CN107203644A - The recommendation method and apparatus of cuisines data - Google Patents
The recommendation method and apparatus of cuisines data Download PDFInfo
- Publication number
- CN107203644A CN107203644A CN201710484775.4A CN201710484775A CN107203644A CN 107203644 A CN107203644 A CN 107203644A CN 201710484775 A CN201710484775 A CN 201710484775A CN 107203644 A CN107203644 A CN 107203644A
- Authority
- CN
- China
- Prior art keywords
- data
- user
- active user
- recommending
- dispensing
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- Databases & Information Systems (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a kind of recommendation method and apparatus of cuisines data, it is related to the technical field of data processing, this method includes:Obtain the dispensing data of target web;It is predicted using the neutral net pre-established to delivering data, to determine recommending data in data are delivered, wherein, recommending data meets the data of default interest-degree for interest-degree in dispensing data, interest-degree is interest-degree of the active user to dispensing data, and the neutral net pre-established is trained obtained neutral net for the user behavior data in advance using different time sections active user;Recommend recommending data to active user, alleviate existing data recommendation scheme and recommend the poor technical problem of precision.
Description
Technical field
The present invention relates to the technical field of data processing, more particularly, to a kind of recommendation method and apparatus of cuisines data.
Background technology
With the fast development of internet, internet provides more easily data acquiring mode, and ditch to everybody
Logical mode.For example, user can browse viewing video in some webpage, wherein, the webpage is also based on being used to browse
The video of viewing is the video that the user recommends that this is used to may like, to improve Consumer's Experience.Traditional suggested design is logical
Cross collection user behavior and content tab is liked to user's mark, then user's content recommendation is given by label.This way of recommendation
Have the disadvantage that content tab is excessively extensive to the mark of user preferences, increase to improve with collection user behavior and recommend to user
The precision of content.
The content of the invention
It is an object of the invention to provide a kind of recommendation method and apparatus of cuisines data, to alleviate existing data recommendation
Scheme recommends the poor technical problem of precision.
According to an aspect of the invention, there is provided a kind of recommendation method of cuisines data, including:Obtain target web
Deliver data;The dispensing data are predicted using the neutral net pre-established, to be determined in the dispensing data
Recommending data, wherein, the recommending data meets the data of default interest-degree, the interest for interest-degree in the dispensing data
The interest-degree to the dispensing data for active user is spent, the neutral net pre-established is to use different time sections in advance
The user behavior data of the active user is trained obtained neutral net;Recommend the recommendation number to the active user
According to.
Further, the dispensing data include a plurality of dispensing data, are thrown using the neutral net pre-established described
Put data to be predicted, to determine that recommending data includes in the dispensing data:Using the neutral net pre-established
The dispensing data are predicted, to obtain every first probability for delivering data, wherein, first probability is represented
The active user is to the current interest-degree for delivering data;Interest-degree in the dispensing data is more than or equal to default interest
The dispensing data of degree are used as the recommending data.
Further, before the dispensing data of target web are obtained, methods described also includes:Judge whether to have set up with
The corresponding neural network model of the active user;If it is judged that having set up the nerve net corresponding with the active user
Network model, then gather the user behavior data of the active user in preset time period, wherein, the user behavior data is root
According to the active user to having delivered the data that the operation performed by data is obtained;Work as using described in the preset time period
The user behavior data of preceding user is trained to the neural network model set up, so that the neural network model pair
Data interested to the active user are remembered.
Further, the user behavior data of the active user includes in collection preset time period:Every predetermined interval
Time gathers the user behavior data of the active user in the preset time period;Worked as using described in the preset time period
The user behavior data of preceding user the neural network model set up is trained including:When the predetermined interval
Between the neural network model set up is entered using the user behavior data of the active user in the preset time period
Row training.
Further, before the recommending data is recommended to the active user, methods described also includes:Obtain described
The subscriber data of active user, wherein, the subscriber data includes the native place of active user, and the sex of active user is current to use
The name at family;The recommending data is screened according to the subscriber data, the recommending data after being screened;To
The active user recommends the recommending data to include:The recommending data after screening is pushed to the active user.
Further, after the recommending data is recommended to the active user, methods described also includes:According to default
Form is stored to the recommending data according to the recommendation time, is stored in database, wherein, the database is used to store
History recommending data;The query statement of user is obtained, and query history is recommended in the database according to the query statement
Data.
