CN110222272B - Potential customer mining and recommending method - Google Patents
Potential customer mining and recommending method Download PDFInfo
- Publication number
- CN110222272B CN110222272B CN201910311200.1A CN201910311200A CN110222272B CN 110222272 B CN110222272 B CN 110222272B CN 201910311200 A CN201910311200 A CN 201910311200A CN 110222272 B CN110222272 B CN 110222272B
- Authority
- CN
- China
- Prior art keywords
- user
- model
- commodity
- users
- 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.)
- Active
Links
Images
Classifications
-
- 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
- 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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Accounting & Taxation (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Finance (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a potential customer mining and recommending method, which is characterized in that personal information and social activity information of a user are obtained from a social platform, are fused with a locally stored user shopping record, and data for training and testing a potential customer classification model are obtained after data cleaning and screening; then constructing a user portrait according to the personal information, social records and shopping records of the user, processing the social records and the shopping records of the user into a feature vector form for the model to use, then training a user interest prediction model, and dividing the user into potential customers and passersby; and finally, identifying and providing more targeted commodity pages for the potential customers according to the interests of the potential customers. The invention can judge the interest of the user while accurately classifying the user; judging to display corresponding products or implement accurate advertisement putting according to the interest of the user, and realizing the conversion of potential customers; targeted recommendations may also be provided for older customers, increasing customer stickiness.
Description
Technical Field
The invention relates to the technical field of client mining methods, in particular to a potential client mining and recommending method.
Background
In the electronic commerce era with increasingly intense competition, new customers are continuously expanded on the basis of original customers, the total quantity of the customers and the viscosity of the customers are increased, enterprises can obtain more economic benefits and market competition advantages, more and more merchants sell goods through online shopping malls, and in the process of sales promotion of the merchants, one very concerned problem is that: how to excavate potential customers according to the information of customers (such as the information of the ages, the sexes, the family addresses and the like of the customers) owned by the merchants at present, and the precise reaching of the customers is realized.
Existing potential customer mining techniques are mainly divided into two categories, namely user attribute tag-based and user browsing behavior-based.
Potential customer mining technology based on user attribute tags, such as an invention patent with the application number of 201510176915 of Ali Baba group holdings GmbH and the invention name of 'method and device for mining potential customers'; application No. 201510696762 of the double-technology company Limited in Beijing, patent application entitled "method and apparatus for mining potential customers", and the like. The technology mainly sorts and screens user attribute labels (gender, age and the like), clusters users according to the attribute labels to form different communities, and judges whether the users are potential clients or not by calculating which community the users belong to.
A potential customer mining technology based on user browsing behaviors, such as an invention patent with the application number of 201510903856 of the focal science and technology limited company and the invention name of "a potential customer mining method based on customer behavior characteristics"; the application number of 201710807133 of Shenzhen science and technology (Shenzhen) Limited company, the invention patent named as 'a potential customer identification method and terminal equipment', and the like. The technology mainly utilizes behaviors of client browsing, collecting and the like, and judges whether the user is a potential client or not by comparing and calculating the similarity between the user browsing behavior and the client browsing behavior.
However, in the prior art, only information such as a customer attribute tag and a behavior record stored in an e-commerce website is used, for a user who newly logs in the website, the problem of low mining precision caused by too little user information exists, and meanwhile, the user interest is difficult to accurately judge, and an enterprise cannot be effectively guided to make a corresponding strategy to realize customer conversion.
With the development of third-party account login technology, more and more users start to choose to log in the e-commerce website by using their own social accounts (QQ, microblog, twitter, facebook, etc.), and the number of cross-clients (both clients of the e-commerce website and clients of other social websites) is gradually increasing. Studies by scholars such as w.x.zhao et al show that social behaviors of users are important sources of information to help predict user purchasing interests. Therefore, how to effectively utilize the social behaviors of cross clients, mine potential clients from new users who log in a website by using social account numbers, and make corresponding strategies to realize client conversion is a problem to be solved.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a potential customer mining and recommending method, which utilizes the social behavior characteristics of cross customers to mine potential customers who log in users of an e-commerce website by using social account numbers, predicts commodities which are possibly purchased by the users and provides accurate reference basis for customer conversion.
