CN110390098A - Method, apparatus, equipment and the storage medium of data dispensing side are chosen based on big data - Google Patents
Method, apparatus, equipment and the storage medium of data dispensing side are chosen based on big data Download PDFInfo
- Publication number
- CN110390098A CN110390098A CN201910527849.7A CN201910527849A CN110390098A CN 110390098 A CN110390098 A CN 110390098A CN 201910527849 A CN201910527849 A CN 201910527849A CN 110390098 A CN110390098 A CN 110390098A
- Authority
- CN
- China
- Prior art keywords
- product
- similarity
- target user
- content
- dispensing side
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention discloses a kind of method, apparatus, equipment and storage medium that data dispensing side is chosen based on big data, and wherein method includes: the user information for receiving multiple first object users and the product information of multiple first object products;Second target user is successively chosen, the scoring of the first product is obtained by default code of points, the content of pages of data dispensing side is obtained in a manner of simulating second target user's specific behavior;Calculate the first similarity of content of pages and multiple product informations, choose corresponding second target product of highest second similarity of similarity, the second product scoring that the second target product is obtained in the scoring of the first product obtains data dispensing side by preset formula and scores the effect of the second target user;It is added up each effect scoring of data dispensing side to obtain the effect score of data dispensing side;Choose the data dispensing side of effect highest scoring.Present invention seek to address that finding the problem for more meeting the better data dispensing side of own service demand, effect.
Description
Technical field
The present invention relates to big data technical field, especially relate to a kind of choose data dispensing side based on big data
Method, apparatus, equipment and storage medium.
Background technique
When doing product advertising dispensing, how needs to find the better ad data dispensing side of effect and cooperate
The advertisement delivery effect of inspection data dispensing side is usually to lean on public praise and in the industry influence power, these data are all third party's offers
, there is certain deviation, it may be also less big with the business correlation of itself.Therefore, there is presently a need to find more meet itself industry
The better data dispensing side of business demand, effect is a problem to be solved.
Summary of the invention
The main object of the present invention be provide it is a kind of based on big data choose the method, apparatus of data dispensing side, equipment and
Storage medium, it is intended to solve the problems, such as to find and more meet the better data dispensing side of own service demand, effect.
In order to achieve the above-mentioned object of the invention, the present invention proposes a kind of method for choosing data dispensing side based on big data, packet
It includes:
The user information of multiple first object users and the product information of multiple first object products are received, user information is extremely
It less include that user draws a portrait;
From multiple first object users choose second target user, according to the user information of the second target user with
The product information of multiple first object products obtains the second target user to multiple first object products according to default code of points
The scoring of the first product, and obtain in a manner of simulating second target user's specific behavior the content of pages of data dispensing side,
Specific behavior includes at least the second target user and carries out behavior relevant to pre-set business;
The first similarity of the product information of content of pages and multiple first object products is calculated, is obtained in the first similarity
Highest second similarity of similarity is chosen the second target corresponding with the second similarity in multiple first object products and is produced
Product obtain the second target user in the scoring of the first product and score the second product of the second target product, similar according to second
Degree and the scoring of the second product obtain data dispensing side by preset formula and score the effect of the second target user;
Data dispensing side adds up to the effect scoring of each second target user, according to cumulative point of effect scoring
Value obtains the effect score of data dispensing side, and effect score is used to evaluate the dispensing effect of data dispensing side;
The effect score of correlation data dispensing side chooses the data dispensing side of effect highest scoring.
Further, product information includes at least product risks, product income, product sales volume, promotion efficiency, according to second
The relationship of the user information of target user and the product information of multiple first object products, obtains second according to default code of points
In the step of target user scores to the first product of multiple first object products, comprising:
According to user's portrait and product risks, multiple first object products are divided into proper product and improper product;
Respectively by proper product and improper product according to product income descending sort, wherein being pressed if product income is identical
According to product sales volume descending sort, if product sales volume is identical not according to promotion efficiency descending sort and product Income Maximum
The sequence of proper product is lower than the smallest proper product of product income, obtains the appropriate degree sequence of first object product;
Descending number is carried out to multiple first object products according to the appropriate degree collating sequence of target product, by product number
It is denoted as the product scoring of first object product, the second target user is obtained and scores the first product of multiple first object products.
Further, and in a manner of simulating second target user's specific behavior the content of pages of data dispensing side is obtained
The step of in, comprising:
The user behavior characteristics of the history of the second target user are obtained from system database;
User behavior characteristics are analyzed, the specific behavior of the second target user is obtained;
According to the specific behavior of the second target user, data throwing is obtained in a manner of simulating second target user's specific behavior
The content of pages for the side of putting.
Further, the step of calculating the first similarity of the product information of content of pages and multiple first object products it
Before, further includes:
Judge whether content of pages is related to pre-set business;
If unrelated with pre-set business, data dispensing side is denoted as zero to the effect scoring of the second target user;
If related to pre-set business, enter the first of the product information for calculating content of pages and multiple first object products
The step of similarity.
Further, product information includes the product content of first object product, calculates content of pages and multiple first mesh
The first similarity of the product information of product is marked, highest second similarity of similarity in the first similarity is obtained, multiple the
The second target product corresponding with the second similarity is chosen in one target product, is obtained the second target in the scoring of the first product and is used
Family scores to the second product of the second target product, is scored according to the second similarity and the second product, is counted by preset formula
In the step of scoring according to effect of the dispensing side to the second target user, comprising:
Word content is converted by content of pages;
It obtains product content from the product information of multiple first object products, calculates the of product content and word content
One similarity;
Obtain highest second similarity of similarity in the first similarity;
The second target product corresponding with the second similarity is chosen in multiple first object products;
The second target user is obtained in the scoring of the first product to score to the second product of the second target product;
It scores to obtain data dispensing side multiplied by the second product for the second similarity to score to the effect of the second target user.
Further, product content is the first product keyword, calculates the step of the similarity of product content and word content
In rapid, comprising:
First product keyword is matched one by one with word content, obtains the second product keyword identical with word content
Quantity;
The second product keyword quantity and the first product keyword ratio of number are calculated, product content and word content are obtained
Similarity.
Further, the first product keyword is matched one by one with word content, obtains identical with word content second
In the step of product keyword quantity, comprising:
Word content is subjected to word segmentation processing;
First product keyword is matched with each participle, obtain and segments identical second product keyword;
Count the second product keyword quantity.
