CN110222272A - A kind of potential customers excavate and recommended method - Google Patents
A kind of potential customers excavate and recommended method Download PDFInfo
- Publication number
- CN110222272A CN110222272A CN201910311200.1A CN201910311200A CN110222272A CN 110222272 A CN110222272 A CN 110222272A CN 201910311200 A CN201910311200 A CN 201910311200A CN 110222272 A CN110222272 A CN 110222272A
- Authority
- CN
- China
- Prior art keywords
- user
- model
- commodity
- vector
- 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.)
- Granted
Links
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)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Accounting & Taxation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides a kind of potential customers and excavates and recommended method, the present invention obtains the personal information and social activity information of user from social platform, and record of doing shopping with the user being locally stored merges, after data cleansing and screening, the data for training and testing potential customers' disaggregated model are obtained;Then according to userspersonal information, social record, shopping record structuring user's portrait, the social of user is recorded and is done shopping record processing as the feature vector form used for model simultaneously, user is divided into potential customers and passerby by then training user's interest prediction model;Finally identification simultaneously provides more targeted commodity page presentation to them according to the interest of potential customers.The present invention can judge the interest of user while exact classification user;Corresponding product is shown according to the judgement of the interest of user or implements accurate advertisement dispensing, realizes the conversion of potential customers;Frequent customer can also be provided and targetedly recommended, client's stickiness is increased.
Description
Technical field
The present invention relates to Customer mining method and technology field, especially a kind of potential customers excavate and recommended method.
Background technique
In this increasingly competitive electronic commerce times, new client is constantly expanded on the basis of existing client,
More economic benefits and market competition advantage can be obtained by increasing client's total amount and client's viscosity, enterprise, more and more
Businessman sells kinds of goods by online shopping mall, and during businessman promotes, a problem being extremely concerned about is: how basis
The information (such as the information such as age, gender, home address of client) for the client that businessman has possessed now, excavates potential visitor
Family realizes that the accurate touching of client reaches.
Existing potential customers' digging technology is broadly divided into based on user property label and based on two class of user browsing behavior.
The application number of potential customers' digging technology such as Alibaba Group Holdings Limited based on user property label
For the 201510176915, patent of invention of entitled " method and apparatus for excavating potential customers ";The double science and technology of Beijing state have
Limit company application No. is 201510696762, the patent application of entitled " method and apparatus for excavating potential customers "
Deng.Above-mentioned technology mainly passes through arrangement screening user property label (gender, age etc.), clusters user according to attribute tags,
Different community is formed, by calculating which community user belongs to, to judge whether user is potential customers.
Potential customers' digging technology such as Focus Technology Corp. based on user browsing behavior application No. is
201510903856, the patent of invention of entitled " a kind of potential customers' method for digging based on customer action feature ";It is flat
Pacify scientific and technological (Shenzhen) Co., Ltd application No. is a kind of 201710807133, entitled " recognition methods of potential customers
And terminal device " patent of invention etc..Above-mentioned technology mainly utilizes the behaviors such as client's browsing, collection, passes through comparing calculation user
Similarity degree between browsing behavior and client's browsing behavior judges whether user is potential customers.
But the prior art has only used the information such as client properties label and the behavior record of electric business website storage inside,
For newly logging in the user of website, there is a problem of that the very few caused excavation precision of user information is low, while being difficult to accurately judge
User interest can not effectively instruct enterprise to formulate corresponding strategy and realize client's conversion.
As third party's account logs in the development of technology, more and more users start the social account that selection uses oneself
(QQ, microblogging, Twitter, Facebook etc.) logs in electric business website, intersect client (be both electric business website client and other
The client of social network sites) quantity gradually increase.W.X.Zhao etc. scholars' studies have shown that the Social behaviors of user be side
Prediction user is helped to buy the important information source of interest.Therefore the Social behaviors for intersecting client how are efficiently used, from using social activity
Account logs in the new user of website and excavates potential customers, and formulates corresponding strategy and realize client's conversion, is to have to be solved ask
Topic.
Summary of the invention
In view of the deficiencies of the prior art, the present invention provides a kind of potential customers and excavates and recommended method, and the present invention is using handing over
The Social behaviors feature of client is pitched, excavates and logs in the potential customers in electric business website user using social account, and predict user
It is possible that the commodity of purchase, provide accurately for client's conversion referring to foundation.