The technical scheme of the invention is as follows: a potential customer mining and recommending method comprises the following steps:
s1), acquiring personal information and social activity data of a user from a social platform, cleaning the acquired rough original data, removing data irrelevant to the user information, such as state codes, and the like, formatting text information, removing special characters in the text, and storing the cleaned data in a social activity database of the user;
s2), screening according to the personal information, social behavior information and shopping behavior information of the user, and selecting key information to construct a user portrait;
s3), carrying out characterization processing on various types of data, and converting the data into a feature vector for model training;
s4), training, updating a factor decomposition model, a model pool consisting of an svm model and an xgboost model, predicting the interest of a user in commodities, using the user portrait and the sorted characteristic vectors as the input of the model pool, performing independent training on the model by using the recent actual purchase result of the user as a label, and then updating the parameters of the model through a gradient descent method according to errors to improve the prediction precision of the model;
s5), testing and adjusting the model, performing cross test on the model by using data of another batch of users different from the training, obtaining the trained model when the test effect is higher than a threshold value, and otherwise, re-training the model by adopting a characteristic screening and parameter adjusting mode according to the step S4);
s6), predicting all users by using the trained model to obtain the latest interest distribution of the users, and updating the interest prediction results of the users by adopting a direct coverage method or a weighted addition method;
s7), according to the browsing records, the shopping records and the time factors of the users, the users are classified into old customers and new users, the new users are used for mining potential customers, then the new users are classified into potential customers and passers-by which purchasing behaviors are unlikely to occur by means of an svm classification model, and the potential customers are mined from the new users;
s8), training and updating the svm classification model, using the characteristics obtained in the step S3) and the step S6) as input of the svm classification model, training the svm classification model by using the actual transformation condition after the user is recommended as a label, and updating parameters of the model according to errors by a random gradient descent method to improve the prediction precision of the model;
s9), testing and adjusting the model, performing cross test by using data of another group of users different from the training users, obtaining the trained model when the test effect is higher than a threshold value, and otherwise, re-training the model by adopting a characteristic screening and parameter adjusting mode according to the step S8);
s10), predicting all new users by using the model trained in the S9), dividing the new users into potential users and passers-by unlikely to generate purchasing behaviors, and updating a new passenger recommendation database;
s11), judging whether a user of a new access platform is a potential client, inquiring information of the access user from a new client recommendation database, and if interest information corresponding to the user exists, screening commodities conforming to the interest of the user to generate a commodity list;
s12), displaying and recommending commodities in the commodity list on a front-end page, or packaging the commodities into advertisements to be accurately delivered to the user.
Further, step S1) adopts a crawler technology to obtain personal information and social activity information of the user from a plurality of social network sites.
Further, in step S2), the personal information of the user mainly includes gender, age, region, work, school, interest tag, user level, user reputation, attention count, and fan count, the social behavior information includes a sending text, an attention topic, and a forwarding comment, and the shopping behavior information includes browsing information, collection information, purchase information, and evaluation information.
Further, in step S3), the feature vector includes a user feature vector and a commodity feature vector; the user feature vector consists of user personal features, user social behavior features and user-commodity interaction features, and the commodity feature vector consists of commodity category features, shopping context features and commodity-user interaction features.
Further, in step S3), performing characterization processing on various types of data, specifically including normalizing numerical data, performing one-hot discrete vectorization on category label type data, performing word2vec word vectorization and doc2vec text vectorization on text content, and completing missing data to finally obtain user vector representation, user social contact and shopping behavior vector representation, and commodity vector representation.
Further, in step S3), vectorization processing is performed on the personal information of the user, such as gender, age, region, work, school, interest tag, user level, user reputation, attention count, fan count, and the like, to construct personal characteristics of the user, where the gender, region, age, and work data are expressed by one-hot vector representation, and data of other numerical types are expressed by max-min normalization.