The present invention also proposes a kind of device that data dispensing side is chosen based on big data, comprising:
Receiving module, for receiving the user information of multiple first object users and the product letter of multiple first object products
Breath, user information are drawn a portrait including at least user;
First grading module, for choosing second target user from multiple first object users, according to the second mesh
The user information of user and the product information of multiple first object products are marked, obtains the second target user according to default code of points
The first product scoring to multiple first object products, and data are obtained in a manner of simulating second target user's specific behavior
The content of pages of dispensing side, specific behavior include at least the second target user and carry out behavior relevant to pre-set business;
Computing module, the first similarity of the product information for calculating content of pages and multiple first object products, is obtained
Highest second similarity of similarity in the first similarity is taken, is chosen in multiple first object products corresponding with the second similarity
The second target product, obtain the second target user in the scoring of the first product and score the second product of the second target product,
It is scored according to the second similarity and the second product, data dispensing side is obtained by preset formula, the effect of the second target user is commented
Point;
Accumulator module, for data dispensing side to add up to the effect scoring of each second target user, according to effect
The cumulative score value of fruit scoring obtains the effect score of data dispensing side, and effect score is used to evaluate the dispensing effect of data dispensing side
Fruit;
Module is chosen, for the effect score of correlation data dispensing side, chooses the data dispensing side of effect highest scoring.
The present invention also proposes that a kind of equipment, including memory and processor, the memory are stored with computer-readable finger
It enables, when the processor executes the computer-readable instruction the step of realization any of the above-described method.
The present invention also provides a kind of computer non-volatile readable storage mediums, are stored thereon with computer-readable instruction,
The step of any of the above-described method is realized when the computer-readable instruction is executed by processor.
The present invention is based on having the beneficial effect that for the method, apparatus of big data selection data dispensing side, equipment and storage medium
Product scoring is carried out to multiple target products respectively by each target user, and is thrown in a manner of modelling customer behavior in data
The side of putting obtains multiple content of pages, and multiple content of pages are carried out similarity calculation with multiple target products respectively, obtain data
Dispensing side scores to the effect of each target user, and effect scoring is added up and obtains the effect score of data dispensing side, according to number
According to the effect score of dispensing side, it can be determined that whether the advertisement serving policy of data dispensing side is more in line with us to target user
Expection, launched to more meet the better data of own service demand, effect according to the selection of the effect score of data dispensing side
Side.
Detailed description of the invention
Fig. 1 is the step schematic diagram that the method for data dispensing side is chosen the present invention is based on big data;
Fig. 2 is the flow diagram that the device of data dispensing side is chosen the present invention is based on big data;
Fig. 3 is the structural schematic block diagram of one embodiment of present device.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
Referring to Fig.1, a method of data dispensing side is chosen based on big data, comprising:
The product information of S1, the user information for receiving multiple first object users and multiple first object products, Yong Huxin
Breath includes at least user and draws a portrait;
S2, second target user is chosen from multiple first object users, believed according to the user of the second target user
The product information of breath and multiple first object products, obtains the second target user to multiple first objects according to default code of points
First product of product scores, and in the page for obtaining data dispensing side in a manner of simulating second target user's specific behavior
Hold, specific behavior includes at least the second target user and carries out behavior relevant to pre-set business;
It is similar to obtain first by S3, the first similarity for calculating content of pages with the product information of multiple first object products
Highest second similarity of similarity in degree chooses the second target corresponding with the second similarity in multiple first object products
Product obtains the second target user in the scoring of the first product and scores the second product of the second target product, according to the second phase
Like degree and the scoring of the second product, data dispensing side is obtained by preset formula and is scored the effect of the second target user;
S4, data dispensing side adds up to the effect scoring of each second target user, according to the tired of effect scoring
Bonus point value obtains the effect score of data dispensing side, and effect score is used to evaluate the dispensing effect of data dispensing side;
S5, correlation data dispensing side effect score, choose effect highest scoring data dispensing side.
In above-mentioned steps S1, by taking loan transaction as an example, filtered out in the user for having applied for loan transaction first more
A first object user, such as representative, generality the first object of a batch is picked out from customer data base by expert
User, multiple first object users can be the user of magnanimity, and user information may include the history that user draws a portrait, user is passing
Behavioural characteristic data etc., while multiple target products are chosen, target product can be loan financial product to be promoted, that is, need
The loan financial product for carrying out advertisement dispensing is also possible to multiple representative, generality loan finance that expert selects
Product, the product information of target product may include product content, product risks, product income, product sales volume, promote sequence
Deng.
In above-mentioned steps S2, second target user is successively chosen from the multiple first object users received,
For each second target user of selection, according to user information and product information, the second mesh is obtained by default code of points
It marks user to score to the product of multiple first object products, that is, it is expected that the second target user sees the phase of multiple first object products
Prestige value such as carries out appropriate degree sequence to multiple first object products according to user information and product information, that is, it is expected the second target
The appropriate degree for the first object product that user sees sorts, and the first object product representative of most suitable second target user most it is expected
The first object product that second target user sees, then the scoring highest of the first object product, obtains the second target with this and uses
It scores the first product of multiple first object products at family.Data throwing is obtained in a manner of simulating second target user's specific behavior
The content of pages for the side of putting, wherein specific behavior includes at least the second target user and carries out behavior relevant to pre-set business, for example,
Our preset business are loans, and the second target user is easy application loan when last year double 11 purchase things, then
The user that qualification similar with the second target user can be simulated in data dispensing side carries out similar behaviour in the similar time
Make, obtains the content of pages of data dispensing side, see whether data dispensing side launches the advertisement of loan product, the loan product of dispensing
It is which product, i.e., the second target user of artificial intelligence when data dispensing side carries out loan operation, throw by data dispensing side
Which kind of loan product what is put is, data dispensing side is the advertisement dispensing channel that product advertising is launched in selection.
Further, product information includes at least product risks, product income, product sales volume, promotion efficiency, in step S2
In, comprising:
S21, according to user portrait and product risks, multiple first object products are divided into proper product and improper production
Product;
S22, respectively by proper product and improper product according to product income descending sort, wherein if product income is identical
Then according to product sales volume descending sort, according to promotion efficiency descending sort and product Income Maximum if product sales volume is identical
Improper product sequence be lower than the smallest proper product of product income, obtain first object product appropriate degree sequence;
S23, descending number is carried out to multiple first object products according to appropriate degree collating sequence, product number is denoted as the
The product of one target product scores, and obtains the second target user and scores the first product of multiple first object products.
In above-mentioned steps S21, user's portrait is taken out according to information such as user's social property, living habit and consumer behaviors
As the user model of the labeling gone out, the content of user's portrait may include gender, age, occupation, repaying ability, value
See etc., user's portrait, which can be, calls directly well-established relevant to target user user portrait, according to user's portrait and
Product risks are divided into proper product and improper product to target product, such as user's portrait of target user shows that target is used
Family is that repaying ability is strong, then properly pushes the high product of risk, then the high first object product of product risks is divided into conjunction
Suitable product, the opposite low product of risk that pushes is exactly less properly, the low first object product of risk to be divided into improper
Product etc..
In above-mentioned steps S22, the target product for being divided into proper product is individually ranked up, while will be also divided into
The target product of improper product is individually ranked up, and is such as set to high target product there are two product risks, then according to target
The product income of product sorts, income it is high come front, the identical then comparative product sales volume of income, product sales volume is high to be come
Front, product sales volume is identical, then is sorted according to promotion efficiency, promotion efficiency it is big come front, promotion efficiency is to preset,
It is to stress the popularization of multiple first object products, the first object model sequencing for being divided into proper product is then come into division
Before the first object product of improper product, i.e., the sequence of the improper product of product Income Maximum is lower than product income
The smallest proper product obtains the second target user with this and sorts for the appropriate degree of multiple first object products, that is, the phase
The second target user is hoped to see the sequence of first object product in the advertisement of dispensing.
In above-mentioned steps S23, for sorted multiple first object products, the first object of foremost is come
Product is that most suitable second target user sees that then the scoring of product is also highest, such as has three sections of target products, is come
The number of the first first object product is three, then the second target user scores also just for the product of the first object product
It is three, the product number for coming last first object product is one, production of second target user for the first object product
Point namely one is judged, the second target user is obtained with this and is scored the first product of multiple first object products.
Further, in step s 2, comprising:
S24, obtained from system database the second target user history user behavior characteristics;
S25, analysis user behavior characteristics, obtain the specific behavior of the second target user;
S26, the specific behavior according to second user obtain data throwing in a manner of simulating second target user's specific behavior
The content of pages for the side of putting.
In the present embodiment, user behavior is primarily referred to as user's behavior online in internet, mobile interchange, such as browses
Webpage, uses application software, network social intercourse behavior etc. at search record.Have in system database for different network datas
The independent user behavior data for collecting each target user in source is to form historical user's behavioural characteristic of user, such as collects and use
Website that family is accessed, search content, access time, web site contents, application program of mobile phone service condition etc..In system data
Historical user's behavioural characteristic data of the second target user are obtained in library, and the user behavior characteristics of acquisition are analyzed, really
Operation behavior of fixed second target user when applying for loan product before, such as the second target user is in last year double 11 purchases
Loan is applied for when vehicle, i.e. when the second target user progress behavior relevant to pre-set business (loan) is double 11
Buy vehicle, it can the specific behavior for obtaining the second target user buys vehicle when being double 11, according to the second obtained mesh
The specific behavior for marking user, is interacted by crawler software with the interface API of data dispensing side, and virtual presses in data dispensing side
It is operated according to second target user's specific behavior, then obtains the content of pages of data dispensing side.
In above-mentioned steps S3, calculate the content of pages of acquisition and the product information of multiple target products first is similar
Degree, analysis content of pages whether there is the first object product for allowing the second target user to see with expectation, similar to content of pages
Highest is spent it may be considered that including the product information of the target product in content of pages, therefore is obtained similar in the first similarity
Highest second similarity is spent, corresponding second target product is chosen in multiple first object users according to the second similarity,
The second target user is obtained in the scoring of the first product to score to the second product of the second target product, has preset formula to calculate
It scores to effect of the data dispensing side to the second target user, the advertisement to target user of the data dispensing side is judged with this
Launch strategy, if be more in line with we itself expectation to target user.
Further, before step S3, comprising:
S31, judge whether content of pages is related to pre-set business;
If S32, unrelated with pre-set business, data dispensing side is zero to the effect scoring of the second target user;
If S33, entrance calculating content of pages and the product information of multiple first object products related to pre-set business
The step of first similarity.
In the present embodiment, it is contemplated that the content of pages that simulation second user specific behavior obtains be possible to we itself
Business is completely irrelevant, and the product information of content of pages and multiple first object products is directly directly calculated similarity, can be compared
Waste time, efficiency also can be slow, it is possible to first with pre-set business go comparison judge, determine content of pages whether contain and we
The relevant content of own service, whether if our business is loan transaction, can search in content of pages has the wordings such as loan,
If no, judge obtain content of pages be it is unrelated with preset loan transaction, then the data dispensing side obtained is current
There is no the information for the second target user not meet in content of pages that is, without the ad content for launching loan financial product
Our expectation, so effect of the data dispensing side to the second target user scores and remembers for the content of pages specifically obtained
It is zero.
Further, product information includes the product content of first object product, in step s3, comprising:
S34, word content is converted by content of pages;
S35, obtain product content from the product information of multiple first object products, calculate the product content with it is described
First similarity of word content;
S36, highest second similarity of similarity in the first similarity is obtained;
S37, it is chosen and highest second target product of the second similarity in multiple first object products;
S38, second product scoring of second target user to the second target product is obtained in first product scoring;
S39, the second similarity is scored to obtain data dispensing side multiplied by the second product, the effect of the second target user is commented
Point.