According to another aspect of the present invention, a kind of recommendation apparatus of cuisines data is additionally provided, including:First obtains single
Member, the dispensing data for obtaining target web;Predicting unit, for using the neutral net pre-established to the dispensing number
According to being predicted, to determine recommending data in the dispensing data, wherein, the recommending data is emerging in the dispensing data
Interesting degree meets the data of default interest-degree, and the interest-degree is interest-degree of the active user to the dispensing data, described advance
The neutral net of foundation is trained obtained god for the advance user behavior data using active user described in different time sections
Through network;Recommendation unit, for recommending the recommending data to the active user.
Further, the dispensing data include a plurality of dispensing data, and the predicting unit includes:Prediction module, is used for
The dispensing data are predicted using the neutral net pre-established, to obtain the first of every dispensing data
Probability, wherein, first probability represents the active user to the current interest-degree for delivering data;Determining module, for inciting somebody to action
Interest-degree is used as the recommending data more than or equal to the dispensing data of default interest-degree in the dispensing data.
Further, described device also includes:Judging unit, for before the dispensing data of target web are obtained, sentencing
It is disconnected whether to have set up the neural network model corresponding with the active user;Collecting unit, for judge to have set up with
In the case of the corresponding neural network model of the active user, user's row of the active user in collection preset time period
For data, wherein, the user behavior data is that the operation performed by having delivered data is obtained according to the active user
Data;Training unit, for the user behavior data using the active user in the preset time period to the institute that has set up
State neural network model to be trained, so that the neural network model remembers the data interested to the active user
Recall.
Further, the collecting unit includes:Acquisition module, for when preset interval time collection is described default
Between in section the active user user behavior data;The training unit includes:Training module, for every between described preset
Every the time using the user behavior data of the active user in the preset time period to the neutral net mould set up
Type is trained.
In embodiments of the present invention, the dispensing data of target web are obtained first;Then, using the nerve net pre-established
Network model is predicted to delivering data, to determine recommending data in data are delivered according to predicting the outcome;Finally, number will be recommended
According to recommending user, wherein, recommending data is delivers the data that interest-degree in data meets default interest-degree, and interest-degree is current
User is to the interest-degree of dispensing data, and the neutral net pre-established is in advance using user's row of different time sections active user
Obtained neutral net is trained for data.Above-mentioned neural network model is a dynamic model, and the model can be to difference
The user behavior data of period is trained, and is trained by more and more behavioral datas, ensure that the model can be increasingly
Accurately, then the data of recommendation are also just more and more accurate, and then alleviate existing data recommendation scheme and recommend precision poor
Technical problem, it is achieved thereby that being accurately the technique effect of user's recommending data.
Brief description of the drawings
, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical scheme of the prior art
The accompanying drawing used required in embodiment or description of the prior art is briefly described, it should be apparent that, in describing below
Accompanying drawing is some embodiments of the present invention, for those of ordinary skill in the art, before creative work is not paid
Put, other accompanying drawings can also be obtained according to these accompanying drawings.
Fig. 1 is a kind of flow chart of the recommendation method of cuisines data according to embodiments of the present invention;
Fig. 2 is the flow chart of according to embodiments of the present invention the first alternatively recommendation method of cuisines data;
Fig. 3 is the flow chart of the alternatively recommendation method of cuisines data of according to embodiments of the present invention second;
Fig. 4 is the flow chart of according to embodiments of the present invention the third the alternatively recommendation method of cuisines data;
Fig. 5 is the flow chart of the alternatively recommendation method of cuisines data of according to embodiments of the present invention the 4th kind;
Fig. 6 is a kind of schematic diagram of the recommendation apparatus of cuisines data according to embodiments of the present invention.
Embodiment
Technical scheme is clearly and completely described below in conjunction with accompanying drawing, it is clear that described implementation
Example is a part of embodiment of the invention, rather than whole embodiments.Based on the embodiment in the present invention, ordinary skill
The every other embodiment that personnel are obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
In the description of the invention, it is necessary to explanation, term " " center ", " on ", " under ", "left", "right", " vertical ",
The orientation or position relationship of the instruction such as " level ", " interior ", " outer " be based on orientation shown in the drawings or position relationship, merely to
Be easy to the description present invention and simplify description, rather than indicate or imply signified device or element must have specific orientation,
With specific azimuth configuration and operation, therefore it is not considered as limiting the invention.In addition, term " first ", " second ",
" the 3rd " is only used for describing purpose, and it is not intended that indicating or implying relative importance.
In the description of the invention, it is necessary to illustrate, unless otherwise clearly defined and limited, term " installation ", " phase
Even ", " connection " should be interpreted broadly, for example, it may be being fixedly connected or being detachably connected, or be integrally connected;Can
To be mechanical connection or electrical connection;Can be joined directly together, can also be indirectly connected to by intermediary, Ke Yishi
The connection of two element internals.For the ordinary skill in the art, with concrete condition above-mentioned term can be understood at this
Concrete meaning in invention.
Embodiment one:
According to embodiments of the present invention there is provided a kind of embodiment of the recommendations method of cuisines data, it is necessary to illustrate,
The step of flow of accompanying drawing is illustrated can perform in the computer system of such as one group computer executable instructions, also,
, in some cases, can be shown to be performed different from order herein although showing logical order in flow charts
The step of going out or describe.
Fig. 1 is a kind of flow chart of the recommendation method of cuisines data according to embodiments of the present invention, as shown in figure 1, the party
Method comprises the following steps:
Step S102, obtains the dispensing data of target web;
In embodiments of the present invention, target web is the cuisines webpage for recommending cuisines data for user, and user can be with
Browse various cuisines in the web page, and every kind of cuisines way.
Above-mentioned dispensing data can be the cuisines data delivered in the cuisines webpage.For example, " garlic stems fry elbow flower " is done
Method is shared with gains in depth of comprehension.
Step S104, is predicted using the neutral net pre-established to delivering data, to be determined in data are delivered
Recommending data, wherein, recommending data is delivers the data that interest-degree in data meets default interest-degree, and interest-degree is active user
Interest-degree to delivering data, the neutral net pre-established is in advance using the user behavior number of different time sections active user
The neutral net obtained according to being trained;
In embodiments of the present invention, the above-mentioned neutral net pre-established is in advance using different time sections active user's
User behavior data is trained obtained neutral net.Can be more accurate by the user behavior data of different time sections
The hobby of user is determined, thinks that user recommends more accurate data.
For example, section at any time, user likes the cuisines browsed to be vegetarian diet, then now, it is possible to pass through the time
The navigation patterns data of user recommend related vegetarian diet for the user in section.
If in next period, user likes the cuisines browsed to be meat.If now also continuing to recommend for user
Vegetarian diet, will influence Consumer's Experience, now just can determine that the user likes browsing by the user behavior data of the period
Food be meat, then can now think that the user recommends carnivorous, to meet the different demands of user not in the same time.
It should be noted that above-mentioned user behavior data can be the click volume of user, the kind for the cuisines that user is clicked on
Class, user clicks on the quantity of identical cuisines, and user clicks on the cuisines that the time user of cuisines is collected, the cuisines of user institute thumb up
Etc..
Step S106, recommending data is recommended to active user.
In embodiments of the present invention, the dispensing data of target web are obtained first;Then, using the nerve net pre-established
Network model is predicted to delivering data, to determine recommending data in data are delivered according to predicting the outcome;Finally, number will be recommended
According to recommending user, wherein, recommending data is delivers the data that interest-degree in data meets default interest-degree, and interest-degree is current
User is to the interest-degree of dispensing data, and the neutral net pre-established is in advance using user's row of different time sections active user
Obtained neutral net is trained for data.Above-mentioned neural network model is a dynamic model, and the model can be to difference
The user behavior data of period is trained, and is trained by more and more behavioral datas, ensure that the model can be increasingly
Accurately, then the data of recommendation are also just more and more accurate, and then alleviate existing data recommendation scheme and recommend precision poor
Technical problem, it is achieved thereby that being accurately the technique effect of user's recommending data.
In an optional embodiment of the embodiment of the present invention, as shown in Fig. 2 obtaining the dispensing data of target web
Before, this method also comprises the following steps:
Step S201, judges whether to have set up the neural network model corresponding with active user;
Step S202, if it is judged that having set up the neural network model corresponding with active user, then when gathering default
Between in section active user user behavior data, wherein, user behavior data is is held according to active user to having delivered data
The data that capable operation is obtained;
Step S203, using the user behavior data of active user in preset time period to the neural network model set up
It is trained, so that neural network model is remembered to the data interested to active user;
In embodiments of the present invention, it is that the user each registered establishes a corresponding neural network model in advance.Cause
This, before the recommendation of cuisines data is carried out for the user, it is first determined whether having built on the corresponding nerve net of active user
Network model.
If it is judged that not pre-establishing neural network model for the user, then step S204 is performed, be the user
Neural network model is set up, and gathers user behavior data of the user in target web in real time, is collected in real time with basis
User behavior data is trained to the neural network model, and next group is delivered according to the neural network model after training
Data are predicted, to select recommending data in next dispensing data for the user.
If it is judged that setting up neural network model for the user, then user's row of the user in preset time period is gathered
For data.For example, the user behavior data of the user within one day is gathered, then, using user's row of the user between one day
Neural network model is trained for data, so that the neural network model (that is, is used the hobby that browses of current time user
Data interested to family) remembered.To browse hobby remember after, it is possible to using training after nerve net
Network model is predicted to the dispensing data at current time, to determine recommending data in the dispensing data.
It should be noted that if delivering data and being cuisines data, then in embodiments of the present invention, neutral net
The input of model can be the content style of cuisines data, the cuisines food materials of cuisines data, cuisines taste, cuisines way, cuisines
Picture etc..
Explanation is needed further exist for, in embodiments of the present invention, could be arranged to first in target web enter in user
During row registration, by above-mentioned steps S201, whether judgement has once set up the neural network model corresponding with active user.Remove
Outside this, it may be arranged as, when user logs in the target web every time, judging once whether set up and active user
Corresponding neural network model.
By foregoing description, in embodiments of the present invention, the training of neural network model is dynamic, passes through dynamic
Adjustment can be interested to exact knowledge current time user cuisines.
Therefore, in another optional embodiment of the embodiment of the present invention, in step S202 collection preset time periods
The user behavior data of active user includes:The user behavior of active user in preset time period is gathered every preset interval time
Data.
In embodiments of the present invention, dynamically neural network model is trained in order to realize, can set every pre-
If the user behavior data of active user in period collection preset time, wherein it is possible to be set to gather default week about
The user behavior data of active user in time.
For example, being gathered on May 7th, 2017 in May 4 to 6 days Mays in 2017 in 2017, the user behavior number of user
According to.Then, gathered on May 14th, 2017 in May 12 to 14 days Mays in 2017 in 2017, the user behavior data of user.
Step S203 is using the user behavior data of the active user in preset time period to the neutral net mould set up
Type be trained including:Every preset interval time using the user behavior data of active user in preset time period to having set up
Neural network model be trained.
Then, collecting in May 4 to 6 days Mays in 2017 in 2017, after the user behavior data of user, using
The user behavior data is trained to neural network model so that neural network model according to the user behavior data to user
Interested data are remembered.And, collecting in May 12 to 14 days Mays in 2017 in 2017, the user of user
After behavioral data, neural network model can also be trained using the user behavior data, so that neural network model
The data interested to user are remembered according to the user behavior data.
In another optional embodiment, as shown in figure 3, delivering data includes a plurality of dispensing data, step S104 is adopted
It is predicted with the neutral net pre-established to delivering data, to determine that recommending data includes following step in data are delivered
Suddenly:
Step S301, is predicted using the neutral net pre-established to delivering data, to obtain every dispensing data
The first probability, wherein, the first probability represents active user to the current interest-degree for delivering data;
Step S302, will deliver interest-degree in data and is used as more than or equal to the dispensing data of default interest-degree and recommend number
According to.
In embodiments of the present invention, after dispensing data are got, it is possible to using the neural network model trained
A plurality of dispensing data are predicted.
Wherein, if dispensing data are cuisines data, then can be beautiful by the cuisines content style included in cuisines data
Eat food materials, cuisines taste, cuisines way and cuisines picture and will be liked probability and do not liked as the input of the neutral net
Probability as the neural network model output.