The technical solution of the present invention is as follows: a kind of potential customers excavate and recommended method, comprising the following steps:
S1), the personal information and social activity data of user are obtained from social platform, and coarse original to what is got
Data are cleaned, the data unrelated with user information such as removal status code, while being formatted to text information, removal text
Spcial character in this, by the data cleaned storage into user social contact activity database;
S2), screened, chosen crucial according to the personal information of user, social behavior information and Shopping Behaviors information
Information architecture user portrait;
S3), characterization is carried out to various types of data, be translated into feature for model training to
Amount;
S4), the model basin that training, updating factor disassembler model, svm model and xgboost model form predicts user
To the interest of commodity, the feature vector for using user to draw a portrait and sort out is as the input of model basin, by the recent reality of user
Purchase result individually trains model as label, then passes through the parameter of gradient descent method more new model according to error,
The precision of prediction of lift scheme;
S5), test, adjust model, cross-beta is carried out to model using the data of another crowd of user different from training,
When test effect is higher than threshold value, then trained model is obtained, Feature Selection, the mode of adjusting parameter and according to step are otherwise taken
Rapid S4) re -training model;
S6), all users are predicted using trained model, obtains the newest interest distribution of user, and use
Direct cladding process updates user interest prediction result by weight additive process;
S7), according to the browsing of user record, shopping record and time factor, user is divided into frequent customer and Xin user,
New user will be used to excavate potential customers, and new user is then divided into potential customers using svm disaggregated model and is less likely to produce
The passerby of raw buying behavior, and potential customers are excavated from new user;
S8), training, update svm disaggregated model, use step S3) and step S6) in obtain feature as svm classification
The input of model, the practical conversion situation after recommending user is as label training svm disaggregated model, then according to accidentally
Difference passes through the parameter of stochastic gradient descent method more new model, the precision of prediction of lift scheme;
S9), model is tested and is adjusted, carries out cross-beta using the data of another crowd of user different from training user,
When test effect is higher than threshold value, then trained model is obtained, Feature Selection, the mode of adjusting parameter and according to step are otherwise taken
Rapid S8) re -training model;
S10), using S9) model of training predicts all new users, new user is divided into potential user and less
There may be the passerbys of buying behavior, and update new objective recommending data library;
S11), for the user of new access platform, judge whether user is potential customers, and from new objective recommending data library
The information of inquiry visiting user then screens the commodity being consistent with user interest if there is the corresponding interest information of user, generates quotient
Product list;
S12), the commodity in front end page displaying and Recommendations list, or commodity packaging is launched at advertisement accurately
To user.
Further, step S1) using crawler technology obtain from multiple social network sites obtain user personal information and
Social activity information.
Further, step S2) in, the userspersonal information mainly includes gender, age, region, work,
School, interest tags, user gradation, user's prestige, attention number, number of fans, the social behavior information include sending information, close
Topic, forwarding comment are infused, the Shopping Behaviors information includes browsing information, Information on Collection, purchase information, evaluation information.
Further, step S3) in, the feature vector includes user characteristics vector sum product features vector;It is described
User characteristics vector be made of individual subscriber feature, user social contact behavioural characteristic and user-commodity interaction feature, the quotient
Product feature vector is made of merchandise classification feature, shopping contextual feature and commodity-user interaction features.
Further, step S3) in, characterization is carried out to various types of data, specifically includes logarithm
Type data are normalized, and carry out one-hot discrete vector to class label type data, carry out word2vec to content of text
Term vector and doc2vec text vector carry out polishing to missing data, it is final obtain user vector expression, user social contact,
Shopping Behaviors vector indicates that commodity vector indicates.
Further, step S3) in, to gender, age, region, work, school, interest tags, user gradation, user
The userspersonal informations such as prestige, attention number, number of fans carry out vectorization processing, structuring user's personal characteristics, wherein gender,
Domain, age, operational data use one-hot vector representation, and the data of other value types normalize table using max-min
Show.
Further, step S3) in, the user social contact behavioural characteristic mainly includes the text sent in user social contact
Information, is spliced into a document for the text information sent in user's certain time, and document is segmented and gone by participle tool
Except stop words, corresponding term vector is then converted for each word by word2vec tool, and all term vectors are added
And normalize, the vector for obtaining corresponding document indicates;The corresponding document vector of document is generated by doc2vec tool simultaneously,
Finally two expression vectors are spliced, as user social contact behavioural characteristic vector.
