on the user portrait and the commodity feature model, recommending advertisement commodities to a user by adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to obtain first recommendation information;
And an optimization module: according to feedback of a user on the first recommendation information, first feedback information is obtained, the first feedback information is combined with the second characteristic information, user portraits are optimized, and a first optimization model is obtained;
and a recommendation adjustment module: continuously adjusting the first recommendation information according to the first optimization model to obtain second recommendation information, monitoring feedback of the user on the second recommendation information according to the first monitoring module to obtain second feedback information, obtaining third behavior data of the user according to the second feedback information and the first optimization model, obtaining third characteristic information of the user according to the third behavior data of the user, and continuously adjusting recommendation strategies according to the third characteristic information of the user.
The working principle of the technical scheme is as follows: and acquiring the operation log and commodity characteristic information of the user from the server background. The operation log of the user comprises behavior operations of the user on the platform, such as commodity browsing, searching, shopping cart adding, ordering, payment, evaluation and the like, and commodity characteristic information comprises text description, category labels, picture characteristics, prices, sales statistics, user evaluation and comment, historical purchase records, associated information and the like of the commodity; and analyzing the operation log of the user, extracting first behavior data of the user, and analyzing and processing the first behavior data to obtain first characteristic information of the user. Such characteristic information may include shopping preferences, interests, purchasing behavior, etc. of the user; and dividing the operation log of the user according to different time nodes to form different log time periods. Thus, the behavior of the user can be better analyzed and speculated, and more accurate recommendation is performed; and combining the user behavior data of different time nodes with the first characteristic information to obtain second behavior data of the user. Then, analyzing and processing according to the second behavior data, extracting second characteristic information of the user, and further optimizing the user portrait; and establishing a user portrait by using a machine learning algorithm, and integrating and analyzing the characteristic information of the user to form a user portrait model. And meanwhile, extracting and encoding commodity characteristic information by using a deep learning algorithm to construct a commodity characteristic model. The commodity feature model comprises semantic information and implicit features of commodities, such as text description, category labels, picture features, prices, sales statistics, user evaluation and comment, historical purchase records, associated information and the like; based on the user portrait and commodity feature model, adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to recommend advertisement commodity to the user. Recommending the most relevant and attractive advertisement goods according to interests and preferences of the user, and generating first recommendation information; and obtaining first feedback information according to feedback of the user on the first recommendation information. And combining the first feedback information with the second characteristic information, optimizing the user portrait, and generating a first optimization model. Continuously adjusting the first recommendation information by using the first optimization model to generate second recommendation information; and monitoring feedback of the user on the second recommendation information to obtain second feedback information. And obtaining third behavior data of the user according to the second feedback information and the first optimization model, and extracting third characteristic information of the user according to the third behavior data. And further optimizing a recommendation strategy according to the third characteristic information of the user, and continuously improving the accuracy and individuation degree of advertisement commodity recommendation.
The technical scheme has the effects that: and (3) establishing a user portrait and commodity characteristic model by analyzing and processing the operation log and commodity characteristic information of the user. Based on the user portrait and commodity feature model, a collaborative filtering algorithm and a content recommendation algorithm are combined, so that accurate advertisement commodity recommendation can be realized, and the most relevant and attractive advertisement commodity is provided for users; personalized recommendation strategies are customized by in-depth analysis of user interests, preferences and purchasing behavior. The user can obtain advertisement commodity recommendation meeting the interests and requirements of the user, so that the user experience and participation are improved, and the click rate and conversion rate of the user on advertisements are increased; accurate advertisement commodity recommendation can improve advertisement click rate and conversion rate, improves brand's awareness and sales volume. Meanwhile, the advertisement effect and the commercial value can be further improved by optimizing the recommendation strategy and continuously adjusting; and continuously monitoring and optimizing the recommendation strategy by combining the feedback of the user on the recommendation information. According to feedback information of the user and the first optimization model, recommended content can be dynamically adjusted, and timeliness and accuracy of advertisement commodity recommendation are ensured; the machine learning algorithm and the deep learning algorithm are used for processing and analyzing the user and commodity data, so that the characteristic information can be automatically extracted, and the user portrait and commodity characteristic model can be established. Thus, the utilization efficiency of resources can be greatly improved, and the manual intervention and the cost are reduced. Through dividing the operation log of the user into different log time according to different time nodes, the contents such as browsing preference of the user in different time can be obtained, for example, the user likes to browse cold drink in noon in summer and to browse milk tea in noon in winter, and waiting can be realized, so that more accurate recommendation can be realized.