In the present embodiment, the content of pages generally obtained is by the collected page pictures of the means such as screenshotss, to page
Face picture can be directed to by carrying out optical character identification (Optical Character Recognition, abbreviation OCR)
Text conversion is become the image file of black and white lattice using optical mode, and will be schemed by identification software by printed character
Text conversion as in is at text formatting.Character recognition is carried out to page pictures using OCR, to identify the text in page pictures
Word content calculates the word content of identification and the product content successively obtained from the product information of multiple first object products
First similarity, the similarity of two sections of character strings can be compared or according to key by python, Java etc. by calculating similarity
Word goes to match, and the first similarity of the product content of word content and multiple first object products is obtained, according to the first similarity
Comparison, which available highest second similarity is, and then is chosen and the second similarity in multiple first object products
Corresponding second target product, with word content similarity highest it may be considered that there is the production of the second target in the content of pages obtained
The product content of product obtains the second target and uses in the first product scoring of second target user to multiple first object products
Second product of the second target product of family scores, and obtains data dispensing side in the product by the second similarity and the scoring of the second product
Effect scoring to the second target user, i.e., the data dispensing side meets the loan advertisement of the intelligent recommendation of the second target user
The expectation score value of our demands.
Further, product content is the first product keyword, in step s 35, comprising:
S351, the first product keyword is matched one by one with word content, obtains the second product identical with word content
Keyword quantity;
S352, the second product keyword quantity and the first product keyword ratio of number are calculated, obtains product content and text
The similarity of word content.
In the present embodiment, it is contemplated that the content of some product informations be it is secondary, we can produce first object
The product content of product is set as several first product keywords, if the type of first object product is that consumption is borrowed, cash is borrowed, credit
Borrow etc., other the first product keywords can also be the time limit, limit amount (highest), make loans time etc., and each first product closes
Key word goes to match with word content, can match, identical second product keyword number and the first total product keyword
First product keyword of the ratio of quantity, such as setting has 5, has 4 the second product keywords, then 4/5=0.8, i.e., similar
Degree is 80%.
Further, in step S351, comprising:
S3511, word content is subjected to word segmentation processing;
S3512, the first product keyword is matched with each participle, obtain and segments identical second product key
Word;
S3513, the second product keyword quantity of statistics.
In the present embodiment, in the present embodiment, N-gram statistics can be used when the word content of conversion being segmented
Model (N-gram Model), Hidden Markov Model (Hidden Markov Model, abbreviation HMM), maximum entropy model
(Maximum Entropy Model) is segmented, segmentation methods can include: Forward Maximum Method, reversed maximum matching are double
It is matched to maximum, shortest path first etc. is matched with the participle of word content with product keyword, due to participle and product
Keyword is all short phrase, multiple in the matched repeated matching that do not need in the process, is obtained with this and segments identical second production
Product keyword counts the quantity of the second product keyword.
In above-mentioned steps S4, the multiple and different of data dispensing side are obtained by simulating a large amount of first object user behaviors
Content of pages obtains data dispensing side and scores corresponding multiple effects of multiple first object users, multiple effects are scored
Cumulative to obtain the final effect score of data dispensing side, effect score is used to evaluate the dispensing effect of data dispensing side, according to number
The expectation of our own service demands whether is more in line with according to the effect score confirmation data dispensing side of dispensing side, i.e. data are launched
Whether the advertisement serving policy of side is more in line with we itself expectation to multiple first object users, and the score value of effect score is high
, the advertisement serving policy for representing data dispensing side is more in line with expectation, it is better to launch effect, then producing to multiple first objects
The advertisement delivery effect of product also can be better, meets own service demand and the good data dispensing side of effect carries out so as to select
The advertisement of multiple first object products is launched.
In above-mentioned steps S5, the advertisement serving policy that the effect score of data dispensing side represents data dispensing side is more accorded with
Conjunction demand, effect score is higher, is just more in line with own service demand and effect is good, so being obtained according to the effect of data dispensing side
Point, it can be scored with the effect of correlation data dispensing side, choose the data dispensing side of effect highest scoring, launched in the data of selection
The advertisement that multiple first object products are carried out in side is launched.
Referring to Fig. 2, the present invention also proposes a kind of device that data dispensing side is chosen based on big data, comprising:
Receiving module 1, for receiving the user information of multiple first object users and the product of multiple first object products
Information, user information are drawn a portrait including at least user;
First grading module 2, for choosing second target user from multiple first object users, according to the second mesh
The user information of user and the product information of multiple first object products are marked, obtains the second target user according to default code of points
The first product scoring to multiple first object products, and data are obtained in a manner of simulating second target user's specific behavior
The content of pages of dispensing side, specific behavior include at least the second target user and carry out behavior relevant to pre-set business;
Computing module 3, the first similarity of the product information for calculating content of pages and multiple first object products, is obtained
Highest second similarity of similarity in the first similarity is taken, is chosen in multiple first object products corresponding with the second similarity
The second target product, obtain the second target user in the scoring of the first product and score the second product of the second target product,
It is scored according to the second similarity and the second product, data dispensing side is obtained by preset formula, the effect of the second target user is commented
Point;
Accumulator module 4, for data dispensing side to add up to the effect scoring of each second target user, according to effect
The cumulative score value of fruit scoring obtains the effect score of data dispensing side, and effect score is used to evaluate the dispensing effect of data dispensing side
Fruit;
Module 5 is chosen, for the effect score of correlation data dispensing side, chooses the data dispensing side of effect highest scoring.
In above-mentioned receiving module 1, by taking loan transaction as an example, filtered out in the user for having applied for loan transaction first
Multiple first object users, such as representative, generality the first mesh of a batch is picked out from customer data base by expert
User is marked, multiple first object users can be the user of magnanimity, and user information, which may include that user draws a portrait, user is passing, to be gone through
History behavioural characteristic data etc., while multiple target products are chosen, target product can be loan financial product to be promoted, that is, need
The loan financial product for carrying out advertisement dispensing is also possible to multiple representative, generality loan gold that expert selects
Melt product, the product information of target product may include product content, product risks, product income, product sales volume, promote sequence
Deng.
In above-mentioned first grading module 2, second mesh is successively chosen from the multiple first object users received
User is marked, each second target user of selection is obtained according to user information and product information by default code of points
Second target user scores to the product of multiple first object products, that is, it is expected that the second target user sees that multiple first objects produce
The desired value of product such as carries out appropriate degree sequence to multiple first object products according to user information and product information, i.e. expectation the
The appropriate degree for the first object product that two target users see sorts, and the first object product of most suitable second target user represents
Most it is expected the first object product that the second target user sees, then the scoring highest of the first object product, obtains second with this
Target user scores to the first product of multiple first object products.It is obtained in a manner of simulating second target user's specific behavior
The content of pages of data dispensing side, wherein specific behavior includes at least the second target user and carries out row relevant to pre-set business
For for example, our preset business are loans, the second target user is easy application when last year double 11 purchase things
Loan, in the similar time, carries out then can simulate the user of qualification similar with the second target user in data dispensing side
Similar operation obtains the content of pages of data dispensing side, sees whether data dispensing side launches the advertisement of loan product, dispensing
Which product loan product is, i.e., the second target user of artificial intelligence is when data dispensing side carries out loan operation, data
Which kind of loan product what dispensing side was launched is, data dispensing side is the advertisement dispensing channel that product advertising is launched in selection.
Further, product information includes at least product risks, product income, product sales volume, promotion efficiency, the first scoring
Module 2 includes:
Submodule is divided, for according to user's portrait and product risks, multiple first object products to be divided into suitable production
Product and improper product;
Sorting sub-module is used for respectively by proper product and improper product according to product income descending sort, wherein if
Product income is identical then according to product sales volume descending sort, according to promotion efficiency descending sort if product sales volume is identical, and
The sequence of the improper product of product Income Maximum is lower than the smallest proper product of product income, obtains the conjunction of first object product
Appropriateness sequence;
Score submodule, for carrying out descending number to multiple first object products according to appropriate degree collating sequence, will produce
Product number is denoted as the product scoring of first object product, obtains the second target user to the first product of multiple first object products
Scoring.
In above-mentioned division submodule, user's portrait is believed according to user's social property, living habit and consumer behavior etc.
Cease the user model of a labeling taken out, the content of user's portrait may include gender, the age, occupation, repaying ability,
Values etc., user's portrait, which can be, calls directly well-established user's portrait relevant to target user, is drawn according to user
Picture and product risks are divided into proper product and improper product to target product, such as user's portrait of target user shows mesh
Mark user is that repaying ability is strong, then properly pushes the high product of risk, then divides the high first object product of product risks
For proper product, the opposite low product of risk that pushes is exactly that less properly, the low first object product of risk is divided into not
Proper product etc..
In above-mentioned sorting sub-module, the target product for being divided into proper product is individually ranked up, while will also be drawn
The target product for being divided into improper product is individually ranked up, and is such as set to high target product there are two product risks, then basis
The product income of target product sorts, income it is high come front, the identical then comparative product sales volume of income, product sales volume is high
Come front, product sales volume is identical, then sorted according to promotion efficiency, promotion efficiency it is big come front, promotion efficiency is preparatory
Setting, is stressed the popularization of multiple first object products, then arranges the first object model sequencing for being divided into proper product
Before the first object product for being divided into improper product, i.e., the sequence of the improper product of product Income Maximum is lower than production
The smallest proper product of product income obtains the second target user with this and sorts for the appropriate degree of multiple first object products,
Exactly it is expected that the second target user sees the sequence of first object product in the advertisement of dispensing.
In above-mentioned scoring submodule, for sorted multiple first object products, the first of foremost is come
Target product is that most suitable second target user sees that then the scoring of product is also highest, such as has three sections of target products,
The number of the first object product to rank the first is three, then the second target user scores for the product of the first object product
Namely three, the product number for coming last first object product is one, and the second target user is for the first object product
Product scoring namely one, the second target user is obtained with this and is scored the first product of multiple first object products.
Further, the first grading module 2 includes:
Acquisition submodule obtains the user behavior characteristics of the history of the second target user from system database;
Submodule is analyzed, user behavior characteristics is analyzed, obtains the specific behavior of the second target user;
Simulation submodule is obtained in a manner of simulating second target user's specific behavior according to the specific behavior of second user
The content of pages fetched according to dispensing side.
In the present embodiment, user behavior is primarily referred to as user's behavior online in internet, mobile interchange, such as browses
Webpage, uses application software, network social intercourse behavior etc. at search record.Have in system database for different network datas
The independent user behavior data for collecting each target user in source is to form historical user's behavioural characteristic of user, such as collects and use
Website that family is accessed, search content, access time, web site contents, application program of mobile phone service condition etc..In system data
Historical user's behavioural characteristic data of the second target user are obtained in library, and the user behavior characteristics of acquisition are analyzed, really
Operation behavior of fixed second target user when applying for loan product before, such as the second target user is in last year double 11 purchases
Loan is applied for when vehicle, i.e. when the second target user progress behavior relevant to pre-set business (loan) is double 11
Buy vehicle, it can the specific behavior for obtaining the second target user buys vehicle when being double 11, according to the second obtained mesh
The specific behavior for marking user, is interacted by crawler software with the interface API of data dispensing side, and virtual presses in data dispensing side
It is operated according to second target user's specific behavior, then obtains the content of pages of data dispensing side.
In above-mentioned computing module 3, the first phase of the content of pages of acquisition and the product information of multiple target products is calculated
Like degree, analyzing content of pages whether there is the first object product for allowing the second target user to see with expectation, with content of pages phase
Like degree highest it may be considered that including the product information of the target product in content of pages, therefore obtain phase in the first similarity
Like highest second similarity is spent, corresponding second target is chosen in multiple first object users according to the second similarity and is produced
Product obtain the second target user in the scoring of the first product and score the second product of the second target product, there is preset formula meter
Calculate and obtain the data side of dispensinging to the scoring of the effect of the second target user, with this come judge the data dispensing side to target user's
Advertisement serving policy, if be more in line with we itself expectation to target user.
Further, in some embodiments, further includes:
Judgment module judges whether content of pages is related to pre-set business;
Second grading module, if unrelated with pre-set business, data dispensing side is to the effect scoring of the second target user
Zero;
Into module, if related to pre-set business, enter the product for calculating content of pages and multiple first object products
The step of first similarity of information.
In the present embodiment, it is contemplated that the content of pages that simulation second user specific behavior obtains be possible to we itself
Business is completely irrelevant, and the product information of content of pages and multiple first object products is directly directly calculated similarity, can be compared
Waste time, efficiency also can be slow, it is possible to first with pre-set business go comparison judge, determine content of pages whether contain and we
The relevant content of own service, whether if our business is loan transaction, can search in content of pages has the wordings such as loan,
If no, judge obtain content of pages be it is unrelated with preset loan transaction, then the data dispensing side obtained is current
There is no the information for the second target user not meet in content of pages that is, without the ad content for launching loan financial product
Our expectation, so effect of the data dispensing side to the second target user scores and remembers for the content of pages specifically obtained
It is zero.