Further, the user-commodity interaction feature refers to that the commodity by user's purchase are expressed as by one-hot
Vector form, that is, the product locations value bought are 1, and the positional value that do not bought is 0.
Further, the shopping contextual feature refers to record the shopping of user and successively carry out according to the time buying
Sequence, structuring user's buy commodity sequence, every commodity are regarded as a word, then by word2vec tool
Skip-gram model is trained, and can obtain the corresponding vector of every commodity, purchase of the vector as commodity after the completion of training
Object contextual feature.
Further, step S4) in, by Factorization machine model, svm model and the average knot of xgboost model prediction
Fruit is responsible for prediction user to the interest of commodity as final output.
Further, step S7), the potential customers refer to the people being likely to purchase, the passerby, which refers to, only to be browsed
But it is less likely the people of purchase commodity.
The invention has the benefit that
1, user includes a large amount of useful information in the personal information and behavioral data of social platform, uses use than other
Potential customers' digging system precision of family browsing behavior is higher, can more accurately judge user interest.
2, the user that commodity are browsed for logging in platform, can give its interested merchandise display according to prediction result
The user promotes conversion ratio.
3, for the user of browsing platform, advertisement dispensing can be carried out to it according to prediction result, realizes client's conversion.
4, frequent customer can also be provided and is targetedly recommended, increase client's stickiness.
Detailed description of the invention
Fig. 1 is flow diagram of the invention;
Fig. 2 is the training flow chart of potential customers' mining model of the present invention;
Fig. 3 is the flow chart that potential customers of the present invention identify and convert.
Specific embodiment
Specific embodiments of the present invention will be further explained with reference to the accompanying drawing:
As shown in Figure 1, a kind of potential customers excavate and recommended method, comprising the following steps:
S1), the personal information and social activity data of user are obtained from social platform using crawler technology, and to acquisition
To coarse initial data cleaned, removal status code etc. data unrelated with user information, while text information is carried out
Format, remove text in spcial character, by the data cleaned storage into user social contact activity database, and periodically from
Corresponding data, the data in real-time update database are obtained in social platform;
S2), screened, chosen crucial according to the personal information of user, social behavior information and Shopping Behaviors information
Information architecture user portrait, wherein the userspersonal information mainly includes gender, age, region, work, school, interest
Label, user gradation, user's prestige, attention number, number of fans, the social behavior information include sending information, concern topic,
The information such as forwarding comment, the Shopping Behaviors information include the letters such as browsing information, Information on Collection, purchase information, evaluation information
Breath;
S3), characterization is carried out to various types of data, be translated into feature for model training to
Amount, specifically includes logarithm type data and is normalized, and one-hot discrete vector is carried out to class label type data, to text
This content carries out word2vec term vector and doc2vec text vector, carries out polishing to missing data, finally obtains user
Vector expression, user social contact, Shopping Behaviors vector indicate that commodity vector indicates, wherein the feature vector includes user spy
Levy vector sum product features vector;The user characteristics vector is by individual subscriber feature, user social contact behavioural characteristic and user-
Commodity interaction feature composition, the product features vector are handed over by merchandise classification feature, shopping contextual feature and commodity-user
Mutual feature composition.
To gender, age, region, work, school, interest tags, user gradation, user's prestige, attention number, number of fans etc.
Userspersonal information carries out vectorization processing, structuring user's personal characteristics, wherein tag types such as gender, region, age, work
Discrete data using one-hot vector indicate, such as [1,0] indicate gender be female, [0,1] indicate gender be male, [0,0] then
Indicate gender it is unknown), wherein the age press one section of every 10 fraction of the year, the data of remaining value type for example user gradation, attention number,
Number of fans etc. uses max-min normalization and indicates.
User social contact behavioural characteristic mainly contains the text information sent in user social contact, and user nearest 3 months are sent out
The text information sent is spliced into a document, and document is segmented to by participle tool and removed stop words, is then passed through
Each word is converted corresponding 100 dimension term vector by word2vec tool, and all term vector phase adductions are normalized, and obtains
The vector of corresponding document indicates;The corresponding document vector of document is generated by doc2vec tool simultaneously, finally indicates two
Vector splicing, as user social contact behavioural characteristic vector.