An embodiment of the invention provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the memory, wherein the processor executes the program to realize any of the advertisement commodity recommendation methods.
In one embodiment of the present invention, a non-transitory computer readable storage medium has a computer program stored thereon, the program being executed by a processor to implement any of the above-described advertisement article recommendation methods.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. A method of recommending advertising merchandise, the method comprising:
obtaining operation logs and commodity characteristic information of a user from a server background;
analyzing first behavior data of the user according to the acquired operation log of the user, and acquiring first characteristic information of the user;
dividing an operation log of a user into different log time according to different time nodes;
Combining the log time with the first behavior data of the user to obtain second behavior data of the user with different time nodes; acquiring second characteristic information of the user according to the second behavior data;
establishing a user image through a machine learning algorithm according to the second characteristic information, extracting and encoding the commodity characteristic information through a deep learning algorithm, and constructing a commodity characteristic model, wherein the commodity characteristic model comprises semantic information and implicit characteristics of a commodity;
based on the user portrait and the commodity feature model, recommending advertisement commodities to a user by adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to obtain first recommendation information;
according to feedback of a user on the first recommendation information, first feedback information is obtained, the first feedback information is combined with the second characteristic information, user portraits are optimized, and a first optimization model is obtained;
continuously adjusting the first recommendation information according to the first optimization model to obtain second recommendation information, monitoring feedback of the user on the second recommendation information according to the first monitoring module to obtain second feedback information, obtaining third behavior data of the user according to the second feedback information and the first optimization model, obtaining third characteristic information of the user according to the third behavior data of the user, and continuously adjusting recommendation strategies according to the third characteristic information of the user.
2. The method of claim 1, wherein the first row of data includes merchandise browsing, searching, shopping cart adding, ordering, payment, evaluation, and communication with merchants; the first characteristic information includes shopping preferences, interests, and purchasing behavior of the user.
3. The method for recommending advertisement goods according to claim 1 or 2, wherein the step of analyzing the first behavior data of the user according to the obtained operation log of the user, and obtaining the first characteristic information of the user comprises the steps of:
cleaning an operation log of a user, removing invalid or error records, and processing missing values and abnormal values;
classifying the behavior types of the user according to the first behavior data of the user in the operation log;
extracting first characteristic information according to the classified behavior types;
based on the extracted first characteristic information, carrying out shopping preference analysis on the user; the shopping preference analysis comprises counting commodity categories, brands and price ranges of user preferences;
deducing the interests and hobbies of the user through the search keywords of the user and the browsed commodity information;
Carrying out shopping behavior analysis through the purchasing behavior data of the user;
if the operation log contains communication information with the merchant, the evaluation, complaint and consultation condition of the user on the goods and/or services are obtained by analyzing the interaction behavior of the user and the merchant.
4. The method of claim 1, wherein the log time comprises a first log time, a second log time, and a third log time; the first log time is divided into a first season time, a second season time, a third season time and a fourth season time according to seasons; the second log time is divided into working day time and rest day time; the third log time is divided into a first stage time, a second stage time, a third stage time and a fourth stage time.
5. The method for recommending advertisement goods according to claim 1 or 4, wherein,
the second behavior data includes first behavior data of a user at the different log times;
the second characteristic information includes second behavior data of the user at the different log times.
6. The method for recommending advertisement commodity according to claim 1, wherein said step of recommending advertisement commodity to the user by combining a collaborative filtering algorithm and a content recommendation algorithm based on said user representation and said commodity feature model comprises the specific steps of:
The second characteristic information of the user is subjected to numerical representation to form a first user characteristic vector;
carrying out numerical processing on the commodity characteristic information to form characteristic vectors of commodities;
matching the first user feature vector with the feature vector of the commodity through a content recommendation algorithm, and calculating the matching degree between the user and the commodity;
sorting according to the matching degree, and recommending advertisement commodities related to the user interests;
carrying out commodity recommendation on the user through a collaborative filtering algorithm;
the collaborative filtering algorithm comprises collaborative filtering based on users and collaborative filtering based on commodities;
the collaborative filtering based on the users comprises the steps of calculating the similarity between the users, finding other users similar to the current user, obtaining advertisement commodities liked by the similar users, and taking the advertisement commodities as recommendation options;
the collaborative filtering based on the commodities comprises the steps of calculating the similarity between advertisement commodities, finding out other commodities similar to the commodities historically liked by the user, and taking the other commodities as recommendation candidates;
comprehensively sequencing recommendation results of the collaborative filtering algorithm and the content recommendation algorithm through recommendation accuracy and individuation degree;
obtaining first recommendation information according to the sorting result, determining a final recommendation list according to the first recommendation information, and filtering out commodities which do not accord with user preference or recommendation strategies;
The ranked and filtered list of recommendations is presented to the user.