Further, in step S3), the user social behavior features mainly include text information sent in user social contact, the text information sent by the user within a certain time is spliced into a document, the document is segmented by a word segmentation tool and stop words are removed, then each word is converted into a corresponding word vector by a word2vec tool, and all the word vectors are added and normalized to obtain a vector representation of the corresponding document; and meanwhile, generating a document vector corresponding to the document through a doc2vec tool, and finally splicing the two expression vectors to serve as the social behavior feature vector of the user.
Further, the user-commodity interaction feature means that commodities purchased by a user are represented in a vector form by one-hot, that is, the position value of a purchased commodity is 1, and the position value of an unpurchased commodity is 0.
Furthermore, the shopping context characteristics refer to the shopping records of the user are sequenced according to the shopping time sequence, a commodity sequence purchased by the user is constructed, each commodity is regarded as a word, then training is carried out through a skip-gram model in a word2vec tool, a vector corresponding to each commodity can be obtained after the training is finished, and the vector is used as the shopping context characteristics of the commodity.
Further, in step S4), the average result predicted by the factorization machine model, the svm model and the xgboost model is used as the final output result, which is responsible for predicting the interest of the user in the commodity.
Further, step S7), the potential customers refer to those who may buy, and the passers-by refers to those who only browse but are unlikely to buy the goods.
The invention has the beneficial effects that:
1. the personal information and behavior data of the user on the social platform contain a large amount of useful information, and the accuracy is higher than that of other potential customer mining systems using the browsing behavior of the user, so that the user interest can be judged more accurately.
2. For the user who browses the commodities on the login platform, the interested commodities can be displayed to the user according to the prediction result, and the conversion rate is improved.
3. And for the user of the browsing platform, advertisement putting can be carried out according to the prediction result, and customer conversion is realized.
4. Targeted recommendations may also be provided for older customers, increasing customer stickiness.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a flow chart of the training of a potential customer mining model of the present invention;
FIG. 3 is a flow chart of potential customer identification and conversion in accordance with the present invention.
Detailed Description
The following further describes embodiments of the present invention with reference to the accompanying drawings:
as shown in fig. 1, a potential customer mining and recommending method includes the following steps:
s1), acquiring personal information and social activity data of a user from a social platform by adopting a crawler technology, cleaning the acquired rough original data, removing data such as state codes and the like which are irrelevant to the user information, formatting text information, removing special characters in the text, storing the cleaned data into a social activity database of the user, acquiring corresponding data from the social platform regularly, and updating the data in the database in real time;
s2) screening according to personal information, social behavior information and shopping behavior information of a user, selecting key information to construct a user portrait, wherein the personal information of the user mainly comprises sex, age, region, work, school, interest tag, user grade, user reputation, attention number and fan number, the social behavior information comprises information such as a sent text, an attention topic and a forwarded comment, and the shopping behavior information comprises information such as browsing information, collection information, purchase information and evaluation information;
s3), performing characterization processing on various types of data, converting the data into a feature vector for model training, specifically normalizing numerical data, performing one-hot discrete vectorization on category label type data, performing word2vec word vectorization and doc2vec text vectorization on text content, supplementing missing data, and finally obtaining user vector representation, user social contact and shopping behavior vector representation and commodity vector representation, wherein the feature vector comprises a user feature vector and a commodity feature vector; the user feature vector consists of user personal features, user social behavior features and user-commodity interaction features, and the commodity feature vector consists of commodity category features, shopping context features and commodity-user interaction features.
The method comprises the steps of vectorizing user personal information such as gender, age, region, work, school, interest labels, user grade, user credit, attention number, fan number and the like, and constructing user personal characteristics, wherein discrete data of the label types such as gender, region, age, work and the like are represented by one-hot vectors, for example [1,0] represents that the gender is female, [0,1] represents that the gender is male, and [0,0] represents that the gender is unknown, wherein the age is divided into segments every 10 years, and data of other numerical value types such as the user grade, the attention number, the fan number and the like are represented by max-min normalization.