The commodity that user buys are expressed as vector form by one-hot, that is, the product locations value bought is 1, is not had
Having the positional value bought is 0, as user-commodity interaction feature.
It inquires and screens when in order to facilitate user's shopping, commodity can be usually divided into different classifications, such as dress ornament, food
Product etc. indicate the corresponding category feature of commodity using one-hot representation.
The shopping of user is recorded and is successively ranked up according to the time buying, structuring user's buy commodity sequence, by every
Commodity are regarded as a word, are then trained by the skip-gram model in word2vec tool, can be with after the completion of training
The corresponding vector of every commodity is obtained, the shopping contextual feature as commodity.And user is for the fancy grade of every commodity
True tag, constructed by the history of user shopping record, if user bought the commodity before, commodity are corresponding
Value is 1;If user on the contrary did not buy the commodity, the corresponding value of commodity is 0.
S4), the model basin that training, updating factor disassembler model, svm model and xgboost model form predicts user
To the interest of commodity, the commodity vector of user vector and every commodity is spliced into the input as model basin, by a collection of feature to
Amount is separately input in 3 models, and is scored using hobby of the model prediction user to commodity, then according to user's physical tags
Value, calculates separately the difference between model prediction result and physical tags result, by batch gradient descent algorithm, to 3 models
Parameter be updated, in such a way that continuous iteration updates, step up model prediction accuracy so that the predicted value of model by
It walks approaching to reality and buys situation, training flow chart is referring to fig. 2.
S5), test, adjust model, cross-beta is carried out to model using the data of another crowd of user different from training,
When test effect is higher than threshold value, then trained model is obtained, Feature Selection, the mode of adjusting parameter and according to step are otherwise taken
Rapid S4) re -training model;
S6), all users are predicted using trained model, obtains the newest interest distribution of user, and use
Direct cladding process updates user interest prediction result by weight additive process;
S7), according to the browsing of user record, shopping record and time factor, user is divided into frequent customer and Xin user,
New user will be used to excavate potential customers, and new user is then divided into potential customers and passerby using svm disaggregated model, and from new
Excavate potential customers in user, and it is pre- every some cycles using svm disaggregated model classify to all new addition users
It surveys;
S8), training, update svm disaggregated model, use step S3) and step S6) in obtain feature as svm classification
The input of model, the practical conversion situation after recommending user is as label training svm disaggregated model, then according to accidentally
Difference passes through the parameter of stochastic gradient descent method more new model, the precision of prediction of lift scheme;
S9), model is tested and is adjusted, carries out cross-beta using the data of another crowd of user different from training user,
When test effect is higher than threshold value, then trained model is obtained, Feature Selection, the mode of adjusting parameter and according to step are otherwise taken
Rapid S8) re -training model;
S10), using S9) model of training predicts all new users, new user is divided into potential user and less
There may be the passerbys of buying behavior, and update new objective recommending data library;
S11), for the user of new access platform, judge whether user is potential customers, and from new objective recommending data library
The information of inquiry visiting user then screens the commodity being consistent with user interest if there is the corresponding interest information of user, generates quotient
Product list;
S12), the commodity in front end page displaying and Recommendations list, or commodity packaging is launched at advertisement accurately
It is specific as shown in Figure 3 to user.
The above embodiments and description only illustrate the principle of the present invention and most preferred embodiment, is not departing from this
Under the premise of spirit and range, various changes and improvements may be made to the invention, these changes and improvements both fall within requirement and protect
In the scope of the invention of shield.