7. The method for recommending advertisement commodity according to claim 1, wherein the first feedback information is obtained according to the feedback of the user on the first recommendation information, the first feedback information is combined with the second feature information, and the user portrait is optimized to obtain a first optimization model; comprising the following steps:
when the user interacts and/or gives feedback to the first recommendation information, recording feedback information of the user, wherein the feedback information comprises preference, satisfaction, clicking behavior and scoring of the user;
extracting information related to the recommended commodity or the recommended result from the user feedback information;
carrying out data fusion on the first feedback information and second characteristic information of the user to obtain a second user characteristic vector;
optimizing the user image according to a machine learning algorithm through the second user feature vector, and training a model according to a random forest algorithm;
and acquiring a first optimization model according to the optimized portrait and the trained model.
8. An advertising merchandise recommendation system, the system comprising:
an information acquisition module: obtaining operation logs and commodity characteristic information of a user from a server background;
And a data analysis module: analyzing first behavior data of the user according to the acquired operation log of the user, and acquiring first characteristic information of the user;
and a log classification module: dividing an operation log of a user into different log time according to different time nodes;
and a data combining module: combining the log time with the first behavior data of the user to obtain second behavior data of the user with different time nodes; acquiring second characteristic information of the user according to the second behavior data;
model construction module: establishing a user image through a machine learning algorithm according to the second characteristic information, extracting and encoding the commodity characteristic information through a deep learning algorithm, and constructing a commodity characteristic model, wherein the commodity characteristic model comprises semantic information and implicit characteristics of a commodity;
and a commodity recommendation module: based on the user portrait and the commodity feature model, recommending advertisement commodities to a user by adopting a mode of combining a collaborative filtering algorithm and a content recommendation algorithm to obtain first recommendation information;
and an optimization module: according to feedback of a user on the first recommendation information, first feedback information is obtained, the first feedback information is combined with the second characteristic information, user portraits are optimized, and a first optimization model is obtained;
And a recommendation adjustment module: continuously adjusting the first recommendation information according to the first optimization model to obtain second recommendation information, monitoring feedback of the user on the second recommendation information according to the first monitoring module to obtain second feedback information, obtaining third behavior data of the user according to the second feedback information and the first optimization model, obtaining third characteristic information of the user according to the third behavior data of the user, and continuously adjusting recommendation strategies according to the third characteristic information of the user.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the memory, the processor executing the program to implement the method of recommending advertising merchandise according to any one of claims 1-8.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the program is executed by a processor to implement the advertisement item recommendation method of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310872267.9A CN116894709A (en) | 2023-07-14 | 2023-07-14 | Advertisement commodity recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310872267.9A CN116894709A (en) | 2023-07-14 | 2023-07-14 | Advertisement commodity recommendation method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116894709A true CN116894709A (en) | 2023-10-17 |
Family
ID=88311849
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310872267.9A Pending CN116894709A (en) | 2023-07-14 | 2023-07-14 | Advertisement commodity recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116894709A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117237024A (en) * | 2023-11-14 | 2023-12-15 | 深圳市诚王创硕科技有限公司 | Popularization method and system for marketing |
| CN117350770A (en) * | 2023-12-05 | 2024-01-05 | 西安柚子信息科技有限公司 | Accurate marketing method and system based on user portrait |
| CN117495458A (en) * | 2023-12-29 | 2024-02-02 | 河北华糖云商营销传播股份有限公司 | Advertisement online pushing method based on user portrait |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106202516A (en) * | 2016-07-24 | 2016-12-07 | 广东聚联电子商务股份有限公司 | A kind of e-commerce platform merchandise display method according to timing node |