Further, product information includes the product content of first object product, and computing module 3 includes:
Submodule is converted, for converting word content for content of pages;
First computational submodule calculates institute for obtaining product content from the product information of multiple first object products
State the first similarity of product content Yu the word content;
First acquisition submodule, for obtaining highest second similarity of similarity in the first similarity;
Submodule is chosen, is chosen and highest second target product of the second similarity in multiple first object products;
Second acquisition submodule, for obtaining the second target user to the second target product in first product scoring
The second product scoring;
Second computational submodule, for scoring to obtain data dispensing side to the second mesh multiplied by the second product for the second similarity
Mark the effect scoring of user.
In the present embodiment, the content of pages generally obtained is by the collected page pictures of the means such as screenshotss, to page
Face picture can be directed to by carrying out optical character identification (Optical Character Recognition, abbreviation OCR)
Text conversion is become the image file of black and white lattice using optical mode, and will be schemed by identification software by printed character
Text conversion as in is at text formatting.Character recognition is carried out to page pictures using OCR, to identify the text in page pictures
Word content calculates the word content of identification and the product content successively obtained from the product information of multiple first object products
First similarity, the similarity of two sections of character strings can be compared or according to key by python, Java etc. by calculating similarity
Word goes to match, and multiple similarities of the product content of word content and multiple first object products is obtained, according to the first similarity
Comparison, which available highest second similarity is, and then is chosen and the second similarity in multiple first object products
Corresponding second target product, with word content similarity highest it may be considered that there is the production of the second target in the content of pages obtained
The product content of product obtains the second target and uses in the first product scoring of second target user to multiple first object products
Second product of the second target product of family scores, and obtains data dispensing side in the product by the second similarity and the scoring of the second product
Effect scoring to the second target user, i.e., the data dispensing side meets the loan advertisement of the intelligent recommendation of the second target user
The expectation score value of our demands.
Further, product content is the first product keyword, and the first computational submodule includes:
Matching unit obtains identical with word content for matching the first product keyword one by one with word content
Second product keyword quantity;
Computing unit obtains product for calculating the second product keyword quantity and the first product keyword ratio of number
The similarity of content and word content.
In the present embodiment, it is contemplated that the content of some product informations be it is secondary, we can produce first object
The product content of product is set as several first product keywords, if the type of first object product is that consumption is borrowed, cash is borrowed, credit
Borrow etc., other the first product keywords can also be the time limit, limit amount (highest), make loans time etc., and each first product closes
Key word goes to match with content of pages, can match, identical second product keyword number and the first total product keyword
First product keyword of the ratio of quantity, such as setting has 5, has 4 the second product keywords, then 4/5=0.8, i.e., similar
Degree is 80%.
Further, matching unit includes:
Subelement is handled, for word content to be carried out word segmentation processing;
Coupling subelement is obtained and segments identical for matching the first product keyword with each participle
Two product keywords;
Count subelement, the second product keyword quantity for statistical match.
In the present embodiment, N-gram statistical model (N-gram can be used when the word content of conversion being segmented
Model), Hidden Markov Model (Hidden Markov Model, abbreviation HMM), maximum entropy model (Maximum Entropy
Model it) is segmented, segmentation methods can include: Forward Maximum Method, reversed maximum matching, two-way maximum matching, shortest path
Diameter algorithm etc. is matched with the participle of word content with product keyword, since participle and product keyword are all short phrases,
It is multiple that repeated matching is not needed during matched, is obtained with this and segments identical second product keyword, statistics second
The quantity of product keyword.
In above-mentioned accumulator module 4, by simulate a large amount of first object user behaviors obtain data dispensing side it is multiple not
Same content of pages obtains data dispensing side and scores corresponding multiple effects of multiple first object users, by multiple effects
Scoring is cumulative to obtain the final effect score of data dispensing side, and effect score is used to evaluate the dispensing effect of data dispensing side, root
The expectation of our own service demands, i.e. data whether are more in line with according to the effect score confirmation data dispensing side of data dispensing side
Whether the advertisement serving policy of dispensing side is more in line with we itself expectation to multiple first object users, point of effect score
It is worth high, the advertisement serving policy for representing data dispensing side is more in line with expectation, and it is better to launch effect, then to multiple first mesh
The advertisement delivery effect for marking product also can be better, meets own service demand and the good data dispensing side of effect so as to select
The advertisement for carrying out multiple first object products is launched.
In above-mentioned selection module 5, the effect score of data dispensing side represents the advertisement serving policy of data dispensing side more
Add and meet demand, effect score is higher, is just more in line with own service demand and effect is good, so according to the effect of data dispensing side
Fruit score can be scored with the effect of correlation data dispensing side, the data dispensing side of effect highest scoring be chosen, in the data of selection
The advertisement that multiple first object products are carried out in dispensing side is launched.
Referring to Fig. 3, a kind of equipment is also provided in the embodiment of the present application, which can be computer or server, in
Portion's structure can be as shown in Figure 3.The equipment includes processor, memory, network interface and the data connected by system bus
Library.Wherein, the processor of the Computer Design is for providing calculating and control ability.The memory of the computer equipment includes non-
Volatile storage medium, built-in storage.The non-volatile memory medium is stored with operating system, computer-readable instruction and data
Library.The internal memory provides environment for the operation of operating system and computer-readable instruction in non-volatile memory medium.This sets
Standby database is for data such as storage configuration item information.The network interface of the computer equipment is used to pass through with external terminal
Network connection communication.The computer-readable instruction when being executed, executes the process of the embodiment such as above-mentioned each method.This field skill
Art personnel are appreciated that structure shown in Fig. 3, only the block diagram of part-structure relevant to application scheme, not structure
The restriction for the computer equipment that pairs of application scheme is applied thereon.