Claims (10)
1. a kind of potential customers excavate and recommended method, which comprises the following steps:
S1), the personal information and social activity data of user are obtained from social platform, and to the coarse initial data got
It is cleaned, the removal status code data unrelated with user information, while text information is formatted, removed in text
Spcial character, by the data cleaned storage into user social contact activity database;
S2), screened according to the personal information of user, social behavior information and Shopping Behaviors information, choose key message
Construct user's portrait;
S3), characterization is carried out to various types of data, is translated into the feature vector for model training;
S4), the model basin that training, updating factor disassembler model, svm model and xgboost model form predicts user to quotient
The interest of product, the feature vector for using user to draw a portrait and sort out is as the input of model basin, by the recent actual purchase of user
As a result model is individually trained as label, then passes through the parameter of gradient descent method more new model according to error, promoted
The precision of prediction of model;
S5), test, adjust model, cross-beta is carried out to model using the data of another crowd of user different from training, works as survey
It tries effect and is higher than threshold value, then obtain trained model, otherwise take Feature Selection, the mode of adjusting parameter and according to step
S4) re -training model;
S6), all users are predicted using trained model, obtains the newest interest distribution of user, and using direct
Cladding process updates user interest prediction result by weight additive process;
S7), according to the browsing of user record, shopping record and time factor, user is divided into frequent customer and Xin user, it is new to use
Family will be used to excavate potential customers, and new user is then divided into potential customers using svm disaggregated model and is less likely to produce purchase
The passerby for buying behavior excavates potential client from new user;
S8), training, update svm disaggregated model, use step S3) and step S6) in acquisition feature as svm disaggregated model
Input, the practical conversion situation after recommending user is then logical according to error as label training svm disaggregated model
Cross the parameter of stochastic gradient descent method more new model, the precision of prediction of lift scheme;
S9), model is tested and adjusted, cross-beta is carried out using the data of another crowd of user different from training user, works as survey
It tries effect and is higher than threshold value, then obtain trained model, otherwise take Feature Selection, the mode of adjusting parameter and according to step
S8) re -training model;
S10), using S9) training model all new users are predicted, by new user be divided into potential user and be less likely
The passerby of buying behavior is generated, and updates new objective recommending data library;
S11), for the user of new access platform, judge whether user is potential customers, and inquired from new objective recommending data library
The information of visiting user then screens the commodity being consistent with user interest if there is the corresponding interest information of user, generates commodity column
Table;
S12), the commodity in front end page displaying and Recommendations list, or commodity packaging is delivered to use at advertisement accurately
Family.
2. a kind of potential customers according to claim 1 excavate and recommended method, it is characterised in that: step S2) in, it is described
Userspersonal information mainly include gender, the age, region, work, school, interest tags, user gradation, user's prestige, close
Number, number of fans are infused, the social behavior information mainly includes sending information, concern topic, forwarding comment, the shopping row
It include browsing information, Information on Collection, purchase information, evaluation information for information.
3. a kind of potential customers according to claim 1 excavate and recommended method, it is characterised in that: step S3) in, to each
The different types of data of kind carry out characterization, specifically include logarithm type data and are normalized, to class label type number
According to one-hot discrete vector is carried out, word2vec term vector and doc2vec text vector are carried out to content of text, it is right
Missing data carries out polishing, final to obtain user vector expression, user social contact, the expression of Shopping Behaviors vector, and commodity vector indicates.
4. a kind of potential customers according to claim 1 excavate and recommended method, it is characterised in that: step S3) in, it is described
Feature vector include user characteristics vector sum product features vector;The user characteristics vector is by individual subscriber feature, use
Family Social behaviors feature and user-commodity interaction feature composition, the product features vector is by merchandise classification feature, shopping
Following traits and commodity-user interaction features composition.
5. a kind of potential customers according to claim 4 excavate and recommended method, it is characterised in that: step S3) in, to property
Not, the age, region, work, school, interest tags, user gradation, user's prestige, attention number, number of fans userspersonal information into
Row vectorization processing, structuring user's personal characteristics, wherein gender, region, age, operational data use one-hot vector table
Show method, the data of other value types are indicated using max-min normalization.
6. a kind of potential customers according to claim 4 excavate and recommended method, it is characterised in that: step S3) in, it is described
User social contact behavioural characteristic mainly include the text information sent in user social contact, the text that will be sent in user's certain time
Information is spliced into a document, and document is segmented to by participle tool and removed stop words, and then passing through word2vec tool will
Each word is converted into corresponding term vector, and all term vector phase adductions are normalized, and obtains the vector table of corresponding document
Show;The corresponding document vector of document is generated by doc2vec tool simultaneously, is finally spliced two expression vectors, as user
Social behaviors feature vector.
7. a kind of potential customers according to claim 4 excavate and recommended method, it is characterised in that: the user-quotient
Product interaction feature refers to that the commodity by user's purchase are expressed as vector form by one-hot, that is, the product locations value bought
It is 1, the positional value that do not bought is 0.
8. a kind of potential customers according to claim 4 excavate and recommended method, it is characterised in that: above and below the shopping
Literary feature refers to record the shopping of user to be successively ranked up according to the time buying, and structuring user's buy commodity sequence, will be every
Part commodity are regarded as a word, are then trained by the skip-gram model in word2vec tool, can after the completion of training
To obtain the corresponding vector of every commodity, shopping contextual feature of the vector as commodity.
9. a kind of potential customers according to claim 1 excavate and recommended method, it is characterised in that: step S4) in, by because
Sub- disassembler model, svm model and the average result of xgboost model prediction are responsible for prediction user as final output
To the interest of commodity.
10. a kind of potential customers according to claim 1 excavate and recommended method, it is characterised in that: step S4) in, it uses
Family constructs, if before user the true tag of the fancy grade of every commodity by the history shopping record of user
The commodity were bought, then the corresponding value of commodity is 1;If user on the contrary did not buy the commodity, the corresponding value of commodity is 0,
It predicts that the commodity vector of user vector and every commodity is spliced the input as model basin to the interest of commodity by user, and makes
Scored with model prediction user the hobbies of commodity, then according to user's physical tags value, calculate separately model prediction result with
Difference between physical tags result is updated the parameter of 3 models, by batch gradient descent algorithm by constantly changing
The mode that generation updates steps up model prediction accuracy.
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 true CN110222272A (en) | 2019-09-10 |
CN110222272B 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) |
Cited By (38)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110889716A (en) * | 2019-09-29 | 2020-03-17 | 清华大学 | Method and device for identifying potential registered user |
CN110933472A (en) * | 2019-12-02 | 2020-03-27 | 深圳市云积分科技有限公司 | Method and device for realizing video recommendation |
CN111127074A (en) * | 2019-11-26 | 2020-05-08 | 杭州聚效科技有限公司 | Data recommendation method |
CN111144986A (en) * | 2019-12-25 | 2020-05-12 | 清华大学 | Commodity recommendation method and device for social e-commerce website 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 |
CN111199421A (en) * | 2019-12-20 | 2020-05-26 | 北京淇瑀信息科技有限公司 | User recommendation method and device based on social relationship and electronic equipment |
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 |
CN111598256A (en) * | 2020-05-18 | 2020-08-28 | 北京互金新融科技有限公司 | Method and device for processing default purchase behavior of target customer |
CN111651686A (en) * | 2019-09-24 | 2020-09-11 | 北京嘀嘀无限科技发展有限公司 | Test processing method and device, electronic equipment and storage medium |
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 |
CN112016003A (en) * | 2020-08-19 | 2020-12-01 | 重庆邮电大学 | Social crowd user tag mining and similar user recommending method based on CNN |
CN112070615A (en) * | 2020-09-02 | 2020-12-11 | 中国银行股份有限公司 | Financial product recommendation method and device based on knowledge graph |
CN112200601A (en) * | 2020-09-11 | 2021-01-08 | 深圳市法本信息技术股份有限公司 | Item recommendation method and device and readable storage medium |
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 |
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 |
CN112288549A (en) * | 2020-11-18 | 2021-01-29 | 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 |
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 |
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 |
CN112561555A (en) * | 2019-09-26 | 2021-03-26 | 北京国双科技有限公司 | Product data processing method and device |
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 |
WO2021135842A1 (en) * | 2020-01-02 | 2021-07-08 | ***通信有限公司研究院 | Method and apparatus for identifying dissatisfied users in group, device, and storage 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 |
CN113393271A (en) * | 2021-06-15 | 2021-09-14 | 湖南汽车工程职业学院 | Product customer big data application matching system and computer storage medium |
CN113744002A (en) * | 2020-05-27 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Method, device, equipment and computer readable medium for pushing information |
CN113822596A (en) * | 2021-10-12 | 2021-12-21 | 深圳市我们在线教育有限公司 | Customer screening method based on big data |
CN114187036A (en) * | 2021-11-30 | 2022-03-15 | 深圳市喂车科技有限公司 | Internet advertisement intelligent recommendation management system based on behavior characteristic recognition |