| CN109034981A (en) * | 2018-08-23 | 2018-12-18 | 上海海事大学 | A kind of electric business collaborative filtering recommending method |
| CN111080413A (en) * | 2019-12-20 | 2020-04-28 | 深圳市华宇讯科技有限公司 | E-commerce platform commodity recommendation method and device, server and storage medium |
| CN111709812A (en) * | 2020-06-19 | 2020-09-25 | 山东财经大学 | E-commerce platform commodity recommendation method and system based on user dynamic classification |
| CN111882361A (en) * | 2020-07-31 | 2020-11-03 | 苏州云开网络科技有限公司 | Audience accurate advertisement pushing method and system based on artificial intelligence and readable storage medium |
| CN115757952A (en) * | 2022-11-17 | 2023-03-07 | 中国平安财产保险股份有限公司 | Content information recommendation method, device, equipment and storage medium |
-
2023
- 2023-07-14 CN CN202310872267.9A patent/CN116894709A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106202516A (en) * | 2016-07-24 | 2016-12-07 | 广东聚联电子商务股份有限公司 | A kind of e-commerce platform merchandise display method according to timing node |
| CN109034981A (en) * | 2018-08-23 | 2018-12-18 | 上海海事大学 | A kind of electric business collaborative filtering recommending method |
| CN111080413A (en) * | 2019-12-20 | 2020-04-28 | 深圳市华宇讯科技有限公司 | E-commerce platform commodity recommendation method and device, server and storage medium |
| CN111709812A (en) * | 2020-06-19 | 2020-09-25 | 山东财经大学 | E-commerce platform commodity recommendation method and system based on user dynamic classification |
| CN111882361A (en) * | 2020-07-31 | 2020-11-03 | 苏州云开网络科技有限公司 | Audience accurate advertisement pushing method and system based on artificial intelligence and readable storage medium |
| CN115757952A (en) * | 2022-11-17 | 2023-03-07 | 中国平安财产保险股份有限公司 | Content information recommendation method, device, equipment and storage medium |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117237024A (en) * | 2023-11-14 | 2023-12-15 | 深圳市诚王创硕科技有限公司 | Popularization method and system for marketing |
| CN117237024B (en) * | 2023-11-14 | 2024-02-20 | 深圳市诚王创硕科技有限公司 | Popularization method and system for marketing |
| CN117350770A (en) * | 2023-12-05 | 2024-01-05 | 西安柚子信息科技有限公司 | Accurate marketing method and system based on user portrait |
| CN117350770B (en) * | 2023-12-05 | 2024-03-08 | 西安柚子信息科技有限公司 | Accurate marketing method and system based on user portrait |
| CN117495458A (en) * | 2023-12-29 | 2024-02-02 | 河北华糖云商营销传播股份有限公司 | Advertisement online pushing method based on user portrait |
| CN117495458B (en) * | 2023-12-29 | 2024-03-26 | 河北华糖云商营销传播股份有限公司 | Advertisement online pushing method based on user portrait |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190220916A1 (en) | Next Generation Improvements In Recommendation Systems | |
| CN102902691B (en) | Recommend method and system | |
| Hung | A personalized recommendation system based on product taxonomy for one-to-one marketing online | |
| US10664889B2 (en) | System and method for combining and optimizing business strategies | |
| Lambrecht et al. | When does retargeting work? Information specificity in online advertising | |
| Chintagunta | Inertia and variety seeking in a model of brand-purchase timing | |
| Changchien et al. | On-line personalized sales promotion in electronic commerce | |
| Liao et al. | Mining customer knowledge to implement online shopping and home delivery for hypermarkets | |
| CN103246980B (en) | Information output method and server | |
| Bauer et al. | Recommender systems based on quantitative implicit customer feedback | |
| CN116894709A (en) | Advertisement commodity recommendation method and system | |
| Im et al. | Deal-seeking versus brand-seeking | |
| Scholz et al. | Using PageRank for non-personalized default rankings in dynamic markets | |
| US20150058119A1 (en) | Automated Advertisement of Products on Online Sites | |
| CN112435067A (en) | Intelligent advertisement putting method and system for cross-e-commerce platform and social platform | |
| WO2022095701A1 (en) | Method and device for recommending objects, equipment, and storage medium | |
| Wu et al. | An enhanced recommendation scheme for online grocery shopping | |
| KR20100123206A (en) | Method and apparatus for ranking analysis based on artificial intelligence, and recording medium thereof | |
| CN114841760B (en) | Advertisement recommendation management method and system based on audience behavior characteristic analysis | |
| CN110020136B (en) | Object recommendation method and related equipment | |
| Kim et al. | An empirical study of the structure of relevant keywords in a search engine using the minimum spanning tree | |
| Ming | Application research of customer big data analysis for online shop based on smart cloud platform tools | |
| JP6679705B1 (en) | Information processing apparatus, information processing method, and information processing program | |
| Yajdani | Factor influencing customer satisfaction of Daraz, Bangladesh | |
| Shruthi et al. | An Effective Product Recommendation System for E-Commerce Website Using Hybrid Recommendation Systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |