CN110472995A - To shop prediction technique, device, readable storage medium storing program for executing and electronic equipment - Google Patents
To shop prediction technique, device, readable storage medium storing program for executing and electronic equipment Download PDFInfo
- Publication number
- CN110472995A CN110472995A CN201910610494.8A CN201910610494A CN110472995A CN 110472995 A CN110472995 A CN 110472995A CN 201910610494 A CN201910610494 A CN 201910610494A CN 110472995 A CN110472995 A CN 110472995A
- Authority
- CN
- China
- Prior art keywords
- shop
- user
- feature
- search
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 71
- 238000003860 storage Methods 0.000 title claims abstract description 11
- 238000000605 extraction Methods 0.000 claims abstract description 47
- 238000012549 training Methods 0.000 claims abstract description 24
- 230000006399 behavior Effects 0.000 claims description 70
- 238000004590 computer program Methods 0.000 claims description 11
- 230000000977 initiatory effect Effects 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 3
- 241000208340 Araliaceae Species 0.000 claims description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 2
- 235000008434 ginseng Nutrition 0.000 claims description 2
- 238000012360 testing method Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 10
- 230000008569 process Effects 0.000 description 10
- 238000012545 processing Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 235000013305 food Nutrition 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 235000013361 beverage Nutrition 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 235000012055 fruits and vegetables Nutrition 0.000 description 1
- 239000009153 huxin Substances 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 235000012054 meals Nutrition 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 235000011888 snacks Nutrition 0.000 description 1
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
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/52—Network services specially adapted for the location of the user terminal
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/029—Location-based management or tracking services
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Accounting & Taxation (AREA)
- Databases & Information Systems (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Signal Processing (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the present application provides one kind to shop prediction technique, device, readable storage medium storing program for executing and electronic equipment, this method comprises: obtaining the searching request that user terminal is initiated；Described search is requested to carry out feature extraction, determines that the feature of the user of the user terminal, described search request the feature in each shop and user-shop cross feature in corresponding shop list；Shop Probabilistic Prediction Model is arrived into the feature of the user, the feature in each shop and the user-shop cross feature input training in advance, determine that the user reaches the probability in each shop；The probability that each shop is reached according to the user, determines whether the user reaches target shop.The application goes out the target shop of user's arrival by the determine the probability in each shop exported to shop Probabilistic Prediction Model, realizes the accurate positionin to the current location of user, improves the reliability in the shop near user terminal current location recommended to the user.
Description
Technical field
This application involves technical field of information processing, more particularly to one kind to arrive shop prediction technique, device, readable storage medium storing program for executing
And electronic equipment.
Background technique
With the development of mobile internet, people easily can access network by mobile device and be serviced with obtaining,
Thus the local life-stylize service of a collection of O2O (Online-to-Offline) (such as: O2O is nearby searched for) has been risen, greatly side
People's lives.By taking O2O is nearby searched for as an example, user by the function of search can check cuisines near front position,
The living informations such as amusement, the specific implementation process of the function of search are as follows: the longitude and latitude mark for collecting each trade company in advance obtains in real time
The positioning for taking family calculates the positioning of user at a distance from each trade company for having marked longitude and latitude, distance is met to the sieve of user
The selection result is obtained after selecting the trade company of range to be ranked up according to certain algorithm, and the selection result is used back to user
Mobile device is shown.
However, the specific implementation process of above-mentioned function of search has a problem that: the selection result is overly dependent upon user's
Positioning, when the not up to certain accuracy of the positioning of user, the position for the user being calculated is deposited at a distance from each trade company
In deviation, a part of trade company is caused as the selection result and user can not to be returned to, has seriously affected the decision of user and made
With experience.
Thus, it is certain to meet the positioning for the user being calculated at a distance from each trade company for having marked longitude and latitude
Accuracy, need to guarantee that the positioning of user meets higher required precision.In a practical situation, the real-time positioning of user is by network
Situation is affected, when (such as: in market) initiates to search near O2O user in some buildings, due to general
Logical location technology can not position the floor information of building, and mobile device can not receive GPS signal indoors,
The real-time positioning of user is realized in the base station that network operator can only be relied on, and the user that the base station for relying on network operator obtains
There are drift phenomenons for positioning in real time, not can guarantee the real-time positioning of user indoors and meet higher required precision.
Therefore, how more accurately to carry out positioning to the current location of user is this field urgent problem.
Summary of the invention
The embodiment of the present application provide one kind to shop prediction technique, device, readable storage medium storing program for executing and electronic equipment, can be in real time
Whether prediction user reaches shop, realizes the precise positioning to the current location of user.
The embodiment of the present application first aspect provides a kind of to shop prediction technique, which comprises
Obtain the searching request that user terminal is initiated；
Described search is requested to carry out feature extraction, determines feature, the described search request of the user of the user terminal
The feature in each shop and user-shop cross feature in corresponding shop list, the user-shop cross feature is pair
The feature of the user and the feature in each shop carry out what characteristic crossover obtained；
The feature of the user, the feature in each shop and the user-shop cross feature input is preparatory
Trained arrives shop Probabilistic Prediction Model, determines that the user reaches the probability in each shop；
The probability that each shop is reached according to the user, determines whether the user reaches target shop, described
Target shop is one in each shop.
Optionally, the probability that each shop is reached according to the user, determines whether the user reaches mesh
The step of marking shop, comprising:
In the case where the probability that the user reaches the target shop is greater than preset probability threshold value, the use is determined
Family reaches the target shop；Or
It is greater than preset probability threshold value in the probability that the user reaches the target shop, and described search request corresponds to
Parameter value in the case where preset come into force in range of parameter values, determine that the user reaches the target shop.
Optionally, the method also includes:
It is determining with user to the associated user behavior type in shop；
Extraction first kind search record and the search of the second class record from the search log of the user terminal, and described first
Class search is recorded as meeting the search record of the user behavior type, and the second class search is recorded as the corresponding search moment
Search of the time difference at the search moment recorded with first kind search in preset duration records；
The search of the user behavior type will be met in first kind search record and second class search record
Recording mark is positive sample, and, the search that the user behavior type is not met in second class search record is recorded
Labeled as negative sample；
According to the positive sample and the negative sample, preset model is trained, is obtained described to shop probabilistic forecasting mould
Type.
Optionally, described that preset model is trained according to the positive sample and the negative sample, it obtains described to shop
The step of Probabilistic Prediction Model, comprising:
Feature extraction is carried out to the positive sample and the negative sample respectively, determines that the positive sample and the negative sample are each
Each sample shop in the feature of self-corresponding sample of users, the positive sample and the corresponding shop list of the negative sample
Feature and sample of users-sample shop cross feature, the sample of users-sample shop cross feature is to the sample
The feature of user and the feature in each sample shop carry out what characteristic crossover obtained；
With the feature of the sample of users, the feature in each sample shop and the sample of users-sample shop
Cross feature is training sample, is trained to the preset model, is obtained described to shop Probabilistic Prediction Model.
Optionally, after the step of determination user reaches the target shop, the method also includes:
Extracting the corresponding search moment from the search log of the user terminal is determining that the user reaches the mesh
Mark the search record after shop；
In the case where the search record of extraction is the search record for the target shop, by the search of the extraction
Recording mark is positive sample, and increases the weight of the positive sample；
If the search record of the extraction is not the searching request for the target shop, the search of the extraction is remembered
Record mark is negative sample, and reduces the weight of the negative sample；And
According to the negative sample after the positive sample and reduction weight after increase weight, carried out to described to shop Probabilistic Prediction Model
It updates.
Optionally, the feature of the user include it is following at least one: the user terminal scanning or the WIFI that is connected to
Title and corresponding signal strength, the device type of the user terminal, the longitude and latitude of the user terminal, the user terminal
IP address, the user user portrait and the user consumption preferences.
Optionally, the feature in each shop include it is following at least one: it is the mark in each shop, described each
The WIFI title in shop, each shop WIFI averagely connect or scan intensity, each shop longitude and latitude, described
The price range for the commodity that classification belonging to each shop, each shop are sold, the clicking rate in each shop and
The visit purchase rate in each shop.
Optionally, the user-shop cross feature is obtained by following at least one mode:
According to the longitude and latitude of the longitude and latitude of the user terminal and each shop, determine the user terminal with it is described
The linear distance in each shop；
The signal strength for the WIFI in shop that the user terminal is scanned or is connected to, with user terminal scanning or
The WIFI in the shop being connected to averagely connects or scans intensity and carries out characteristic crossover；
The signal strength for the WIFI in shop that the user terminal is scanned or is connected to, with the user terminal with it is described
The linear distance in the shop that user terminal is scanned or is connected to carries out characteristic crossover；And/or
Feature friendship is carried out to user's click of the user terminal or consumption price and the price per capita in each shop
Fork.
The embodiment of the present application second aspect provides one kind to shop prediction meanss, and described device includes:
Module is obtained, for obtaining the searching request of user terminal initiation；
Characteristic extracting module carries out feature extraction for requesting described search, determines the user's of the user terminal
Feature, described search request the feature in each shop and user-shop cross feature, the use in corresponding shop list
Family-shop cross feature is to carry out characteristic crossover to the feature of the user and the feature in each shop to obtain；
Probabilistic forecasting module, for by the feature of the feature of the user, each shop and the user-shop
Shop Probabilistic Prediction Model is arrived in cross feature input training in advance, determines that the user reaches the probability in each shop；With
And
Determining module determines whether the user reaches for reaching the probability in each shop according to the user
Target shop, the target shop are one in each shop.
Optionally, the determining module includes:
First determining module, the probability for reaching the target shop in the user are greater than preset probability threshold value
In the case of, determine that the user reaches the target shop；Or
Second determining module, the probability for reaching the target shop in the user are greater than preset probability threshold value,
And described search requests corresponding parameter value in the case where preset come into force in range of parameter values, determines that the user reaches institute
State target shop.
Optionally, described device further include:
Third determining module, for it is determining with user to the associated user behavior type in shop；
First extraction module, for extracting first kind search record and the second class from the search log of the user terminal
Search record, the first kind search are recorded as meeting the search record of the user behavior type, the second class search note
Record is that search of the time difference at the search moment of corresponding search moment and first kind search record in preset duration is remembered
Record；
Mark module, for user's row will to be met in first kind search record and second class search record
It is positive sample for the search recording mark of type, and, the user behavior class will not be met in second class search record
The search recording mark of type is negative sample；
Training module obtains described arrive for being trained to preset model according to the positive sample and the negative sample
Shop Probabilistic Prediction Model.
Optionally, the training module includes:
Feature extraction submodule, for carrying out feature extraction respectively to the positive sample and the negative sample, determine described in
The feature of positive sample and the corresponding sample of users of the negative sample, the positive sample and the corresponding shop of the negative sample
Spread the feature and sample of users-sample shop cross feature in each sample shop in list, the sample of users-sample shop
Cross feature is to carry out characteristic crossover to the feature of the sample of users and the feature in each sample shop to obtain；
Training submodule, for the feature of the sample of users, the feature in each sample shop and the sample
This user-sample shop cross feature is training sample, is trained to the preset model, is obtained described to shop probabilistic forecasting
Model.
Optionally, described device further include:
Second extraction module, for extracting the corresponding search moment from the search log of the user terminal determining
It states user and reaches search record after the target shop；
First weight adjusts module, is the feelings for the search record in the target shop for the search record in extraction
It is positive sample by the search recording mark of the extraction, and increase the weight of the positive sample under condition；
Second weight adjusts module, is not the search note for the target shop for the search record in the extraction
It is negative sample by the search recording mark of the extraction, and reduce the weight of the negative sample in the case where record；And
Update module, for according to increase weight after positive sample and reduce weight after negative sample, to it is described to shop it is general
Rate prediction model is updated.
Optionally, the feature of the user include it is following at least one: the user terminal scanning or the WIFI that is connected to
Title and corresponding signal strength, the device type of the user terminal, the longitude and latitude of the user terminal, the user terminal
IP address, the user user portrait and the user consumption preferences.
Optionally, the feature in each shop include it is following at least one: it is the mark in each shop, described each
The WIFI title in shop, each shop WIFI averagely connect or scan intensity, each shop longitude and latitude, described
The price range for the commodity that classification belonging to each shop, each shop are sold, the clicking rate in each shop and
The visit purchase rate in each shop.
Optionally, the user-shop cross feature is obtained by following at least one mode:
According to the longitude and latitude of the longitude and latitude of the user terminal and each shop, determine the user terminal with it is described
The linear distance in each shop；
The signal strength for the WIFI in shop that the user terminal is scanned or is connected to, with user terminal scanning or
The WIFI in the shop being connected to averagely connects or scans intensity and carries out characteristic crossover；
The signal strength for the WIFI in shop that the user terminal is scanned or is connected to, with the user terminal with it is described
The linear distance in the shop that user terminal is scanned or is connected to carries out characteristic crossover；And/or
Feature friendship is carried out to user's click of the user terminal or consumption price and the price per capita in each shop
Fork.
The embodiment of the present application third aspect provides a kind of computer readable storage medium, is stored thereon with computer program,
The step in the method as described in the application first aspect is realized when the program is executed by processor.
The embodiment of the present application fourth aspect provides a kind of electronic equipment, including memory, processor and is stored in memory
Computer program that is upper and can running on a processor, the processor realize method described in the application first aspect when executing
The step of.
Shop prediction technique is arrived using one kind provided by the embodiments of the present application, first acquisition Client-initiated searching request, so
Feature extraction is carried out to searching request afterwards, and (includes: feature, the searching request of the user of user terminal by the feature extracted
The feature in each shop and user-shop cross feature in corresponding shop list) shop Probabilistic Prediction Model is input to obtain
The probability in each shop in the list of shop is reached to user, finally obtains the target shop of user's arrival further according to these probabilistic forecastings
Paving.The search that the application initiates user terminal from user's dimension, shop dimension and user-dimension of shop cross-dimension three
Request carries out feature extraction, improves the accuracy of the result exported to shop Probabilistic Prediction Model and finally predicts obtained mesh
The accuracy in shop is marked, in addition, going out user's arrival by the determine the probability in each shop exported to shop Probabilistic Prediction Model
Target shop realizes the accurate positionin to the current location of user, improves user terminal current location recommended to the user
The reliability in neighbouring shop enhances the usage experience of user.
Detailed description of the invention
Technical solution in ord to more clearly illustrate embodiments of the present application, below by institute in the description to the embodiment of the present application
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the application
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is the schematic diagram that the user terminal that each embodiment of the application provides and background server interact；
Fig. 2 is that the application one implements a kind of flow chart to shop prediction technique exemplified；
Fig. 3 is that the application one implements the acquisition user-shop cross feature flow chart exemplified；
Fig. 4 is the flow chart that the application one implements a kind of training method to shop Probabilistic Prediction Model exemplified；
Fig. 5 be the application one implement the another kind that exemplifies to shop Probabilistic Prediction Model training method flow chart；
Fig. 6 be the application one implement a kind of update for exemplifying to shop Probabilistic Prediction Model method flow chart；
Fig. 7 is that the application one implements a kind of schematic diagram to shop prediction meanss exemplified.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiment is some embodiments of the present application, instead of all the embodiments.Based on this Shen
Please in embodiment, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall in the protection scope of this application.
Before being illustrated to each embodiment of the application, the relevant technologies are illustrated first.To obtain user's
More accurately position, in the related art, in such a way that the positioning to the user of acquisition is modified, such as therein one
Kind correcting mode are as follows: reference source is found in a certain range of the current location of user, such as: if the network state that user is current
It is to have connected WIFI, then can be using the coordinate of the WIFI as a reference source, then the positioning of user is modified；Another example is:
When user is in inside building (such as: inside megastore), since GPS signal is blocked, then can according to
The Geomagnetism Information on the periphery of family history positioning is modified the positioning of user.