For example, input can be the content style of cuisines data 1, food materials, taste, way and picture, output can be to work as
Preceding user is to the interest-degree (that is, the first probability, that is, the probability liked) of current cuisines data 1, and it can be current use to export this
Dislike degree (that is, the probability that does not like) of the family to current cuisines data 1.
Determine every dispensing data like probability and the probability that does not like after, it is possible to according to like probability and
The determine the probability recommending data not liked.For example, probability will be liked to be more than 0.6 cuisines data as recommending data, can be with
Cuisines data of the probability more than 0.7 will be liked as recommending data, specifically, user can enter to the threshold value according to actual needs
Row adjustment.
In embodiments of the present invention, it is being predicted, is being recommended to delivering data using the neutral net pre-established
After data, it is possible to recommend the recommending data to user.But, before recommending data is recommended to active user, such as Fig. 4 institutes
Show, this method also includes:
Step S401, obtains the subscriber data of active user, wherein, subscriber data includes the native place of active user, currently
The sex of user, the name of active user;
Step S402, is screened according to subscriber data to recommending data, the recommending data after being screened;
Step S403, active user is pushed to by the recommending data after screening.
After recommending data is obtained by above-mentioned steps S104, recommending data can also further be screened.
Specifically, the subscriber data of user can be combined, for example, the native place of user, sex, name is screened to recommending data.
For example, user is mother of baby, then the user can input the information, i.e. identity can be set in registration
It is set to mother of baby.So now, after prediction obtains recommending data, it is possible to reference to the identity, in recommending data
Recommend to meet the cuisines data of its identity for the user.
In another example, user is fitness, then the user can input the information, i.e. occupation can be set in registration
It is set to fitness.So now, after prediction obtains recommending data, it is possible to reference to the occupation, be in recommending data
The user recommends to meet its professional cuisines data.
It should be noted that in addition to recommending data is recommended into active user, can also be by the recommendation after screening
Data-pushing to active user, wherein, when recommending to user, can be marked by different marks.For example, adopting
The recommending data after screening is marked with label symbol is paid close attention to, and uses the label symbol commonly paid close attention to screening
Recommending data before is marked.By the mark mode, enable to user accurate and be quickly found out the U.S. oneself admired
Eclipse number evidence, without searching the data admired in numerous cuisines data.
In another optional embodiment, after recommending data is recommended to active user, as shown in figure 5, this method
Also comprise the following steps:
Step S501, stores according to the recommendation time to recommending data according to preset format, is stored in database, its
In, database is used to store history recommending data;
Step S502, obtain user query statement, and according to query statement in database query history recommending data.
In embodiments of the present invention, can also be according to preset format pair after the recommending data is recommended to active user
Recommending data is stored, for example, can be stored every recommending data according to recommendation time and release time.In storage
When, the cuisines content style of the recommending data, food materials, taste, the information such as way and picture can be stored.
User's cuisines data interested before can also inquiring about at any time, user can send inquiry to server and refer to
Order, for example, the data recommended during inquiry May 1 to 8 days Mays in 2017 in 2017, and in the data recommended by with
The data that family is browsed.Now, server can just inquire about corresponding recommending data according to the query statement in database, and will
The recommending data is shown to user and checked.
By the set-up mode, enable to what user became apparent to know oneself in the food interested to different time sections
Which thing has, to help user to remember the food interested to oneself.
Embodiment two:
The embodiment of the present invention additionally provides a kind of recommendation apparatus of cuisines data, and the recommendation apparatus of the cuisines data is mainly used
In the recommendation method for performing the cuisines data that the above of the embodiment of the present invention is provided, below to provided in an embodiment of the present invention
The recommendation apparatus of cuisines data does specific introduction.
Fig. 6 is a kind of schematic diagram of the recommendation apparatus of cuisines data according to embodiments of the present invention, as shown in fig. 6, the U.S.
The recommendation apparatus of eclipse number evidence mainly includes:First acquisition unit 61, predicting unit 62 and recommendation unit 63, wherein:
First acquisition unit 61, the dispensing data for obtaining target web;
Predicting unit 62, for being predicted using the neutral net pre-established to delivering data, to deliver data
Middle determination recommending data, wherein, recommending data is delivers the data that interest-degree in data meets default interest-degree, and interest-degree is to work as
Preceding user is to the interest-degree of dispensing data, and the neutral net pre-established is in advance using the user of different time sections active user
Behavioral data is trained obtained neutral net;
Recommendation unit 63, for recommending recommending data to active user.