The social behavior characteristics of the user mainly comprise text information sent in the social contact of the user, the text information sent in the last 3 months of the user is spliced into a document, the document is segmented by a word segmentation tool and stop words are removed, then each word is converted into a corresponding 100-dimensional word vector by a word2vec tool, and all the word vectors are added and normalized to obtain the vector representation of the corresponding document; and meanwhile, generating a document vector corresponding to the document through a doc2vec tool, and finally splicing the two expression vectors to serve as the social behavior feature vector of the user.
And representing the commodities purchased by the user in a vector form through one-hot, namely, the position value of the purchased commodities is 1, and the position value of the purchased commodities is 0, wherein the commodity is taken as the user-commodity interaction feature.
In order to facilitate the inquiry and screening when a user shops, the commodities are generally divided into different categories, such as clothes, foods and the like, and the category characteristics corresponding to the commodities are represented by using a one-hot representation.
Sequencing shopping records of users according to the shopping time sequence, constructing a commodity sequence purchased by the users, regarding each commodity as a word, then training through a skip-gram model in a word2vec tool, and obtaining a vector corresponding to each commodity after training is completed as the shopping context characteristics of the commodity. The real label of the preference degree of each commodity of the user is constructed through the historical shopping record of the user, and if the user purchases the commodity before, the corresponding value of the commodity is 1; otherwise, if the user has not purchased the product, the value corresponding to the product is 0.
S4), training, updating a factorization model, a model pool composed of an svm model and an xgboost model, predicting the interest of a user to commodities, splicing a user vector and a commodity vector of each commodity as the input of the model pool, respectively inputting a batch of feature vectors into 3 models, using the models to predict the preference scores of the user to the commodities, respectively calculating the difference between the model prediction result and the actual label result according to the actual label value of the user, updating the parameters of the 3 models through a batch gradient descent algorithm, and gradually improving the model prediction precision in a continuous iterative updating mode to ensure that the prediction value of the model gradually approaches to the real purchasing condition, wherein a training flow chart of the model is shown in FIG. 2.
S5), testing and adjusting the model, performing cross test on the model by using data of another batch of users different from the training, obtaining the trained model when the test effect is higher than a threshold value, and otherwise, re-training the model by adopting a characteristic screening and parameter adjusting mode according to the step S4);
s6), predicting all users by using the trained model to obtain the latest interest distribution of the users, and updating the interest prediction results of the users by adopting a direct coverage method or a weighted addition method;
s7), dividing the users into old customers and new users according to browsing records, shopping records and time factors of the users, wherein the new users are used for mining potential customers, then dividing the new users into potential customers and passers-by using an svm classification model, mining the potential customers from the new users, and performing classification prediction on all newly added users by using the svm classification model at regular intervals;
s8), training and updating the svm classification model, using the features obtained in the step S3) and the step S6) as the input of the svm classification model, training the svm classification model by using the actual transformation condition after the recommendation to the user as a label, and updating the parameters of the model according to errors by a stochastic gradient descent method to improve the prediction precision of the model;
s9), testing and adjusting the model, performing cross test by using data of another group of users different from the training users, obtaining the trained model when the test effect is higher than a threshold value, and re-training the model in a characteristic screening and parameter adjusting mode according to the step S8) if the test effect is not higher than the threshold value;
s10), predicting all new users by using the model trained in the S9), dividing the new users into potential users and passers-by unlikely to generate purchasing behaviors, and updating a new passenger recommendation database;
s11), judging whether a user of a new access platform is a potential client, inquiring information of the access user from a new client recommendation database, and if interest information corresponding to the user exists, screening commodities conforming to the interest of the user to generate a commodity list;
s12), displaying and recommending commodities in the commodity list on a front-end page, or packaging the commodities into advertisements to be accurately delivered to the user, as shown in fig. 3.
The foregoing embodiments and description have been presented only to illustrate the principles and preferred embodiments of the invention, and various changes and modifications may be made therein without departing from the spirit and scope of the invention as hereinafter claimed.