However, the mode that the positioning of the above two user to acquisition is modified needs to obtain corresponding data information,
It is of a high price, such as: when the coordinate using WIFI is as reference source, need additionally to obtain each in region locating for user
The coordinate information of WIFI need to additionally obtain region locating for user when being modified using Geomagnetism Information to the positioning of user
In Geomagnetism Information.
In the related art, the mode of another positioning for obtaining user is: the WIFI in the shop that collection user is connected to,
User formulates corresponding rule then in conjunction with business in the purchase information in the shop residence time and user in the shop
Then, judge user whether in the shop further according to rule.But which only considers the WIFI in the shop of user's connection, uses
Whether residence time and user of the family in the shop only judge user from user side in the purchase information in the shop
In the shop, in addition, whether the program is overly dependent upon shop whether WIFI, user have buying behavior, do not have yet compared with
Strong universality, therefore the accuracy that this positioning method positions obtained positioning result to user is not still high.
In order to improve user positioning accuracy, the embodiment of the present application is local using O2O (Online-to-Offline)
The characteristic that many business can only be completed in businessman shop in life-stylize service obtains the searching request of user in real time and marks use
Family is made that typical behaviour (such as: purchasing by group in participation shop tests certificate activity, favour is dodged in purchase, registers, uploads UGC etc.) for shop
Consumption data, then carry out features from the multiple dimensions such as the intersection in user, shop, user and shop for these consumption datas and mention
It takes, it is then that the feature extracted input is trained to shop Probabilistic Prediction Model in advance, it obtains user and reaches each shop
Probability finally obtains the target shop that user is actually reached further according to these probability.
One kind provided by the embodiments of the present application will be described in detail to shop prediction technique below.
Fig. 1 is the schematic diagram that the user terminal that each embodiment of the application provides and background server interact.Reference
Fig. 1, background server by network and one or more user terminals (such as: user terminal 1 in Fig. 1 to user terminal n)
It is communicatively coupled, to realize communication interaction.Background server can be network server, database server etc..User is whole
End can be PC (personal computer, PC), tablet computer, smart phone etc..
Fig. 2 is that the application one implements a kind of flow chart to shop prediction technique exemplified, and this method is applied in Fig. 1
Background server.Referring to Fig. 2, one embodiment of the application provide to shop prediction technique the following steps are included:
Step S11: the searching request that user terminal is initiated is obtained.
In the present embodiment, user can input searching request in user terminal, and background server is for receiving user terminal
The searching request of transmission.Wherein, user terminal is equipped with the terminal application software that can support neighbouring function of search, such as: search
Class application software, shopping class application software can provide the application software of other services for user.By this kind of application software,
User may search for any target in the peripheral extent of current location.User terminal receives the searching request of user's input
Afterwards, searching request is sent to background server, then receives the search knot that background server is returned for the secondary searching request
Search result is simultaneously showed user in the form of the list of shop on the page by fruit.
After user terminal receives the searching request of user, it will be actively sent to for the user data of this searching request
Background server, alternatively, being asked in background server next time to user terminal after user terminal receives the searching request of user
When seeking user data, then it will be sent to background server for the user data of this searching request, it is of course also possible to use its
Its mode will be sent to background server for the user data of searching request, and each embodiment of the application is not made this specifically
Limitation.
Wherein, user data includes the device status information of user behavior data, user information and user terminal.User
Behavioral data refers to: the data of dynamic generation, user behavior when user makes user behavior to any shop in the list of shop
It can be the behaviors such as general user's behavior, such as click, collection, sharing；Be also possible to to the relevant behavior in shop (i.e.: only
Reach the behavior that could be completed behind shop), such as: it dodges favour and checks, purchases by group and test certificate, Self-help ordering, the number of taking queuing, upload UGC, use
Family is registered, it is of course also possible to be other types of user behavior, the application includes but is not limited to the above-mentioned a variety of use enumerated
Family behavior.User information refers to: user's portrait (such as: age, occupation, gender of user etc.), user consumption preferences (example
Such as: the classifications of the commodity often bought, price consume section) and other characterization users personal characteristics information.User is whole
The device status information at end refers to: the WIFI title and corresponding signal strength, user terminal that user terminal is scanned or is connected to
Device type, longitude and latitude, the IP address of user terminal of user terminal etc..
Illustratively, background server is that the background server of certain personal consumption class APP is correspondingly installed on user terminal
The terminal application software for supporting neighbouring function of search be with background server carry out communication interaction personal consumption class APP.
User inputs " chafing dish " in the search column of personal consumption class APP, then will pop up on personal consumption class APP displaying have it is multiple with
The page in " chafing dish " relevant shop, user can arbitrarily click a shop, inquire the relevant information in the shop, participate in the shop
Interior sudden strain of a muscle favour checks, purchase by group test certificate, the activities such as user registers.Personal consumption class APP records user in real time and asks for the secondary search
The user behavior data of generation is sought, while user information is obtained by the personal account of user, and by user terminal
Application program obtains the device status information of user terminal, utilizes setting for user behavior data, user information and user terminal
Standby status information generates the user data for this searching request and is sent to background server.
Step S12: described search is requested to carry out feature extraction, determines the feature, described of the user of the user terminal
The feature and user-shop cross feature in each shop in the corresponding shop list of searching request, the user-shop intersect
It is characterized in carrying out what characteristic crossover obtained to the feature of the user and the feature in each shop.
In each embodiment of the application, feature extraction is carried out to searching request from three dimensions, three dimensions are respectively as follows:
The cross-dimension of user's dimension, shop dimension and user and shop.Described search is requested to carry out feature extraction, specifically:
Feature extraction is carried out to user data corresponding with searching request.
Specifically, it is determined that the feature of the user of user terminal refers to: being extracted from user data and meet user's dimension
The feature of the user of data, the feature of the user as user terminal, such as user terminal may is that user terminal scanning or connect
The WIFI title and corresponding signal strength that are connected to, the device type of user terminal, the longitude and latitude of user terminal, user terminal
IP address, user's portrait of user and the consumption preferences of user.Certainly, the feature of user can also include it is other can be with table
Take over the feature of the individual character at family for use, the application includes but is not limited to the feature of the above-mentioned user enumerated.
Background server is previously stored with the information in all shops registered in terminal application software, such as: terminal
When application software is personal consumption class APP, background server, which is previously stored with, all have been registered on personal consumption class APP
The information in shop.After initiating searching request due to user terminal, background server can be returned to user terminal meets search condition
Shop list, accordingly, it is determined that the feature in each shop refers in the corresponding shop list of searching request: from pre-stored institute
There is the feature that each shop in the list of shop is extracted in the information in registered shop, wherein each shop in the list of shop
Feature may is that the mark in each shop, the WIFI title in each shop, each shop WIFI averagely connect or scan by force
The price range for the commodity that degree, the longitude and latitude in each shop, classification, each shop belonging to each shop are sold, each shop
Clicking rate and each shop visit purchase rate.Specifically, classification belonging to each shop characterizes the quotient that each shop is sold
The type of product, such as: the commodity that some shop is sold are dress ornaments, then classification belonging to the shop is dress ornament class, another example is: certain
The commodity that a shop is sold are snack or beverage, then classification belonging to the shop is food and drink class.The visit purchase rate in each shop is
Refer to: within a certain period of time, accessing in all customers in the shop, produce percentage shared by the customer of buying behavior.
User-shop cross feature is the linked character of user Yu each shop.Fig. 3 is that the implementation of the application one exemplifies
Obtain user-shop cross feature flow chart.Referring to Fig. 3, user-shop cross feature can specifically be obtained by following steps
:
Step S121: according to the longitude and latitude of the longitude and latitude of the user terminal and each shop, the user is determined
The linear distance of terminal and each shop.
Step S122: the signal strength of the WIFI in the shop that the user terminal is scanned or is connected to, with the user
The WIFI in the shop that terminal is scanned or is connected to averagely connects or scans intensity and carries out characteristic crossover.
Feature A and feature B progress characteristic crossover is referred to: feature A and feature B being counted using preset calculation method
It calculates to obtain the linked character C of feature A Yu feature B, such as: feature A is the WIFI in the shop that user terminal is scanned or is connected to
Signal strength, feature B is that the WIFI in shop for scanning or being connected to user terminal averagely connect or scan intensity, preset
Calculation method is characterized the ratio of A and feature B, then linked character C is characterized A divided by the obtained value of feature B.
Step S123: the signal strength of the WIFI in the shop that the user terminal is scanned or is connected to, with the user
The linear distance in the shop that terminal and the user terminal are scanned or be connected to carries out characteristic crossover；And/or it is whole to the user
The user at end clicks or the progress characteristic crossover of price per capita of consumption price and each shop.
The price per capita in each shop, that is, each shop pre-capita consumption price, user click price, that is, user and clicked
The price of commodity, customer consumption price are the price for the commodity that user bought.The present embodiment is in order to guarantee finally to measure in advance
The accuracy in the target shop arrived, in the signal strength of the WIFI in the shop that user terminal is scanned or is connected to, with user's end
After the WIFI in end scanning or the shop being connected to averagely connects or scan intensity progress characteristic crossover, following three can also be passed through
Any mode in kind mode carries out characteristic crossover:
1) signal strength of the WIFI in the shop that user terminal is scanned or is connected to is scanned with user terminal and user terminal
Or the linear distance in the shop being connected to carries out characteristic crossover.
2) price per capita that the user of user terminal clicks price (or customer consumption price) and each shop is subjected to spy
Sign is intersected.
3) signal strength of the WIFI in the shop that user terminal is scanned or is connected to, sweeps with user terminal and user terminal
The linear distance in the shop retouched or be connected to carries out characteristic crossover, and the user of user terminal is clicked price (or customer consumption
Price) with the price per capita in each shop carry out characteristic crossover.
Step S13: by the feature of the user, the feature in each shop and the user-shop cross feature
Shop Probabilistic Prediction Model is arrived in input training in advance, determines that the user reaches the probability in each shop.
It wherein, is to be trained to obtain to preset model using the search log of user terminal to shop Probabilistic Prediction Model
's.Specific training process will hereinafter be described in detail.
It is general that shop is input to using the feature of user, the feature in each shop and user-shop cross feature as input value
After rate prediction model, the probability value in each shop that user reaches in the list of shop can be exported to shop Probabilistic Prediction Model.User
It reaches the probability value in each shop and a possibility that user reaches the shop is directly proportional, probability value is higher, and user reaches the shop
A possibility that it is bigger, a possibility that probability value is lower, and user reaches the shop, is smaller.
Step S14: the probability in each shop is reached according to the user, determines whether the user reaches target shop
Paving, the target shop are one in each shop.
Target shop is the shop that is currently located of user, determines target shop, namely determines current accurate of user
Position.Determine whether user reaches the specific steps in target shop and be described below.
In one embodiment, background server is the background server of certain personal consumption class APP, is pacified on user terminal
The software for supporting neighbouring function of search of dress is the personal consumption class APP that communication interaction is carried out with background server, works as user
When inside megastore and by the neighbouring shop of personal consumption class APP search, background server is searched according to user's
Rope request return it is multiple meet search condition (search condition can be arranged from many aspects, such as: apart from user current location
Distance, the price range in shop, the favorable comment degree ranking of user) shop, so that user checks.Such as user can be in search column
It inputs " chafing dish ", and it is in the range of one km of current location that search condition, which is arranged, personal consumption class APP meets all
The shop relevant to " chafing dish " of search condition is shown in the page, and user can check or do to interested shop
Out relevant to some shop user behavior (as it was noted above, user behavior may include: general user's behavior and with to shop phase
The behavior of pass), personal consumption class APP records the user behavior data of user's generation in real time, and by user behavior data, Yong Huxin
It ceases, the device status information of user terminal is as the current searching request of user, (search name is that the search of " chafing dish " is asked
Ask) user data be sent to background server.Background server carries out feature extraction to user data, and by the feature of extraction
It is input to shop Probabilistic Prediction Model, to obtain the general of each shop relevant to " chafing dish " in user's arrival search result
Then rate obtains the shop where user's current time further according to each probabilistic forecasting.In each embodiment of the application, currently
Moment is one section of shorter duration at the time of initiating searching request comprising user.
In the embodiment of the present application, Client-initiated searching request is obtained first, and feature then is carried out to searching request and is mentioned
It takes, and (includes: each shop in the corresponding shop list of feature, searching request of the user of user terminal by the feature extracted
The feature of paving and user-shop cross feature) shop Probabilistic Prediction Model is input to obtain user and reach in the list of shop respectively
The probability in a shop finally obtains the target shop of user's arrival further according to these probabilistic forecastings.The application is from user's dimension, shop
Paving dimension and user-dimension of shop cross-dimension three carry out feature extraction to the searching request that user terminal is initiated, and improve
The accuracy of the result exported to shop Probabilistic Prediction Model and the accuracy for finally predicting obtained target shop, in addition,
The target shop for going out user's arrival by the determine the probability in each shop exported to shop Probabilistic Prediction Model, realizes to user
Current location positioning, improve the reliability in the shop near user terminal current location recommended to the user, enhance
The usage experience of user.
Specifically, step S14 may include:
Step S141: in the case where the probability that the user reaches the target shop is greater than preset probability threshold value,
Determine that the user reaches the target shop；Or
Step S142: it is greater than preset probability threshold value in the probability that the user reaches the target shop, and described searches
Rope requests corresponding parameter value in the case where preset come into force in range of parameter values, determines that the user reaches the target shop
Paving.
In the present embodiment, judge whether user reaches mode there are two types of target shops, the first decision procedure is: will arrive
The probability of shop Probabilistic Prediction Model output is compared with preset probability threshold value, and probability is greater than to the shop of preset probability threshold value
Paving is as a user to the target shop reached；Second of decision procedure is: probability being greater than in the shop of preset probability threshold value, is searched
Rope requests corresponding parameter value in the preset shop come into force in range of parameter values as a user to the target shop reached, and search is asked
The relationship of corresponding parameter value and the preset range of parameter values that comes into force is asked to determine whether user reaches target shop for assisting.
Illustratively, if preset probability threshold value is 0.8, only one in all probability exported to shop Probabilistic Prediction Model
The probability of a shop M is 0.9, then shop M is the target shop that user reaches according to the first decision procedure；According to second
Kind decision procedure can set user terminal for the corresponding parameter value of searching request if shop M is a food and drink class shop
At the time of current, by the preset range of parameter values that comes into force be set as the lunchtime (such as: 11:00-13:00), if user terminal
Preset come into force in range of parameter values is just fallen at the time of current, it may be determined that user reaches shop M.Certainly, searching request pair
The parameter value and the preset range of parameter values that comes into force answered can be determined according to the demand in the application actual application.
It in a practical situation, can if the period for the searching request that background server acquisition user terminal is initiated is longer
The phenomenon that probability value that can exist to multiple shops of shop Probabilistic Prediction Model output is greater than preset threshold.If background server
The period for obtaining the searching request that user terminal is initiated is shorter, then it is usual greater than the quantity in the shop of preset threshold to meet probability value
It is less, even zero, in this case, if the shop that a probability value is greater than preset threshold is only existed, by the shop
As the target shop that user is currently located, the shop of preset threshold is greater than if there is multiple probability values, then by multiple shops
The shop that the middle highest shop of probability value is currently located as user.Preset threshold is by calculating to shop Probabilistic Prediction Model
Accuracy rate and recall rate, the empirical value that obtained user to shop determines whether to come into force, it may be assumed that preset threshold is will to use
The target shop that shop Probabilistic Prediction Model predicts, after being compared and analyzed with the shop of user being actually reached, in conjunction with specific
The obtained optimal probability value of business, only under the optimal probability value, target shop that the user that predicts reaches
Accuracy highest.