One embodiment of the application also provides a kind of computer non-volatile readable storage medium, and being stored thereon with computer can
Reading instruction, the computer-readable instruction when being executed, execute the process of the embodiment such as above-mentioned each method, comprising: receive multiple
The product information of the user information of first object user and multiple first object products, user information are drawn a portrait including at least user;
Second target user is chosen from multiple first object users, according to the user information of the second target user and multiple first
The product information of target product obtains the second target user according to default code of points and produces to the first of multiple first object products
It judges point, and obtains the content of pages of data dispensing side, specific behavior in a manner of simulating second target user's specific behavior
Behavior relevant to pre-set business is carried out including at least the second target user;Calculate content of pages and multiple first object products
First similarity of product information obtains highest second similarity of similarity in the first similarity, produces in multiple first objects
The second target product corresponding with the second similarity is chosen in product, obtains the second target user to second in the scoring of the first product
Second product of target product scores, and is scored according to the second similarity and the second product, obtains data dispensing side by preset formula
Effect scoring to the second target user;Data dispensing side adds up to the effect scoring of each second target user, root
The effect score of data dispensing side is obtained according to the cumulative score value that effect scores, effect score is used to evaluate the dispensing of data dispensing side
Effect;The effect score of correlation data dispensing side chooses the data dispensing side of effect highest scoring.
Product scoring is carried out to multiple target products respectively by each target user, and in a manner of modelling customer behavior
Multiple content of pages are obtained in data dispensing side, multiple content of pages are subjected to similarity calculation with multiple target products respectively,
It obtains data dispensing side to score to the effect of each target user, the cumulative effect for obtaining data dispensing side of effect scoring is obtained
Point, according to the effect score of data dispensing side, it can be determined that whether the advertisement serving policy of data dispensing side is more in line with us
Expection to target user, so that it is more preferable more to meet own service demand, effect according to the selection of the effect score of data dispensing side
Data dispensing side.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
Any reference used in provided herein and embodiment to memory, storage, database or other media,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double speed are according to rate SDRAM (SSRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, device, article or the method that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, device, article or method institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, device of element, article or method.
The foregoing is merely preferred embodiment of the present application, are not intended to limit the scope of the patents of the application, all utilizations
Equivalent structure or equivalent flow shift made by present specification and accompanying drawing content is applied directly or indirectly in other correlations
Technical field, similarly include in the scope of patent protection of the application.
Claims (10)
1. a kind of method for choosing data dispensing side based on big data characterized by comprising
The user information of multiple first object users and the product information of multiple first object products are received, the user information is extremely
It less include that user draws a portrait;
From multiple first object users choose second target user, according to the user information of the second target user with it is multiple
The product information of first object product obtains the second target user to multiple first object products according to default code of points
The scoring of the first product, and obtain in a manner of simulating second target user's specific behavior the content of pages of data dispensing side,
The specific behavior includes at least the second target user and carries out behavior relevant to pre-set business;
The first similarity of the product information of the content of pages and multiple first object products is calculated, obtains described first
Highest second similarity of similarity in similarity is chosen corresponding with second similarity in multiple first object products
Second target product obtains the second target user in first product scoring and comments the second product of the second target product
Point, it is scored according to second similarity and second product, data dispensing side is obtained by preset formula, the second target is used
The effect at family scores;
Data dispensing side adds up to the effect scoring of each second target user, is obtained according to the cumulative score value that effect scores
To the effect score of data dispensing side, the effect score is used to evaluate the dispensing effect of data dispensing side;
The effect score of correlation data dispensing side, chooses the data dispensing side of the effect highest scoring.
2. the method according to claim 1 for choosing data dispensing side based on big data, which is characterized in that the product letter
Breath includes at least product risks, product income, product sales volume, promotion efficiency, the user information according to the second target user
With the product information of multiple first object products, the second target user is obtained to multiple first mesh according to default code of points
In the step of marking the first product scoring of product, comprising:
According to user portrait and the product risks, multiple first object products are divided into proper product and improper production
Product;
Respectively by proper product and improper product according to the product income descending sort, wherein if the product income is identical
Then according to the product sales volume descending sort, according to the promotion efficiency descending sort if the product sales volume is identical, and
The sequence of the improper product of product Income Maximum is lower than the smallest proper product of product income, obtains the conjunction of first object product
Appropriateness sequence;
Descending number is carried out to multiple first object products according to appropriate degree collating sequence, number is denoted as first object product
Product scoring obtains the second target user and scores first product of multiple first object products.
3. it is according to claim 1 based on big data choose data dispensing side method, which is characterized in that it is described and with
It simulates in the step of the mode of second target user's specific behavior obtains the content of pages of data dispensing side, comprising:
The user behavior characteristics of the history of the second target user are obtained from system database;
The user behavior characteristics are analyzed, the specific behavior of the second target user is obtained;
According to the specific behavior of the second target user, number is obtained in a manner of simulating specific behavior described in the second target user
According to the content of pages of dispensing side.
4. the method according to claim 1 for choosing data dispensing side based on big data, which is characterized in that the calculating institute
Before the step of stating the first similarity of the product information of content of pages and multiple first object products, further includes:
Judge whether the content of pages is related to the pre-set business;
If unrelated with the pre-set business, data dispensing side is denoted as zero to the effect scoring of the second target user;
If related to the pre-set business, enter the product letter for calculating the content of pages and multiple first object products
The step of first similarity of breath.
5. the method according to claim 1 for choosing data dispensing side based on big data, which is characterized in that the product letter
Breath includes the product content of first object product, the product for calculating the content of pages and multiple first object products
First similarity of information obtains highest second similarity of similarity in first similarity, produces in multiple first objects
The second target product corresponding with second similarity is chosen in product, is obtained the second target in first product scoring and is used
Family scores to the second product of the second target product, is scored according to second similarity and second product, by default public affairs
Formula obtained in the step of data dispensing side scores to the effect of the second target user, comprising:
Word content is converted by the content of pages;
Obtain the product content from the product information of multiple first object products, calculate the product content with it is described
First similarity of word content;
Obtain highest second similarity of similarity in first similarity;
The second target product corresponding with second similarity is chosen in multiple first object products;
The second target user is obtained in first product scoring to score to second product of the second target product;
Second similarity is scored to obtain data dispensing side multiplied by second product, the effect of the second target user is commented
Point.
6. the method according to claim 5 for choosing data dispensing side based on big data, which is characterized in that in the product
In the step of holding is the first product keyword, the similarity for calculating the product content and the word content, comprising:
The first product keyword is matched one by one with the word content, identical with the word content second is obtained and produces
Product keyword quantity;
Calculate the second product keyword quantity and the first product keyword ratio of number, obtain the product content with
The similarity of the word content.