CN114267457A (en) * | 2021-12-03 | 2022-04-01 | 爱优牙信息技术(深圳)有限公司 | Dentist service platform matched with oriented client |
CN114387064A (en) * | 2022-01-13 | 2022-04-22 | 福州大学 | E-commerce platform potential customer recommendation method and system based on comprehensive similarity |
CN114595323A (en) * | 2022-03-04 | 2022-06-07 | 北京百度网讯科技有限公司 | Portrait construction, recommendation, model training method, apparatus, device and storage medium |
CN114841760A (en) * | 2022-06-30 | 2022-08-02 | 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 |
CN116308464A (en) * | 2023-05-11 | 2023-06-23 | 广州钛动科技股份有限公司 | 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 |
CN116523572A (en) * | 2023-06-28 | 2023-08-01 | 悦享星光(北京)科技有限公司 | Client mining method and system based on client behavior characteristics |
CN116894692A (en) * | 2023-09-11 | 2023-10-17 | 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 |
CN116957691A (en) * | 2023-09-19 | 2023-10-27 | 翼果(深圳)科技有限公司 | Cross-platform intelligent advertisement putting method and system for commodities of e-commerce merchants |
CN117035853A (en) * | 2023-10-09 | 2023-11-10 | 销生客(北京)数字科技有限公司 | Potential customer identity marking system based on big data |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
US20160267498A1 (en) * | 2015-03-10 | 2016-09-15 | Wipro Limited | Systems and methods for identifying new users using trend analysis |
WO2017157146A1 (en) * | 2016-03-15 | 2017-09-21 | 平安科技(深圳)有限公司 | User portrait-based personalized recommendation method and apparatus, server, and storage medium |
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 |
CN109376766A (en) * | 2018-09-18 | 2019-02-22 | 平安科技(深圳)有限公司 | A kind of portrait prediction classification method, device and equipment |
-
2019
- 2019-04-18 CN CN201910311200.1A patent/CN110222272B/en active Active
Patent Citations (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 |
WO2017157146A1 (en) * | 2016-03-15 | 2017-09-21 | 平安科技(深圳)有限公司 | User portrait-based personalized recommendation method and apparatus, server, and storage medium |
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 |
CN109376766A (en) * | 2018-09-18 | 2019-02-22 | 平安科技(深圳)有限公司 | A kind of portrait prediction classification method, device and equipment |
Non-Patent Citations (1)
Title |
---|
李飞等: "基于决策树C4.5算法的大数据保险业模型研究", 《中国市场》 * |
Cited By (59)
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 |
CN111651686A (en) * | 2019-09-24 | 2020-09-11 | 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 |
CN111127074A (en) * | 2019-11-26 | 2020-05-08 | 杭州聚效科技有限公司 | Data recommendation method |
CN111127074B (en) * | 2019-11-26 | 2023-04-25 | 杭州聚效科技有限公司 | Data recommendation method |
CN110933472A (en) * | 2019-12-02 | 2020-03-27 | 深圳市云积分科技有限公司 | Method and device for realizing video recommendation |
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 |
CN111199421A (en) * | 2019-12-20 | 2020-05-26 | 北京淇瑀信息科技有限公司 | User recommendation method and device based on social relationship and electronic equipment |
CN111177581A (en) * | 2019-12-25 | 2020-05-19 | 清华大学 | Multi-platform-based social e-commerce website commodity recommendation method and device |
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 |
CN111144986A (en) * | 2019-12-25 | 2020-05-12 | 清华大学 | Commodity recommendation method and device for social e-commerce website based on sharing behavior |
WO2021135842A1 (en) * | 2020-01-02 | 2021-07-08 | ***通信有限公司研究院 | Method and apparatus for identifying dissatisfied users in group, device, 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 |
CN111598256A (en) * | 2020-05-18 | 2020-08-28 | 北京互金新融科技有限公司 | Method and device for processing default purchase behavior of target customer |
CN113744002A (en) * | 2020-05-27 | 2021-12-03 | 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 |
CN112016003A (en) * | 2020-08-19 | 2020-12-01 | 重庆邮电大学 | Social crowd user tag mining and similar user recommending method based on CNN |
CN112070615A (en) * | 2020-09-02 | 2020-12-11 | 中国银行股份有限公司 | Financial product recommendation method and device based on knowledge graph |
CN112200601A (en) * | 2020-09-11 | 2021-01-08 | 深圳市法本信息技术股份有限公司 | Item recommendation method and device and readable storage medium |
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 |
CN112288549A (en) * | 2020-11-18 | 2021-01-29 | 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 |
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 |
CN113177151A (en) * | 2021-05-28 | 2021-07-27 | 中山世达模型制造有限公司 | Potential customer screening method |
WO2022246923A1 (en) * | 2021-05-28 | 2022-12-01 | 中山世达模型制造有限公司 | Method for screening potential customer |