Claims (10)
1. A potential customer mining and recommending method is characterized by comprising the following steps:
s1), acquiring personal information and social activity data of a user from a social platform, cleaning the acquired rough original data, removing data of a state code irrelevant to the user information, formatting text information, removing special characters in the text, and storing the cleaned data in a social activity database of the user;
s2), screening according to the personal information, social behavior information and shopping behavior information of the user, and selecting key information to construct a user portrait;
s3), carrying out characterization processing on various types of data, and converting the data into a feature vector for model training;
s4), training, updating a factor decomposition model, a model pool consisting of an svm model and an xgboost model, predicting the interest of a user in commodities, using the user portrait and the sorted characteristic vectors as the input of the model pool, performing independent training on the model by using the recent actual purchase result of the user as a label, and then updating the parameters of the model through a gradient descent method according to errors to improve the prediction precision of the model;
s5), testing and adjusting the model, performing cross test on the model by using data of another batch of users different from the training, obtaining the trained model when the test effect is higher than a threshold value, and otherwise, re-training the model by adopting a characteristic screening and parameter adjusting mode according to the step S4);
s6), predicting all users by using the trained model to obtain the latest interest distribution of the users, and updating the interest prediction results of the users by adopting a direct coverage method or a weighted addition method;
s7), classifying users into old customers and new users according to browsing records, shopping records and time factors of the users, wherein the new users are used for mining potential customers, then classifying the new users into potential customers and passers-by unlikely to generate purchasing behaviors by utilizing an svm classification model, and mining the potential customers from the new users;
s8), training and updating the svm classification model, using the features obtained in the step S3) and the step S6) as the input of the svm classification model, training the svm classification model by using the actual transformation condition after the recommendation to the user as a label, and updating the parameters of the model according to errors by a stochastic gradient descent method to improve the prediction precision of the model;
s9), testing and adjusting the model, performing cross test by using data of another group of users different from the training users, obtaining the trained model when the test effect is higher than a threshold value, and re-training the model in a characteristic screening and parameter adjusting mode according to the step S8) if the test effect is not higher than the threshold value;
s10), predicting all new users by using the model trained in the S9), dividing the new users into potential users and passers-by unlikely to generate purchasing behaviors, and updating a new passenger recommendation database;
s11), judging whether a user of a new access platform is a potential client, inquiring information of the access user from a new client recommendation database, and if interest information corresponding to the user exists, screening commodities conforming to the interest of the user to generate a commodity list;
s12), displaying and recommending commodities in the commodity list on a front-end page, or packaging the commodities into advertisements to be accurately delivered to the user.
2. The method of claim 1, wherein the method comprises the following steps: in step S2), the personal information of the user mainly includes gender, age, region, work, school, interest tag, user level, user reputation, attention count, and fan count, the social behavior information mainly includes a sending text, an attention topic, and a forwarding comment, and the shopping behavior information includes browsing information, collection information, purchase information, and evaluation information.
3. The method of claim 1, wherein the method comprises the following steps: in the step S3), various types of data are subjected to characterization processing, specifically, numerical data are normalized, category label type data are subjected to one-hot discrete vectorization, word2vec word vectorization and doc2vec text vectorization are performed on text contents, missing data are supplemented, and finally user vector representation, user social contact and shopping behavior vector representation and commodity vector representation are obtained.
4. The method of claim 1, wherein the method comprises the following steps: in the step S3), the characteristic vector comprises a user characteristic vector and a commodity characteristic vector; the user feature vector consists of user personal features, user social behavior features and user-commodity interaction features, and the commodity feature vector consists of commodity category features, shopping context features and commodity-user interaction features.
5. The method of claim 4, wherein the step of mining and recommending potential customers comprises: and step S3), vectorizing personal information of the users such as gender, age, region, work, school, interest tag, user grade, user credit, attention number and fan number, and constructing personal characteristics of the users, wherein the gender, region, age and work data are expressed by a one-hot vector representation method, and data of other numerical types are expressed by max-min normalization.
6. The method of claim 4, wherein the step of mining and recommending potential customers comprises: in the step S3), the social behavior characteristics of the user mainly comprise text information sent in the user social contact, the text information sent in a certain time by the user is spliced into a document, the document is segmented by a word segmentation tool and stop words are removed, then each word is converted into a corresponding word vector by a word2vec tool, and all the word vectors are added and normalized to obtain the vector representation of the corresponding document; and meanwhile, generating a document vector corresponding to the document through a doc2vec tool, and finally splicing the two expression vectors to serve as the social behavior feature vector of the user.