In embodiments of the present invention, the dispensing data of target web are obtained first;Then, using the nerve net pre-established
Network model is predicted to delivering data, to determine recommending data in data are delivered according to predicting the outcome;Finally, number will be recommended
According to recommending user, wherein, recommending data is delivers the data that interest-degree in data meets default interest-degree, and interest-degree is current
User is to the interest-degree of dispensing data, and the neutral net pre-established is in advance using user's row of different time sections active user
Obtained neutral net is trained for data.Above-mentioned neural network model is a dynamic model, and the model can be to difference
The user behavior data of period is trained, and is trained by more and more behavioral datas, ensure that the model can be increasingly
Accurately, then the data of recommendation are also just more and more accurate, and then alleviate existing data recommendation scheme and recommend precision poor
Technical problem, it is achieved thereby that being accurately the technique effect of user's recommending data.
Alternatively, delivering data includes a plurality of dispensing data, and predicting unit includes:Prediction module, builds in advance for using
Vertical neutral net is predicted to delivering data, to obtain the first probability of every dispensing data, wherein, the first probability is represented
Active user is to the current interest-degree for delivering data;Determining module, is more than or equal in advance for that will deliver interest-degree in data
If the dispensing data of interest-degree are used as recommending data.
Alternatively, the device also includes:Judging unit, for before the dispensing data of target web are obtained, judgement to be
It is no to have set up the neural network model corresponding with active user;Collecting unit, for judging to have set up and active user
In the case of corresponding neural network model, the user behavior data of active user in collection preset time period, wherein, user
Behavioral data is the data obtained according to active user to the operation performed by having delivered data;Training unit, for using pre-
If the user behavior data of active user is trained to the neural network model set up in the period, so that neutral net mould
Type is remembered to the data interested to active user.
Alternatively, collecting unit includes:Acquisition module, for every current in preset interval time collection preset time period
The user behavior data of user;Training unit includes:Training module, for being used every preset interval time in preset time period
The user behavior data of active user is trained to the neural network model set up.
Alternatively, the device also includes:Second acquisition unit, for before recommending data is recommended to active user, obtaining
The subscriber data of active user is taken, wherein, subscriber data includes the native place of active user, the sex of active user, active user
Name;Recommending data is screened according to subscriber data, the recommending data after being screened;Recommendation unit includes:Push away
Module is recommended, for the recommending data after screening to be pushed into active user.
Alternatively, the device also includes:Memory cell, for after recommending data is recommended to active user, according to pre-
If form is stored to recommending data according to the recommendation time, it is stored in database, wherein, database is pushed away for storing history
Recommend data;3rd acquiring unit, the query statement for obtaining user, and query history is pushed away in database according to query statement
Recommend data.
Finally it should be noted that:Various embodiments above is merely illustrative of the technical solution of the present invention, rather than its limitations;To the greatest extent
The present invention is described in detail with reference to foregoing embodiments for pipe, it will be understood by those within the art that:Its according to
The technical scheme described in foregoing embodiments can so be modified, or which part or all technical characteristic are entered
Row equivalent substitution;And these modifications or replacement, the essence of appropriate technical solution is departed from various embodiments of the present invention technology
The scope of scheme.
Claims (10)
1. a kind of recommendation method of cuisines data, it is characterised in that including:
Obtain the dispensing data of target web;
The dispensing data are predicted using the neutral net pre-established, to determine to recommend number in the dispensing data
According to, wherein, the recommending data meets the data of default interest-degree for interest-degree in the dispensing data, and the interest-degree is to work as
Preceding user to the interest-degrees of the dispensing data, the neutral net pre-established in advance using described in different time sections when
The user behavior data of preceding user is trained obtained neutral net;
Recommend the recommending data to the active user.
2. recommendation method according to claim 1, it is characterised in that the dispensing data include a plurality of dispensing data, adopt
The dispensing data are predicted with the neutral net pre-established, to determine recommending data bag in the dispensing data
Include:
The dispensing data are predicted using the neutral net pre-established, described data are delivered to obtain every
First probability, wherein, first probability represents the active user to the current interest-degree for delivering data;
It regard the dispensing data that interest-degree in the dispensing data is more than or equal to default interest-degree as the recommending data.