CN113379529A (en) * | 2021-06-07 | 2021-09-10 | 广发银行股份有限公司 | Collaborative decision engine application framework |
CN113393271A (en) * | 2021-06-15 | 2021-09-14 | 湖南汽车工程职业学院 | Product customer big data application matching system and computer storage medium |
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 |
CN113822596A (en) * | 2021-10-12 | 2021-12-21 | 深圳市我们在线教育有限公司 | Customer screening method based on big data |
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 |
CN114187036A (en) * | 2021-11-30 | 2022-03-15 | 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 |
CN114267457A (en) * | 2021-12-03 | 2022-04-01 | 爱优牙信息技术(深圳)有限公司 | Dentist service platform matched with oriented 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 |
CN114387064A (en) * | 2022-01-13 | 2022-04-22 | 福州大学 | E-commerce platform potential customer recommendation method and system based on comprehensive similarity |
CN114595323A (en) * | 2022-03-04 | 2022-06-07 | 北京百度网讯科技有限公司 | Portrait construction, recommendation, model training method, apparatus, device and storage medium |
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 |
CN114841760A (en) * | 2022-06-30 | 2022-08-02 | 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 |
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 |
CN116308464A (en) * | 2023-05-11 | 2023-06-23 | 广州钛动科技股份有限公司 | Target client acquisition system and method |
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 |
CN116523572A (en) * | 2023-06-28 | 2023-08-01 | 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 |
CN116894692A (en) * | 2023-09-11 | 2023-10-17 | 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 |
CN116957691A (en) * | 2023-09-19 | 2023-10-27 | 翼果(深圳)科技有限公司 | Cross-platform intelligent advertisement putting method and system for commodities of e-commerce merchants |
CN117035853A (en) * | 2023-10-09 | 2023-11-10 | 销生客(北京)数字科技有限公司 | Potential customer identity marking system based on big data |
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 |
Also Published As
Publication number | Publication date |
---|---|
CN110222272B (en) | 2022-10-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110222272A (en) | A kind of potential customers excavate and recommended method | |
Liu et al. | Analyzing changes in hotel customers’ expectations by trip mode | |
Kirby-Hawkins et al. | An investigation into the geography of corporate e-commerce sales in the UK grocery market | |
US20210158187A1 (en) | System and method for detecting friction in websites | |
Li et al. | Is peer evaluation of consumer online reviews socially embedded?–An examination combining reviewer’s social network and social identity | |
Kuo et al. | Application of quality function deployment to improve the quality of Internet shopping website interface design | |
Luo | Analyzing the impact of social networks and social behavior on electronic business during COVID-19 pandemic | |
CN103886074B (en) | Commercial product recommending system based on social media | |
CN103246980B (en) | Information output method and server | |
CN105488697A (en) | Potential customer mining method based on customer behavior characteristics | |
CN106327227A (en) | Information recommendation system and information recommendation method | |
CN102968506A (en) | Personalized collaborative filtering recommendation method based on extension characteristic vectors | |
Rezaeinia et al. | Recommender system based on customer segmentation (RSCS) | |
WO2022095701A1 (en) | Method and device for recommending objects, equipment, and storage medium | |
CN104615721B (en) | For the method and system based on return of goods related information Recommendations | |
Mei et al. | Overview of Web mining technology and its application in e-commerce | |
CN104408648A (en) | Method and device for choosing items | |
CN104574153A (en) | Method for quickly applying negative sequence mining patterns to customer purchasing behavior analysis | |
CN104537553A (en) | Application of repeated negative sequence pattern in customer purchase behavior analysis | |
Niu et al. | Product hierarchy-based customer profiles for electronic commerce recommendation | |
Lin et al. | A consumer review-driven recommender service for web e-commerce | |
CN109919728A (en) | A kind of Method of Commodity Recommendation in difference quotient city | |
Čavoški et al. | Agent-based modelling and simulation in the analysis of customer behaviour on B2C e-commerce sites | |
Faraone et al. | Using context to improve the effectiveness of segmentation and targeting in e-commerce | |
Nurmi et al. | Promotionrank: Ranking and recommending grocery product promotions using personal shopping lists |
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 |