In the embodiment of the present application, the method that can determine that whether user reaches target shop provided with two kinds enhances this
The flexibility of shop prediction technique in actual application is arrived in application, is asked in addition, being additionally arranged search in second of decision procedure
Ask corresponding parameter value whether in the preset range of parameter values that comes into force this auxiliary judgement condition, improve determine result standard
Exactness.
The training process to shop Probabilistic Prediction Model will be illustrated below.
Fig. 4 is the flow chart that the application one implements a kind of training method to shop Probabilistic Prediction Model exemplified.Reference
Fig. 4, the training method include:
Step S21: it is determining with user to the associated user behavior type in shop.
Refer to user to the associated user behavior in shop: user must reach the user behavior that could be completed behind shop,
Such as: it dodges favour and checks, purchases by group and test certificate, Self-help ordering, the number of taking are lined up, upload UGC, user registers, connects WIFI etc..For one
Shop, if user produce it is larger to a possibility that the relevant user behavior in shop, user is actually reached the shop.
Step S22: extracting first kind search record and the second class search record from the search log of the user terminal,
The first kind search is recorded as meeting the search record of the user behavior type, and the second class search is recorded as corresponding
Search of the time difference at the search moment of moment and first kind search record in preset duration is searched for record.
It is recorded in the search log of user terminal comprising all search, searching request of Client-initiated can correspond to
A plurality of search record, a plurality of search record generated when initiating searching request every time constitute the user behavior of the secondary searching request
Data.For example, user inputs " dress ornament " in search column, then being with the searching request that the secondary search name is " dress ornament "
Searching request X, after clicking search, user terminal can show multiple shops for meeting " Chinese meal " this search name, if user 1
It has checked shop A and has completed to register in the A of shop, recorded then user 1- searching request X- shop A- registers for a search,
If user 1, which has checked shop A and taken part in purchase by group in the A of shop, tests certificate activity, user 1- searching request X- shop A-
Purchasing by group and testing certificate is another search record, and similarly, user 1, which is also based in other shops, generates a plurality of search record.
Wherein, first kind search is recorded as meeting the search record of user behavior type, as long as carried in search record
User behavior is that user must reach the user behavior that could be generated behind corresponding shop, this article search record can be by as the
One kind search record.Such as: a search is recorded as user 1- searching request X- shop A- and purchases by group to test certificate, tests certificate due to purchasing by group and is
It is required that user reaches and could complete behind corresponding shop, therefore this search for record can be used as a first kind search record,
Another example is: a search is recorded as user 1- searching request X- shop A- sharing, user is not required to reach accordingly due to sharing
Shop, therefore this search record cannot function as a first kind search record.
Optionally, after selecting a plurality of first kind search record, screening conditions, such as screening item can also be further set
Part can be time range condition, and the corresponding search moment in all search record is located at the first kind in certain a period of time and is searched
Suo Jilu is screened, and searches for record as the new first kind.
In the present embodiment, first kind search record only searches for the search record for meeting user behavior type in log
In a part search record, for make acquisition search record be reasonably distributed, also need acquisition the second class search record.Second class
Search is recorded as search of the time difference at the search moment of corresponding search moment and first kind search record in preset duration
Record.For example, the search moment of a first kind search record is 10:00, if preset duration is 1 minute, by this
The second class search record that first kind search record obtains can be the search record that user generates between 9:59-10:01.
In the present embodiment using extraction first kind search record and the search of the second class record by the way of ensure that and extract
Search record have preferable distributivity (such as: in advance extract Annual distribution it is relatively reasonable the first kind search record, then
The second class search record is extracted based on these first kind search record), it avoids not extracting and a certain number of meets use
Family behavior type search record the phenomenon that, or extract search record Annual distribution it is unreasonable (such as: search record
Concentrations are in some period) the phenomenon that.
Step S23: the user behavior class will be met in first kind search record and second class search record
The search recording mark of type is positive sample, and, the user behavior type will not be met in second class search record
Search recording mark is negative sample.
Positive sample indicates that user theoretically reaches shop, and negative sample indicates user theoretically without reaching shop.Due to
User behavior in all first kind search records be with user to the associated user behavior in shop, thus, all the
One kind search record is positive sample.Due to meeting searching for user behavior type in first kind search record only search log
A part search record in Suo Jilu, thus, there is likely to be have a plurality of to meet user behavior class in the second class search record
The search of type records, therefore also needs to search for the second class in record when dividing positive negative sample and meet searching for user behavior type
Rope recording mark is positive sample, is negative sample by the search recording mark for not meeting user behavior type.
Step S24: according to the positive sample and the negative sample, being trained preset model, obtains described general to shop
Rate prediction model.
Obtained positive sample and negative sample will be divided in the present embodiment as input value, using machine learning algorithm (such as:
Logistic regression algorithm) preset model is trained to obtain one two classification prediction model (that is: to shop Probabilistic Prediction Model),
Effect is: when the feature that input is extracted from searching request, can export each shop in user's arrival shop list
Probability, shop list are the corresponding search result of searching request.
Fig. 5 be the application one implement the another kind that exemplifies to shop Probabilistic Prediction Model training method flow chart.Ginseng
According to Fig. 5, step S24 includes:
Step S241: feature extraction is carried out to the positive sample and the negative sample respectively, determines the positive sample and institute
It states each in the corresponding shop list of feature, the positive sample and the negative sample of the corresponding sample of users of negative sample
The feature and sample of users in a sample shop-sample shop cross feature, the sample of users-sample shop cross feature are
The feature of feature and each sample shop to the sample of users carries out what characteristic crossover obtained.
Step S242: with the feature of the sample of users, the feature and the sample of users-in each sample shop
Sample shop cross feature is training sample, is trained to the preset model, is obtained described to shop Probabilistic Prediction Model.
In the present embodiment, every search record also carries the device status information of user information and user terminal, In
After obtaining positive sample and negative sample, the feature of corresponding sample of users, corresponding sample can be extracted from each sample
The feature and sample of users in shop-sample shop cross feature, then again by this Partial Feature input preset model and to pre-
If model is trained, obtain to shop Probabilistic Prediction Model.
Fig. 6 be the application one implement a kind of update for exemplifying to shop Probabilistic Prediction Model method flow chart.Reference
6, this method comprises:
Step S31: extracted from the search log of the user terminal the corresponding search moment determine the user to
Up to the search record after the target shop.
After according to the target shop for predicting to obtain user's arrival to shop Probabilistic Prediction Model, from the search of user terminal
Search record of the corresponding search moment after determining that user reaches target shop is extracted in log, this part searches is recorded
As feedback record, the adjustment to sample weights may be implemented using feedback record, convenient for carrying out to shop Probabilistic Prediction Model
It updates.Such as: user actually only reaches shop A, but may be made that other shops associated to shop with user
User behavior, can be at this time that positive sample and adjust weighted value by sample labeling relevant to shop A, by the sample in other shops
This is labeled as positive and negative and adjusts weighted value, reliability when guaranteeing that each sample uses.
Step S32: it in the case where the search record of extraction is the search record for the target shop, is mentioned described
The search recording mark taken is positive sample, and increases the weight of the positive sample.
For example, the target shop that certain prediction obtains user's arrival is shop A, can be with then in feedback record
It regard search record relevant to shop A as the biggish positive sample of weighted value, such as: when in search relevant to shop A record
User behavior be when purchasing by group certificate behavior of testing (or other user behaviors, such as: uploading UGC, connection WIFI), can will
The search recording mark is the biggish positive sample of weighted value.
Optionally, according to the difference of the user behavior in feedback record, the present embodiment is to different positive sample (feedback records
In search record relevant to target shop) setting weighted value of different sizes, such as: in positive sample, user behavior is to purchase by group
It is weighted value corresponding to the sample of click behavior that weighted value corresponding to the sample of certificate behavior, which is tested, greater than user behavior.
Step S33:, will in the case where the search record of the extraction is not the search record for the target shop
The search recording mark of the extraction is negative sample, and reduces the weight of the negative sample.
In the present embodiment, the weight of unrelated with shop A search record can also be adjusted, specifically, will with shop A without
The search recording mark of pass is negative sample, and reduces the weight of the negative sample.Such as: the user in extracted search record
It, can be by the search when whether behavior purchases by group certificate behavior of testing (or other user behaviors, such as: uploading UGC, connection WIFI)
Recording mark is the lesser negative sample of weighted value.
Optionally, the size of the weighted value of different negative samples is also adjustable, according to the user in feedback record
The difference of behavior, the present embodiment is to different negative sample (search unrelated with target shop records in feedback record) setting sizes
Different weighted values.Such as: in negative sample, user behavior is to purchase by group weighted value corresponding to the sample for testing certificate greater than user behavior
It is weighted value corresponding to the sample of click behavior.
Step S34: according to the negative sample after the positive sample and reduction weight after increase weight, shop probabilistic forecasting is arrived to described
Model is updated.
In the present embodiment, using all feedback records as after to sample when being updated to shop Probabilistic Prediction Model
This, realizes the prediction process excavated feedback record from search log and react on model, is arrived according to the user determined
The target shop that reaches and the feedback behavior for determining that user generates after target shop, in feedback record positive sample and negative sample
The adjustment of this progress weighted value ensure that the reliability of the sample used when updating to shop Probabilistic Prediction Model, in reliable sample
In this fairly large number of situation, realizes to the continuous iteration optimization to shop Probabilistic Prediction Model, effectively improve to shop
The accuracy of the prediction result of Probabilistic Prediction Model, and the mesh predicted according to the prediction result to shop Probabilistic Prediction Model
Mark the accuracy in shop.
Can more precisely be positioned to shop prediction technique to user in the embodiment of the present application, improves near O2O
The accuracy that distance calculates in search, suitable for the scene of a variety of distances to user that need to calculate shop, such as all kinds of service types
The Perimeter service (including: fresh, fresh flower dispatching of cuisines, fruits and vegetables nearby etc.) of software, can provide more good search for user
Cable body is tested.In addition, the application proposition is not only simple and easy, at low cost to shop prediction technique, but also feedback record can also be utilized
To constantly iteration updates to shop prediction model, allow the accuracy for the output result for obtaining shop prediction model with search industry
The accumulation of the development of business scale and search data volume (such as: search record) and it is higher and higher, to realize to user more
Accurately position.
Based on the same inventive concept, one embodiment of the application provides a kind of to shop prediction meanss.Fig. 7 is that the application one is implemented
A kind of schematic diagram to shop prediction meanss exemplified.Referring to Fig. 7, which includes:
Module 701 is obtained, for obtaining the searching request of user terminal initiation；
Characteristic extracting module 702 carries out feature extraction for requesting described search, determines the user of the user terminal
Feature, described search request the feature in each shop and user-shop cross feature, the use in corresponding shop list
Family-shop cross feature is to carry out characteristic crossover to the feature of the user and the feature in each shop to obtain；
Probabilistic forecasting module 703, for by the feature of the user, the feature in each shop and the user-
Shop Probabilistic Prediction Model is arrived in shop cross feature input training in advance, determines that the user reaches the general of each shop
Rate；
Determining module 704 determines whether the user arrives for reaching the probability in each shop according to the user
Up to target shop, the target shop is one in each shop.
Optionally, the determining module includes:
First determining module, the probability for reaching the target shop in the user are greater than preset probability threshold value
In the case of, determine that the user reaches the target shop；Or
Second determining module, the probability for reaching the target shop in the user are greater than preset probability threshold value,
And described search requests corresponding parameter value in the case where preset come into force in range of parameter values, determines that the user reaches institute
State target shop.
Optionally, described device further include:
Third determining module, for it is determining with user to the associated user behavior type in shop；
First extraction module, for extracting first kind search record and the second class from the search log of the user terminal
Search record, the first kind search are recorded as meeting the search record of the user behavior type, the second class search note
Record is that search of the time difference at the search moment of corresponding search moment and first kind search record in preset duration is remembered
Record；
Mark module, for user's row will to be met in first kind search record and second class search record
It is positive sample for the search recording mark of type, and, the user behavior class will not be met in second class search record
The search recording mark of type is negative sample；
Training module obtains described arrive for being trained to preset model according to the positive sample and the negative sample
Shop Probabilistic Prediction Model.
Optionally, the training module includes:
Feature extraction submodule, for carrying out feature extraction respectively to the positive sample and the negative sample, determine described in
The feature of positive sample and the corresponding sample of users of the negative sample, the positive sample and the corresponding shop of the negative sample
Spread the feature and sample of users-sample shop cross feature in each sample shop in list, the sample of users-sample shop
Cross feature is to carry out characteristic crossover to the feature of the sample of users and the feature in each sample shop to obtain；
Training submodule, for the feature of the sample of users, the feature in each sample shop and the sample
This user-sample shop cross feature is training sample, is trained to the preset model, is obtained described to shop probabilistic forecasting
Model.
Optionally, described device further include:
Second extraction module, for extracting the corresponding search moment from the search log of the user terminal determining
It states user and reaches search record after the target shop；
First weight adjusts module, is the feelings for the search record in the target shop for the search record in extraction
It is positive sample by the search recording mark of the extraction, and increase the weight of the positive sample under condition；
Second weight adjusts module, is not the search note for the target shop for the search record in the extraction
It is negative sample by the search recording mark of the extraction, and reduce the weight of the negative sample in the case where record；And
Update module, for according to increase weight after positive sample and reduce weight after negative sample, to it is described to shop it is general
Rate prediction model is updated.
Optionally, the feature of the user include it is following at least one: the user terminal scanning or the WIFI that is connected to
Title and corresponding signal strength, the device type of the user terminal, the longitude and latitude of the user terminal, the user terminal
IP address, the user user portrait and the user consumption preferences.
Optionally, the feature in each shop include it is following at least one: it is the mark in each shop, described each
The WIFI title in shop, each shop WIFI averagely connect or scan intensity, each shop longitude and latitude, described
The price range for the commodity that classification belonging to each shop, each shop are sold, the clicking rate in each shop and
The visit purchase rate in each shop.
Optionally, the user-shop cross feature is obtained by following at least one mode:
According to the longitude and latitude of the longitude and latitude of the user terminal and each shop, determine the user terminal with it is described
The linear distance in each shop；
The signal strength for the WIFI in shop that the user terminal is scanned or is connected to, with user terminal scanning or
The WIFI in the shop being connected to averagely connects or scans intensity and carries out characteristic crossover；
The signal strength for the WIFI in shop that the user terminal is scanned or is connected to, with the user terminal with it is described
The linear distance in the shop that user terminal is scanned or is connected to carries out characteristic crossover；And/or
Feature friendship is carried out to user's click of the user terminal or consumption price and the price per capita in each shop
Fork.
Based on the same inventive concept, another embodiment of the application provides a kind of computer readable storage medium, stores thereon
There is computer program, the step in the method as described in any of the above-described embodiment of the application is realized when which is executed by processor
Suddenly.
Based on the same inventive concept, another embodiment of the application provides a kind of electronic equipment, including memory, processor and
The computer program that can be run on a memory and on a processor is stored, the processor realizes the application above-mentioned when executing
Step in method described in one embodiment.
For device embodiment, since it is basically similar to the method embodiment, related so being described relatively simple
Place illustrates referring to the part of embodiment of the method.
All the embodiments in this specification are described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiments, the same or similar parts between the embodiments can be referred to each other.
It should be understood by those skilled in the art that, the embodiments of the present application may be provided as method, apparatus or calculating
Machine program product.Therefore, the embodiment of the present application can be used complete hardware embodiment, complete software embodiment or combine software and
The form of the embodiment of hardware aspect.Moreover, the embodiment of the present application can be used one or more wherein include computer can
With in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of program code
The form of the computer program product of implementation.
The embodiment of the present application is referring to according to the method for the embodiment of the present application, terminal device (system) and computer program
The flowchart and/or the block diagram of product describes.It should be understood that flowchart and/or the block diagram can be realized by computer program instructions
In each flow and/or block and flowchart and/or the block diagram in process and/or box combination.It can provide these
Computer program instructions are set to general purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminals
Standby processor is to generate a machine, so that being held by the processor of computer or other programmable data processing terminal devices
Capable instruction generates for realizing in one or more flows of the flowchart and/or one or more blocks of the block diagram
The device of specified function.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing terminal devices
In computer-readable memory operate in a specific manner, so that instruction stored in the computer readable memory generates packet
The manufacture of command device is included, which realizes in one side of one or more flows of the flowchart and/or block diagram
The function of being specified in frame or multiple boxes.
These computer program instructions can also be loaded into computer or other programmable data processing terminal devices, so that
Series of operation steps are executed on computer or other programmable terminal equipments to generate computer implemented processing, thus
The instruction executed on computer or other programmable terminal equipments is provided for realizing in one or more flows of the flowchart
And/or in one or more blocks of the block diagram specify function the step of.
Although preferred embodiments of the embodiments of the present application have been described, once a person skilled in the art knows bases
This creative concept, then additional changes and modifications can be made to these embodiments.So the following claims are intended to be interpreted as
Including preferred embodiment and all change and modification within the scope of the embodiments of the present application.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that process, method, article or terminal device including a series of elements not only wrap
Those elements are included, but also including other elements that are not explicitly listed, or further includes for this process, method, article
Or the element that terminal device is intrinsic.In the absence of more restrictions, being wanted by what sentence "including a ..." limited
Element, it is not excluded that there is also other identical elements in process, method, article or the terminal device for including the element.
Above to one kind provided herein to shop prediction technique, device, readable storage medium storing program for executing and electronic equipment, carry out
It is discussed in detail, specific examples are used herein to illustrate the principle and implementation manner of the present application, above embodiments
Explanation be merely used to help understand the present processes and its core concept；At the same time, for those skilled in the art,
According to the thought of the application, there will be changes in the specific implementation manner and application range, in conclusion in this specification
Hold the limitation that should not be construed as to the application.
Claims (11)
1. one kind arrives shop prediction technique, which is characterized in that the described method includes:
Obtain the searching request that user terminal is initiated；
Described search is requested to carry out feature extraction, determines that the feature of the user of the user terminal, described search request correspond to
Shop list in each shop feature and user-shop cross feature, the user-shop cross feature is to described
The feature of user and the feature in each shop carry out what characteristic crossover obtained；
By the feature of the user, the feature in each shop and the user-shop cross feature input training in advance
Arrive shop Probabilistic Prediction Model, determine that the user reaches the probability in each shop；
The probability that each shop is reached according to the user, determines whether the user reaches target shop, the target
Shop is one in each shop.
2. the method according to claim 1, wherein described reach the general of each shop according to the user
Rate determines the step of whether user reaches target shop, comprising:
In the case where the probability that the user reaches the target shop is greater than preset probability threshold value, determine that the user arrives
Up to the target shop；Or
It is greater than preset probability threshold value in the probability that the user reaches the target shop, and described search requests corresponding ginseng
Numerical value determines that the user reaches the target shop in the case where preset come into force in range of parameter values.
3. the method according to claim 1, wherein the method also includes:
It is determining with user to the associated user behavior type in shop；
First kind search record and the second class search record are extracted from the search log of the user terminal, the first kind is searched
Rope is recorded as meeting the search record of the user behavior type, and the second class search is recorded as corresponding search moment and institute
State search record of the time difference at the search moment of first kind search record in preset duration；
The search record of the user behavior type will be met in first kind search record and second class search record
Labeled as positive sample, and, the search recording mark of the user behavior type will not be met in second class search record
For negative sample；
According to the positive sample and the negative sample, preset model is trained, is obtained described to shop Probabilistic Prediction Model.
4. according to the method described in claim 3, it is characterized in that, described according to the positive sample and the negative sample, to pre-
If model is trained, obtain it is described to shop Probabilistic Prediction Model the step of, comprising:
Feature extraction is carried out to the positive sample and the negative sample respectively, determines that the positive sample and the negative sample are respectively right
The spy in each sample shop in the feature for the sample of users answered, the positive sample and the corresponding shop list of the negative sample
Sign and sample of users-sample shop cross feature, the sample of users-sample shop cross feature is to the sample of users
Feature and the feature in each sample shop carry out characteristic crossover and obtain；
Intersected with the feature of the sample of users, the feature in each sample shop and the sample of users-sample shop
Feature is training sample, is trained to the preset model, is obtained described to shop Probabilistic Prediction Model.
5. according to the method described in claim 2, it is characterized in that, reaching the target shop in the determination user
After step, the method also includes:
Extracting the corresponding search moment from the search log of the user terminal is determining that the user reaches the target shop
Search record after paving；
In the case where the search record of extraction is the search record for the target shop, the search of the extraction is recorded
Labeled as positive sample, and increase the weight of the positive sample；
In the case where the search record of the extraction is not the search record for the target shop, by searching for the extraction
Rope recording mark is negative sample, and reduces the weight of the negative sample；And
According to the negative sample after the positive sample and reduction weight after increase weight, carried out more to described to shop Probabilistic Prediction Model
Newly.
6. the method according to claim 1, wherein the feature of the user include it is following at least one: it is described
User terminal scanning or the WIFI title being connected to and corresponding signal strength, the device type of the user terminal, the use
The longitude and latitude of family terminal, the IP address of the user terminal, user's portrait of the user and the consumption of the user are inclined
It is good.
7. the method according to claim 1, wherein the feature in each shop include it is following at least one:
The mark in each shop, the WIFI title in each shop, each shop WIFI averagely connect or scan by force
The price area for the commodity that degree, the longitude and latitude in each shop, classification, each shop belonging to each shop are sold
Between, the visit purchase rate in the clicking rate in each shop and each shop.
8. the method according to the description of claim 7 is characterized in that the user-shop cross feature be by it is following at least
What a kind of mode obtained:
According to the longitude and latitude of the longitude and latitude of the user terminal and each shop, determine the user terminal with it is described each
The linear distance in shop；
The signal strength of the WIFI in the shop that the user terminal is scanned or is connected to is scanned or is connect with the user terminal
The WIFI in the shop arrived averagely connects or scans intensity and carries out characteristic crossover；
The signal strength of the WIFI in the shop that the user terminal is scanned or is connected to, with the user terminal and the user
The linear distance in the shop that terminal is scanned or is connected to carries out characteristic crossover；And/or
Characteristic crossover is carried out to user's click of the user terminal or consumption price and the price per capita in each shop.
9. one kind arrives shop prediction meanss, which is characterized in that described device includes:
Module is obtained, for obtaining the searching request of user terminal initiation；
Characteristic extracting module, for described search request carry out feature extraction, determine the user of the user terminal feature,
Described search requests the feature in each shop and user-shop cross feature, the user-shop in corresponding shop list
Cross feature is to carry out characteristic crossover to the feature of the user and the feature in each shop to obtain；
Probabilistic forecasting module, for intersecting the feature of the feature of the user, each shop and the user-shop
Shop Probabilistic Prediction Model is arrived in feature input training in advance, determines that the user reaches the probability in each shop；And
Determining module determines whether the user reaches target for reaching the probability in each shop according to the user
Shop, the target shop are one in each shop.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
The step in method a method as claimed in any one of claims 1-8 is realized when execution.
11. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine program, which is characterized in that the step of processor realizes method a method as claimed in any one of claims 1-8 when executing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910610494.8A CN110472995A (en) | 2019-07-08 | 2019-07-08 | To shop prediction technique, device, readable storage medium storing program for executing and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910610494.8A CN110472995A (en) | 2019-07-08 | 2019-07-08 | To shop prediction technique, device, readable storage medium storing program for executing and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110472995A true CN110472995A (en) | 2019-11-19 |
Family
ID=68507089
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910610494.8A Pending CN110472995A (en) | 2019-07-08 | 2019-07-08 | To shop prediction technique, device, readable storage medium storing program for executing and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110472995A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111194005A (en) * | 2019-12-05 | 2020-05-22 | 中国科学院地理科学与资源研究所 | Indoor pedestrian semantic position extraction method and prediction method |
| CN111275479A (en) * | 2020-01-07 | 2020-06-12 | 北京爱笔科技有限公司 | People flow prediction method, device and system |
| CN112561569A (en) * | 2020-12-07 | 2021-03-26 | 上海明略人工智能(集团)有限公司 | Dual-model-based arrival prediction method and system, electronic device and storage medium |
| CN112950865A (en) * | 2019-12-10 | 2021-06-11 | 东芝泰格有限公司 | Shopping support device, server, storage medium, and shopping support method |
| CN113543117A (en) * | 2020-04-22 | 2021-10-22 | 中国移动通信集团重庆有限公司 | Prediction method and device for number portability user and computing equipment |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663033A (en) * | 2012-03-23 | 2012-09-12 | 汉海信息技术(上海)有限公司 | Method for searching interest points in designated area of map by hand-drawing way |
| US20140188897A1 (en) * | 2013-01-02 | 2014-07-03 | CrowdChunk LLC | CrowdChunk System, Method and Computer Program Product for Searching Summaries of Mobile Apps Reviews |
| CN105446972A (en) * | 2014-06-17 | 2016-03-30 | 阿里巴巴集团控股有限公司 | Search method, device and system based on and fusing with user relation data |
| JP2016091535A (en) * | 2014-10-31 | 2016-05-23 | 株式会社東芝 | Item recommendation system, item recommendation method and program |
| CN106055607A (en) * | 2016-05-25 | 2016-10-26 | 百度在线网络技术（北京）有限公司 | User visiting prediction model establishment and user visiting prediction method and apparatus |
| CN106485562A (en) * | 2015-09-01 | 2017-03-08 | 苏宁云商集团股份有限公司 | A kind of commodity information recommendation method based on user's history behavior and system |
| CN107203518A (en) * | 2016-03-16 | 2017-09-26 | 阿里巴巴集团控股有限公司 | Method, system and device, the electronic equipment of on-line system personalized recommendation |
| CN107481093A (en) * | 2017-07-21 | 2017-12-15 | 北京京东尚科信息技术有限公司 | Personalized shop Forecasting Methodology and device |
| CN107609960A (en) * | 2017-10-18 | 2018-01-19 | 口碑(上海)信息技术有限公司 | Rationale for the recommendation generation method and device |
| CN108230040A (en) * | 2018-01-17 | 2018-06-29 | 口碑(上海)信息技术有限公司 | To shop Forecasting Methodology and device |
-
2019
- 2019-07-08 CN CN201910610494.8A patent/CN110472995A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663033A (en) * | 2012-03-23 | 2012-09-12 | 汉海信息技术(上海)有限公司 | Method for searching interest points in designated area of map by hand-drawing way |
| US20140188897A1 (en) * | 2013-01-02 | 2014-07-03 | CrowdChunk LLC | CrowdChunk System, Method and Computer Program Product for Searching Summaries of Mobile Apps Reviews |
| CN105446972A (en) * | 2014-06-17 | 2016-03-30 | 阿里巴巴集团控股有限公司 | Search method, device and system based on and fusing with user relation data |
| JP2016091535A (en) * | 2014-10-31 | 2016-05-23 | 株式会社東芝 | Item recommendation system, item recommendation method and program |
| CN106485562A (en) * | 2015-09-01 | 2017-03-08 | 苏宁云商集团股份有限公司 | A kind of commodity information recommendation method based on user's history behavior and system |
| CN107203518A (en) * | 2016-03-16 | 2017-09-26 | 阿里巴巴集团控股有限公司 | Method, system and device, the electronic equipment of on-line system personalized recommendation |
| CN106055607A (en) * | 2016-05-25 | 2016-10-26 | 百度在线网络技术（北京）有限公司 | User visiting prediction model establishment and user visiting prediction method and apparatus |
| CN107481093A (en) * | 2017-07-21 | 2017-12-15 | 北京京东尚科信息技术有限公司 | Personalized shop Forecasting Methodology and device |
| CN107609960A (en) * | 2017-10-18 | 2018-01-19 | 口碑(上海)信息技术有限公司 | Rationale for the recommendation generation method and device |
| CN108230040A (en) * | 2018-01-17 | 2018-06-29 | 口碑(上海)信息技术有限公司 | To shop Forecasting Methodology and device |
Non-Patent Citations (1)
| Title |
|---|
| 李峰;李军怀;王瑞林;张;: "基于商品特征的个性化推荐算法", 计算机工程与应用, no. 17, 11 June 2007 (2007-06-11) * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111194005A (en) * | 2019-12-05 | 2020-05-22 | 中国科学院地理科学与资源研究所 | Indoor pedestrian semantic position extraction method and prediction method |
| CN112950865A (en) * | 2019-12-10 | 2021-06-11 | 东芝泰格有限公司 | Shopping support device, server, storage medium, and shopping support method |
| CN112950865B (en) * | 2019-12-10 | 2023-04-07 | 东芝泰格有限公司 | Shopping support device, server, storage medium, and shopping support method |
| CN111275479A (en) * | 2020-01-07 | 2020-06-12 | 北京爱笔科技有限公司 | People flow prediction method, device and system |
| CN111275479B (en) * | 2020-01-07 | 2023-11-10 | 北京爱笔科技有限公司 | People flow prediction method, device and system |
| CN113543117A (en) * | 2020-04-22 | 2021-10-22 | 中国移动通信集团重庆有限公司 | Prediction method and device for number portability user and computing equipment |
| CN113543117B (en) * | 2020-04-22 | 2022-10-04 | 中国移动通信集团重庆有限公司 | Prediction method and device for number portability user and computing equipment |
| CN112561569A (en) * | 2020-12-07 | 2021-03-26 | 上海明略人工智能(集团)有限公司 | Dual-model-based arrival prediction method and system, electronic device and storage medium |
| CN112561569B (en) * | 2020-12-07 | 2024-02-27 | 上海明略人工智能(集团)有限公司 | Dual-model-based store arrival prediction method, system, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110472995A (en) | To shop prediction technique, device, readable storage medium storing program for executing and electronic equipment | |
| US11449919B2 (en) | Commodity recommendation method and commodity recommendation device | |
| CN104965920B (en) | Method and apparatus for determining destination address | |
| CN107465741B (en) | Information pushing method and device | |
| US9349128B1 (en) | Targeted content delivery | |
| US8775351B2 (en) | System and method for providing recommendations with a location-based service | |
| CN105247915B (en) | Context-sensitive is associated engine platform | |
| KR102340463B1 (en) | Sample weight setting method and device, electronic device | |
| JP2019508766A (en) | System, method, and device for generating a heat map of a geographical area | |
| CN108022152A (en) | The automatic commending system of user's commodity and recommendation method based on image recognition | |
| CN109151890A (en) | A kind of mobile terminal locating method and device | |
| JP2013058213A5 (en) | ||
| CA2829252A1 (en) | System and method for providing recommendations with a location-based service | |
| JPWO2010140504A1 (en) | RECOMMENDATION INFORMATION PROVIDING SYSTEM, DEVICE, METHOD, AND PROGRAM | |
| US11729576B2 (en) | Targeted content delivery | |
| WO2012147374A1 (en) | Information processing system, information processing method, program, and information recording medium | |
| US10445813B2 (en) | Tap walls that recommend products based on in-store and out-of-store interactions of anonymous users | |
| CN109711917A (en) | Information-pushing method and device | |
| CN104751345A (en) | Position based data processing method and position based data processing device | |
| CN106878938A (en) | A kind of information-pushing method, location positioning method and equipment | |
| CN113360790A (en) | Information recommendation method and device and electronic equipment | |
| CN112395486B (en) | Broadband service recommendation method, system, server and storage medium | |
| CN109615412A (en) | Cell average price predictor method, electronic device and storage medium | |
| KR101865216B1 (en) | A method of user preference customized evaluation management | |
| JP2020190964A (en) | Device, method, and program for processing information |
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 |