3. recommendation method according to claim 1, it is characterised in that before the dispensing data of target web are obtained, institute
Stating method also includes:
Judge whether to have set up the neural network model corresponding with the active user;
If it is judged that having set up the neural network model corresponding with the active user, then gather described in preset time period
The user behavior data of active user, wherein, the user behavior data for according to the active user to having delivered data institute
The data that the operation of execution is obtained;
Using the user behavior data of the active user in the preset time period to the neural network model set up
It is trained, so that the neural network model is remembered to the data interested to the active user.
4. recommendation method according to claim 3, it is characterised in that
The user behavior data of the active user includes in collection preset time period:Gather described pre- every preset interval time
If the user behavior data of the active user in the period;
Using the user behavior data of the active user in the preset time period to the neutral net mould set up
Type be trained including:Every user behavior of the preset interval time using the active user in the preset time period
Data are trained to the neural network model set up.
5. recommendation method according to claim 3, it is characterised in that
Before the recommending data is recommended to the active user, methods described also includes:Obtain the use of the active user
Family data, wherein, the subscriber data includes the native place of active user, the sex of active user, the name of active user;According to
The subscriber data is screened to the recommending data, the recommending data after being screened;
The recommending data is recommended to include to the active user:The recommending data after screening is pushed to described current
User.
6. recommendation method according to claim 1, it is characterised in that recommending the recommending data to the active user
Afterwards, methods described also includes:
The recommending data is stored according to the recommendation time according to preset format, is stored in database, wherein, the number
It is used to store history recommending data according to storehouse;
The query statement of user is obtained, and according to query statement query history recommending data in the database.
7. a kind of recommendation apparatus of cuisines data, it is characterised in that including:
First acquisition unit, the dispensing data for obtaining target web;
Predicting unit, for being predicted using the neutral net pre-established to the dispensing data, with the dispensing number
According to middle determination recommending data, wherein, the recommending data meets the data of default interest-degree for interest-degree in the dispensing data,
The interest-degree is interest-degree of the active user to the dispensing data, and the neutral net pre-established is in advance using not
User behavior data with active user described in the period is trained obtained neutral net;
Recommendation unit, for recommending the recommending data to the active user.
8. recommendation apparatus according to claim 7, it is characterised in that the dispensing data include a plurality of dispensing data, institute
Stating predicting unit includes:
Prediction module, the neutral net for being pre-established described in is predicted to the dispensing data, to obtain every
First probability for delivering data, wherein, first probability represents the active user to the current interest for delivering data
Degree;
Determining module, the dispensing data for interest-degree in the dispensing data to be more than or equal to default interest-degree are used as institute
State recommending data.
9. recommendation apparatus according to claim 7, it is characterised in that described device also includes:
Judging unit, for before the dispensing data of target web are obtained, judging whether to have set up and active user's phase
Corresponding neural network model;
Collecting unit, in the case where judging to have set up the neural network model corresponding with the active user, adopting
Collect the user behavior data of the active user in preset time period, wherein, the user behavior data is according to described current
User is to having delivered the data that the operation performed by data is obtained;
Training unit, for the user behavior data using the active user in the preset time period to described in having set up
Neural network model is trained, so that the neural network model is remembered to the data interested to the active user
Recall.
10. recommendation apparatus according to claim 9, it is characterised in that
The collecting unit includes:Acquisition module, for gathering described in the preset time period work as every preset interval time
The user behavior data of preceding user;
The training unit includes:Training module, for using institute in the preset time period every the preset interval time
The user behavior data for stating active user is trained to the neural network model set up.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710484775.4A CN107203644A (en) | 2017-06-23 | 2017-06-23 | The recommendation method and apparatus of cuisines data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710484775.4A CN107203644A (en) | 2017-06-23 | 2017-06-23 | The recommendation method and apparatus of cuisines data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107203644A true CN107203644A (en) | 2017-09-26 |
Family
ID=59908233
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710484775.4A Pending CN107203644A (en) | 2017-06-23 | 2017-06-23 | The recommendation method and apparatus of cuisines data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107203644A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109587530A (en) * | 2018-11-22 | 2019-04-05 | 广州虎牙信息科技有限公司 | A kind of data processing method, device, terminal device and storage medium |
CN110968768A (en) * | 2018-09-28 | 2020-04-07 | 北京易数科技有限公司 | Information generation method and device |
CN113491432A (en) * | 2020-04-07 | 2021-10-12 | 添可智能科技有限公司 | Automatic cooking method and system of cooking machine and cooking machine |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105844508A (en) * | 2016-03-22 | 2016-08-10 | 天津中科智能识别产业技术研究院有限公司 | Dynamic periodic neural network-based commodity recommendation method |
CN106327240A (en) * | 2016-08-11 | 2017-01-11 | 中国船舶重工集团公司第七0九研究所 | Recommendation method and recommendation system based on GRU neural network |
EP3173983A1 (en) * | 2015-11-26 | 2017-05-31 | Siemens Aktiengesellschaft | A method and apparatus for providing automatically recommendations concerning an industrial system |
-
2017
- 2017-06-23 CN CN201710484775.4A patent/CN107203644A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3173983A1 (en) * | 2015-11-26 | 2017-05-31 | Siemens Aktiengesellschaft | A method and apparatus for providing automatically recommendations concerning an industrial system |
CN105844508A (en) * | 2016-03-22 | 2016-08-10 | 天津中科智能识别产业技术研究院有限公司 | Dynamic periodic neural network-based commodity recommendation method |
CN106327240A (en) * | 2016-08-11 | 2017-01-11 | 中国船舶重工集团公司第七0九研究所 | Recommendation method and recommendation system based on GRU neural network |
Non-Patent Citations (1)
Title |
---|
宋文官 等: "《"电子商务网站建设与维护实训》", 31 May 2008, 高等教育出版社 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110968768A (en) * | 2018-09-28 | 2020-04-07 | 北京易数科技有限公司 | Information generation method and device |
CN110968768B (en) * | 2018-09-28 | 2023-11-24 | 北京易数科技有限公司 | Information generation method and device |
CN109587530A (en) * | 2018-11-22 | 2019-04-05 | 广州虎牙信息科技有限公司 | A kind of data processing method, device, terminal device and storage medium |
CN109587530B (en) * | 2018-11-22 | 2021-06-08 | 广州虎牙信息科技有限公司 | Data processing method and device, terminal equipment and storage medium |
CN113491432A (en) * | 2020-04-07 | 2021-10-12 | 添可智能科技有限公司 | Automatic cooking method and system of cooking machine and cooking machine |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104602042B (en) | Label setting method based on user behavior | |
KR101770683B1 (en) | Method, apparatus, server, program and computer-readable recording medium of dispalying social network information flow | |
CN106303955B (en) | For carrying out matched method and apparatus to hotspot and POI | |
JP6262764B2 (en) | Method and system for pushing mobile applications | |
US9607273B2 (en) | Optimal time to post for maximum social engagement | |
US10025807B2 (en) | Dynamic data acquisition method and system | |
CN103136253A (en) | Method and device of acquiring information | |
CA2828380A1 (en) | Computer system, database and uses thereof | |
CN105868291A (en) | Website address recommendation method, apparatus and system | |
CN103399861B (en) | A kind of network address in Web side navigation recommends methods, devices and systems | |
CN107203644A (en) | The recommendation method and apparatus of cuisines data | |
CN102682046A (en) | Member searching and analyzing method in social network and searching system | |
CN103780625B (en) | User interest finds method and apparatus | |
WO2012174174A2 (en) | System and method for user preference augmentation through social network inner-circle knowledge discovery | |
US9449111B2 (en) | System and method for generating and accessing trails | |
CN102542055A (en) | Website directory display method and system | |
CN101493818A (en) | Network information searching method based on human relation network | |
CN110020152B (en) | Application recommendation method and device | |
CN105898425A (en) | Video recommendation method and system and server | |
CN110929058B (en) | Trademark picture retrieval method and device, storage medium and electronic device | |
US20170331909A1 (en) | System and method of monitoring and tracking online source content and/or determining content influencers | |
JP2006053616A (en) | Server device, web site recommendation method and program | |
CN106937173A (en) | Video broadcasting method and device | |
CN103593455B (en) | File recommendation method and file recommendation device | |
CN110188277A (en) | A kind of recommended method and device of resource |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170926 |
|
RJ01 | Rejection of invention patent application after publication |