7. The method of claim 4, wherein the step of mining and recommending potential customers comprises: the user-commodity interactive feature means that commodities purchased by a user are represented in a vector form through one-hot, namely the position value of purchased commodities is 1, and the position value of unpurchased commodities is 0.
8. The method of claim 4, wherein the step of mining and recommending potential customers comprises: the shopping context characteristics refer to the steps that the shopping records of the user are sequenced according to the shopping time sequence, a commodity sequence purchased by the user is constructed, each commodity is regarded as a word, then training is carried out through a skip-gram model in a word2vec tool, a vector corresponding to each commodity can be obtained after the training is finished, and the vector is used as the shopping context characteristics of the commodity.
9. The method of claim 1, wherein the method comprises the following steps: in the step S4), the average result predicted by the factorization machine model, the svm model and the xgboost model is used as a final output result to be responsible for predicting the interest of the user in the commodity.
10. The method of claim 1, wherein the method comprises the following steps: in the step S4), the real label of the preference degree of each commodity is constructed by the historical shopping record of the user, and if the user purchases the commodity before, the corresponding value of the commodity is 1; otherwise, if the user does not purchase the commodity, the value corresponding to the commodity is 0, the interest of the user to the commodity is predicted, the user vector and the commodity vector of each commodity are spliced to be used as the input of a model pool, the preference of the user to the commodity is predicted by using the model, then the difference value between the model prediction result and the actual label result is respectively calculated according to the actual label value of the user, the parameters of the 3 models are updated through a batch gradient descent algorithm, and the model prediction precision is gradually improved through a continuous iteration updating mode.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910311200.1A CN110222272B (en) | 2019-04-18 | 2019-04-18 | Potential customer mining and recommending method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910311200.1A CN110222272B (en) | 2019-04-18 | 2019-04-18 | Potential customer mining and recommending method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110222272A CN110222272A (en) | 2019-09-10 |
| CN110222272B true CN110222272B (en) | 2022-10-14 |
Family
ID=67822621
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910311200.1A Active CN110222272B (en) | 2019-04-18 | 2019-04-18 | Potential customer mining and recommending method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110222272B (en) |
Families Citing this family (41)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111651686B (en) * | 2019-09-24 | 2021-02-26 | 北京嘀嘀无限科技发展有限公司 | Test processing method and device, electronic equipment and storage medium |
| CN112561555A (en) * | 2019-09-26 | 2021-03-26 | 北京国双科技有限公司 | Product data processing method and device |
| CN110889716A (en) * | 2019-09-29 | 2020-03-17 | 清华大学 | Method and device for identifying potential registered user |
| CN111127074B (en) * | 2019-11-26 | 2023-04-25 | 杭州聚效科技有限公司 | Data recommendation method |
| CN110933472B (en) * | 2019-12-02 | 2021-10-22 | 深圳市云积分科技有限公司 | Method and device for realizing video recommendation |
| CN111199421B (en) * | 2019-12-20 | 2023-09-29 | 北京淇瑀信息科技有限公司 | Social relationship-based user recommendation method and device and electronic equipment |
| CN111144986B (en) * | 2019-12-25 | 2024-05-31 | 清华大学 | Social electronic commerce website commodity recommendation method and device based on sharing behavior |
| CN111177581A (en) * | 2019-12-25 | 2020-05-19 | 清华大学 | Multi-platform-based social e-commerce website commodity recommendation method and device |
| CN113065880A (en) * | 2020-01-02 | 2021-07-02 | 中国移动通信有限公司研究院 | Group dissatisfaction user identification method, device, equipment and storage medium |