7. it is according to claim 6 based on big data choose the data side of dispensinging method, which is characterized in that it is described will described in
First product keyword matches one by one with the word content, obtains the second product number of keyword identical with the word content
In the step of amount, comprising:
The word content is subjected to word segmentation processing;
The first product keyword is matched with each participle, obtain and segments the identical second product key
Word;
Count the second product keyword quantity.
8. a kind of device for choosing data dispensing side based on big data characterized by comprising
Receiving module, for receiving the user information of multiple first object users and the product information of multiple first object products,
The user information is drawn a portrait including at least user;
First grading module is used for choosing second target user from multiple first object users according to the second target
The product information of the user information at family and multiple first object products obtains the second target user according to default code of points
The first product scoring to multiple first object products, and data are obtained in a manner of simulating second target user's specific behavior
The content of pages of dispensing side, the specific behavior include at least the second target user and carry out behavior relevant to pre-set business;
Computing module, it is similar to the first of the product information of multiple first object products for calculating the content of pages
Degree obtains highest second similarity of similarity in first similarity, chosen in multiple first object products with it is described
Corresponding second target product of second similarity obtains the second target user in first product scoring and produces to the second target
Second product of product scores, and is scored according to second similarity and second product, obtains data by preset formula and launches
It scores the effect of the second target user side;
Accumulator module is commented for adding up data dispensing side to the effect scoring of each second target user according to effect
The cumulative score value divided obtains the effect score of data dispensing side, and the effect score is used to evaluate the dispensing effect of data dispensing side
Fruit;
Module is chosen, the effect for correlation data dispensing side scores, and the data for choosing the effect highest scoring are launched
Side.
9. a kind of equipment, including memory and processor, the memory are stored with computer-readable instruction, which is characterized in that
The processor realizes the step of any one of claim 1 to 7 the method when executing the computer-readable instruction.
10. a kind of computer non-volatile readable storage medium, is stored thereon with computer-readable instruction, which is characterized in that institute
State the step of any one of claims 1 to 7 the method is realized when computer-readable instruction is executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910527849.7A CN110390098B (en) | 2019-06-18 | 2019-06-18 | Method, device, equipment and storage medium for selecting data delivery party based on big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910527849.7A CN110390098B (en) | 2019-06-18 | 2019-06-18 | Method, device, equipment and storage medium for selecting data delivery party based on big data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110390098A true CN110390098A (en) | 2019-10-29 |
CN110390098B CN110390098B (en) | 2022-08-19 |
Family
ID=68285730
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910527849.7A Active CN110390098B (en) | 2019-06-18 | 2019-06-18 | Method, device, equipment and storage medium for selecting data delivery party based on big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110390098B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102930452A (en) * | 2012-09-24 | 2013-02-13 | 曹欢欢 | Method and device for releasing home decoration advertisements accurately |
CN104794631A (en) * | 2015-03-31 | 2015-07-22 | 北京奇艺世纪科技有限公司 | Verification method and device for advertisement putting effect |
CN107563808A (en) * | 2017-08-29 | 2018-01-09 | 天脉聚源(北京)科技有限公司 | A kind of advertisement putting data statistical approach and device |
CN108960899A (en) * | 2018-06-11 | 2018-12-07 | 广东因特利信息科技股份有限公司 | The user information exchange method and system launched for advertisement |
CN109583970A (en) * | 2018-12-14 | 2019-04-05 | 深圳壹账通智能科技有限公司 | Advertisement placement method, device, computer equipment and storage medium |
-
2019
- 2019-06-18 CN CN201910527849.7A patent/CN110390098B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102930452A (en) * | 2012-09-24 | 2013-02-13 | 曹欢欢 | Method and device for releasing home decoration advertisements accurately |
CN104794631A (en) * | 2015-03-31 | 2015-07-22 | 北京奇艺世纪科技有限公司 | Verification method and device for advertisement putting effect |
CN107563808A (en) * | 2017-08-29 | 2018-01-09 | 天脉聚源(北京)科技有限公司 | A kind of advertisement putting data statistical approach and device |
CN108960899A (en) * | 2018-06-11 | 2018-12-07 | 广东因特利信息科技股份有限公司 | The user information exchange method and system launched for advertisement |
CN109583970A (en) * | 2018-12-14 | 2019-04-05 | 深圳壹账通智能科技有限公司 | Advertisement placement method, device, computer equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110390098B (en) | 2022-08-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111177569B (en) | Recommendation processing method, device and equipment based on artificial intelligence | |
CN110532451A (en) | Search method and device for policy text, storage medium, electronic device | |
CN110209805B (en) | Text classification method, apparatus, storage medium and computer device | |
CN109509033B (en) | Big data prediction method for user purchasing behavior in consumption financial scene | |
CN109766438A (en) | Biographic information extracting method, device, computer equipment and storage medium | |
CN109767318A (en) | Loan product recommended method, device, equipment and storage medium | |
CN112508609B (en) | Crowd expansion prediction method, device, equipment and storage medium | |
CN112631997B (en) | Data processing method, device, terminal and storage medium | |
CN108363821A (en) | A kind of information-pushing method, device, terminal device and storage medium | |
WO2018040069A1 (en) | Information recommendation system and method | |
CN106919575B (en) | Application program searching method and device | |
CN103870523A (en) | Analyzing content to determine context and serving relevant content based on the context | |
CN106445963B (en) | Advertisement index keyword automatic generation method and device of APP platform | |
CN110325986A (en) | Article processing method, device, server and storage medium | |
CN110222260A (en) | A kind of searching method, device and storage medium | |
CN109801101A (en) | Label determines method, apparatus, computer equipment and storage medium | |
CN106919588A (en) | A kind of application program search system and method | |
CN111582932A (en) | Inter-scene information pushing method and device, computer equipment and storage medium | |
CN110389963A (en) | The recognition methods of channel effect, device, equipment and storage medium based on big data | |
CN112685639A (en) | Activity recommendation method and device, computer equipment and storage medium | |
Diniz et al. | Github label embeddings | |
CN110766465A (en) | Financial product evaluation method and verification method and device thereof | |
CN110390098A (en) | Method, apparatus, equipment and the storage medium of data dispensing side are chosen based on big data | |
CN109684467A (en) | A kind of classification method and device of text | |
Pîrvu et al. | Predicting user intent from search queries using both cnns and rnns |
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 |