| CN111598613A (en) * | 2020-04-28 | 2020-08-28 | 杭州沃朴物联科技有限公司 | Red packet issuing method, device, equipment and medium |
| CN111598256B (en) * | 2020-05-18 | 2023-08-08 | 北京互金新融科技有限公司 | Processing method and device for default purchase behavior of target client |
| CN113744002B (en) * | 2020-05-27 | 2024-07-19 | 北京沃东天骏信息技术有限公司 | Method, device, equipment and computer readable medium for pushing information |
| CN111814092A (en) * | 2020-07-21 | 2020-10-23 | 上海数鸣人工智能科技有限公司 | Data preprocessing method for artificial intelligence algorithm based on user internet behavior |
| CN112016003B (en) * | 2020-08-19 | 2022-07-12 | 重庆邮电大学 | Social crowd user tag mining and similar user recommending method based on CNN |
| CN112070615B (en) * | 2020-09-02 | 2024-06-28 | 中国银行股份有限公司 | Financial product recommendation method and device based on knowledge graph |
| CN112200601B (en) * | 2020-09-11 | 2024-05-14 | 深圳市法本信息技术股份有限公司 | Item recommendation method, device and readable storage medium |
| CN112287213A (en) * | 2020-10-21 | 2021-01-29 | 重庆电子工程职业学院 | Information analysis method based on big data |
| CN112269911A (en) * | 2020-11-11 | 2021-01-26 | 深圳视界信息技术有限公司 | Equipment information identification method, model training method, device, equipment and medium |
| CN112288549B (en) * | 2020-11-18 | 2024-05-31 | 杭州拼便宜网络科技有限公司 | Commodity recommendation list generation method, device and equipment and readable storage medium |
| CN112508603A (en) * | 2020-11-26 | 2021-03-16 | 泰康保险集团股份有限公司 | Method and device for mining potential customer information of endowment community |
| CN112435067A (en) * | 2020-11-30 | 2021-03-02 | 翼果（深圳）科技有限公司 | Intelligent advertisement putting method and system for cross-e-commerce platform and social platform |
| CN112667919A (en) * | 2020-12-28 | 2021-04-16 | 山东大学 | Personalized community correction scheme recommendation system based on text data and working method thereof |
| CN112667911A (en) * | 2021-01-14 | 2021-04-16 | 中山世达模型制造有限公司 | Method for searching potential customers by using social software big data |
| CN113268656A (en) * | 2021-04-15 | 2021-08-17 | 北京沃东天骏信息技术有限公司 | User recommendation method and device, electronic equipment and computer storage medium |
| CN113204577A (en) * | 2021-04-15 | 2021-08-03 | 北京沃东天骏信息技术有限公司 | Information pushing method and device, electronic equipment and computer readable medium |
| CN113177151A (en) * | 2021-05-28 | 2021-07-27 | 中山世达模型制造有限公司 | Potential customer screening method |
| CN113379529A (en) * | 2021-06-07 | 2021-09-10 | 广发银行股份有限公司 | Collaborative decision engine application framework |
| CN113393271B (en) * | 2021-06-15 | 2022-09-23 | 湖南汽车工程职业学院 | Product customer big data application matching system and computer storage medium |
| CN113626704A (en) * | 2021-08-10 | 2021-11-09 | 平安国际智慧城市科技股份有限公司 | Method, device and equipment for recommending information based on word2vec model |
| CN113822596B (en) * | 2021-10-12 | 2023-08-29 | 深圳市单仁牛商科技股份有限公司 | Customer screening method based on big data |
| CN114187036B (en) * | 2021-11-30 | 2022-10-11 | 深圳市喂车科技有限公司 | Internet advertisement intelligent recommendation management system based on behavior characteristic recognition |
| CN114267457B (en) * | 2021-12-03 | 2022-11-29 | 爱优牙信息技术(深圳)有限公司 | Dentist service platform matched with directional client |
| CN114387064B (en) * | 2022-01-13 | 2024-07-19 | 福州大学 | Electronic commerce platform potential customer recommendation method and system based on comprehensive similarity |
| CN114595323B (en) * | 2022-03-04 | 2023-03-10 | 北京百度网讯科技有限公司 | Portrait construction, recommendation, model training method, apparatus, device and storage medium |
| CN114841760B (en) * | 2022-06-30 | 2022-09-02 | 北京聚云数字信息技术有限公司 | Advertisement recommendation management method and system based on audience behavior characteristic analysis |
| CN116308464B (en) * | 2023-05-11 | 2023-09-08 | 广州市沃钛移动科技有限公司 | Target client acquisition system and method |
| CN116485560A (en) * | 2023-06-21 | 2023-07-25 | 杭州大鱼网络科技有限公司 | Target user screening method and system based on feedback mechanism |
| CN116523572B (en) * | 2023-06-28 | 2023-09-08 | 悦享星光(北京)科技有限公司 | Client mining method and system based on client behavior characteristics |
| CN116894692B (en) * | 2023-09-11 | 2023-11-24 | 北京亿家老小科技有限公司 | Method and system for analyzing and monitoring potential demands of online network sales users |
| CN116957691B (en) * | 2023-09-19 | 2024-01-19 | 翼果（深圳）科技有限公司 | Cross-platform intelligent advertisement putting method and system for commodities of e-commerce merchants |
| CN117035853B (en) * | 2023-10-09 | 2024-02-02 | 销生客(北京)数字科技有限公司 | Potential customer identity marking system based on big data |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160267498A1 (en) * | 2015-03-10 | 2016-09-15 | Wipro Limited | Systems and methods for identifying new users using trend analysis |
| CN105488697A (en) * | 2015-12-09 | 2016-04-13 | 焦点科技股份有限公司 | Potential customer mining method based on customer behavior characteristics |
| CN105824912A (en) * | 2016-03-15 | 2016-08-03 | 平安科技（深圳）有限公司 | Personalized recommending method and device based on user portrait |
| CN108090162A (en) * | 2017-12-13 | 2018-05-29 | 北京百度网讯科技有限公司 | Information-pushing method and device based on artificial intelligence |
| CN109376766B (en) * | 2018-09-18 | 2023-10-24 | 平安科技（深圳）有限公司 | Portrait prediction classification method, device and equipment |
-
2019
- 2019-04-18 CN CN201910311200.1A patent/CN110222272B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110222272A (en) | 2019-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222272B (en) | Potential customer mining and recommending method | |
| CN111444334B (en) | Data processing method, text recognition device and computer equipment | |
| CN102902691B (en) | Recommend method and system | |
| CN109711955B (en) | Poor evaluation early warning method and system based on current order and blacklist base establishment method | |
| CN104462333A (en) | Shopping search recommending and alarming method and system | |
| CN111047412A (en) | Big data electricity merchant operation platform | |
| CN108921602B (en) | User purchasing behavior prediction method based on integrated neural network | |
| KR20190142508A (en) | Method and apparatus for providing product information | |
| Yan et al. | Implementation of a product-recommender system in an IoT-based smart shopping using fuzzy logic and apriori algorithm | |
| CN111738805B (en) | Behavior log-based search recommendation model generation method, device and storage medium | |
| CN107169806A (en) | For determining method and device of the item property for the disturbance degree of purchase decision | |
| CN103246991A (en) | Data mining-based customer relationship management method and data mining-based customer relationship management system | |
| JP6976207B2 (en) | Information processing equipment, information processing methods, and programs | |
| CN112200601A (en) | Item recommendation method and device and readable storage medium | |
| CN110955690A (en) | Self-service data labeling platform and self-service data labeling method based on big data technology | |
| CN115409577A (en) | Intelligent container repurchase prediction method and system based on user behavior and environmental information | |
| CN111177581A (en) | Multi-platform-based social e-commerce website commodity recommendation method and device | |
| CN113516496A (en) | Advertisement conversion rate pre-estimation model construction method, device, equipment and medium thereof | |
| CN108475387A (en) | Increase selection using Individual Motivation using social media data to share | |
| CN109615437A (en) | Sale obtains objective method for tracking and managing | |
| CN111784428A (en) | Information pushing method and device, electronic commerce system and storage medium | |
| Alizadeh et al. | Suitable delivery system in small e-commerce companies | |
| CN117455621A (en) | Personalized recommendation method and device, storage medium and computer equipment | |
| CN111861679A (en) | Commodity recommendation method based on artificial intelligence | |
| Wei et al. | Online shopping behavior analysis for smart business using big data analytics and blockchain security |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |