CN109636481A - User's portrait construction method and device towards domestic consumer - Google Patents
User's portrait construction method and device towards domestic consumer Download PDFInfo
- Publication number
- CN109636481A CN109636481A CN201811568282.XA CN201811568282A CN109636481A CN 109636481 A CN109636481 A CN 109636481A CN 201811568282 A CN201811568282 A CN 201811568282A CN 109636481 A CN109636481 A CN 109636481A
- Authority
- CN
- China
- Prior art keywords
- domestic consumer
- label
- user
- kinsfolk
- portrait
- 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
- 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
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of user's portrait construction method and device towards domestic consumer, is related to user's portrait building field.User's portrait construction method towards domestic consumer, it include: the multimedia data for acquiring domestic consumer in preset time period, according to the multimedia data, kinsfolk's structure of the domestic consumer is predicted, drawn a portrait according to the user that kinsfolk's structure constructs the domestic consumer.It realizes and deepization, fining analysis is carried out to domestic consumer, probe into household internal member composition, and then platform can also provide more fine service according to kinsfolk's structure for family, improve home user experience degree.
Description
Technical field
The present invention relates to user's portrait building fields, in particular to a kind of user's portrait structure towards domestic consumer
Construction method and device.
Background technique
Under " internet+" background, internet television industry is rapidly developed.It is quick-fried with internet television user
Hairdo increases, and the demand for experience of user is more diversified.The experience of internet television user is improved by building user portrait
Degree is the development upsurge of current internet television industry.
Existing user's portrait construction method is based primarily upon the static attribute data and dynamic behaviour data of user, and building is used
The method of family portrait, this building user portrait usually regards user as an entirety and black box, i.e., using equipment as description object
To construct user's portrait.
Since existing user portrait does not distinguish practical viewing object, so that platform can not provide fining for user
Service, user experience be not high.
Summary of the invention
It is an object of the present invention in view of the deficiency of the prior art, provide a kind of user towards domestic consumer
Draw a portrait construction method and device, regard user as entirety in the method to solve conventional construction user portrait so that platform without
The problem of method provides fining service for user.
To achieve the above object, technical solution used in the embodiment of the present invention is as follows:
In a first aspect, the embodiment of the invention provides a kind of methods for constructing user's portrait towards domestic consumer, comprising:
Acquire the multimedia data of domestic consumer in preset time period；
According to the multimedia data, kinsfolk's structure of the domestic consumer is predicted；
User's portrait of the domestic consumer is constructed according to kinsfolk's structure.
Optionally, it according to the multimedia data, before the kinsfolk's structure for predicting the domestic consumer, also wraps
It includes:
According to the multimedia programming of preset kind, tape label sample is constructed；
Extract the characteristic information in the tape label sample with mark meaning；
The characteristic information is weighted by time decay factor, and does normalized, obtains the tape label
Sample.
Optionally, the characteristic information is weighted by time decay factor, and does normalized, obtained described
After tape label sample, further includes:
The tape label sample is divided are as follows: training set and test set；
The training set predicts kinsfolk's structure of the domestic consumer based on preset algorithm；
The test set assesses the performance of domestic consumer kinsfolk's structure.
Optionally, the characteristic information includes following one or more:
In the preset time period user be switched on number, watch multimedia programming total duration, watch programme televised live
Total duration, watch the preset kind the multimedia programming duration, watch the multimedia section of the preset kind
Purpose number.
Optionally, further includes: according to kinsfolk's structure of the domestic consumer, predict the family of the domestic consumer at
The Annual distribution of member.
Optionally, further includes: the various dimensions tag library towards the domestic consumer is constructed according to the domestic consumer.
Optionally, the various dimensions tag library includes: level-one label and second level label；
The level-one label includes: domestic consumer's label, kinsfolk's label；
The second level label includes: family structure label, Annual distribution label, user's essential attribute label, interest preference
Label, viewing period preference label, home audience diversity label, user's active state label and user are worth label.
Second aspect, the embodiment of the invention also provides a kind of devices that user's portrait is constructed towards domestic consumer, comprising:
Acquisition module, for acquiring the multimedia data of domestic consumer in preset time period；
Prediction module, for predicting kinsfolk's structure of the domestic consumer according to the multimedia data；
Module is constructed, the user for constructing the domestic consumer according to kinsfolk's structure draws a portrait.
Optionally, further includes:
Module is constructed, the multimedia programming according to preset kind is also used to, constructs tape label sample；
Extraction module, for extracting the characteristic information in the tape label sample with mark meaning；
Processing module for being weighted to the characteristic information by time decay factor, and is done normalized, is obtained
To the tape label sample；
Optionally, the tape label sample is divided are as follows: training set and test set；Further include:
Prediction module is also used to kinsfolk's structure that the training set predicts the domestic consumer based on preset algorithm；
Evaluation module assesses the performance of kinsfolk's structure of the domestic consumer for the test set.
The beneficial effects of the present invention are:
User's portrait construction method provided by the invention towards domestic consumer, passes through more matchmakers in acquisition preset time period
Body played data, predicts kinsfolk's structure according to collected multimedia data, constructs house according to kinsfolk's structure
The user of front yard user draws a portrait, and deepization, fining analysis can be carried out to domestic consumer, probes into household internal member composition, into
And platform can also provide more fine service according to kinsfolk's structure for family, improve home user experience degree.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 is the flow diagram that user's portrait method is constructed towards domestic consumer that one embodiment of the application provides；
Fig. 2 is the flow diagram that kinsfolk's structure is predicted towards domestic consumer that one embodiment of the application provides；
Fig. 3 is the process signal that kinsfolk's Annual distribution is predicted towards domestic consumer that one embodiment of the application provides
Figure；
Fig. 4 is the structural schematic diagram that user's portrait device is constructed towards domestic consumer that one embodiment of the application provides；
Fig. 5 is the structural schematic diagram that user's portrait device is constructed towards domestic consumer that another embodiment of the application provides.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.
Fig. 1 is a kind of flow diagram that user's portrait method is constructed towards domestic consumer provided by the present application.The present invention
The realization of embodiment is construction device of being drawn a portrait based on user, is realized in the form of software functional units.The device can integrate
In equipment such as terminal, servers.The terminals such as general smart television, intelligent players can play some more matchmakers by internet
Body resource, these resources can be provided by the server on backstage, which can be understood as multi-media network platform, simple below
Claim " platform ".
As shown in Figure 1, provided in this embodiment include: towards domestic consumer's building user's portrait method
S110: the multimedia data of domestic consumer in acquisition preset time period.
Above-mentioned preset time period can be one week, one month etc., during this period of time to more matchmakers of different domestic consumers
Body played data is acquired.Wherein, different domestic consumers can be distinguished by user identifier, such as the account for passing through user
Number information is distinguished, herein with no restrictions.
Collect multimedia broadcasting receipt can by platform from be sent to terminal request acquisition and also from terminal oneself to
Platform reports, and can specifically use Flume (result collection system) and Sqoop (SQL-to-Hadoop, abbreviation Sqoop) technology,
The data for collecting multiple data sources, such as Relational DBMS (My Structured Query Language, abbreviation
) and journal file Mysql.
Wherein, Flume is result collection system, is that a High Availabitity, highly reliable, distributed massive logs are adopted
Collection, polymerization and Transmission system.Flume supports to customize Various types of data sender in result collection system, for collecting data.Together
When, Flume, which is provided, carries out simple process to data, and writes the ability of various data receivings (customizable).
Sqoop is one and is used to distributed system (Hadoop Distributed File System, abbreviation
Hadoop the tool that the data) and in Relational DBMS mutually shift, can will be in a relevant database
Data lead enter in Hadoop distributed file system (Hadoop Distributed File System, referred to as
HDFS in), the data of HDFS can also be led and is entered in relevant database.HDFS can carry out the mass data being collected into
Distributed storage.
The multimedia data being collected into may include: that (such as place provinces and cities, open device model for the attribute information of user
Family time, the client used), user watched behavioural information and video content information etc..
Wherein, period, each period that the viewing behavior of user can be collected into user's viewing video watch anything
The data such as the viewing duration of the program of type and every kind of program.Video content information can be collected into what relevant user was watched
The data such as video content, director, performer, show time, area, video length and video subject.
S120: according to multimedia data, kinsfolk's structure of domestic consumer is predicted.
Kinsfolk's structure is used to describe the composition of domestic consumer member, for example, this family include which specific type at
Member, such as: in addition old man, young man, children can also distinguish male, women etc..
Specific type member is by taking old man, young man, children as an example.The step predominantly judges whether contain in certain domestic consumer
There are the process of old man, young man and children.
It should be noted that kinsfolk's structure of prediction domestic consumer, for judging whether contain one in domestic consumer
A little specific type members, such as: old man, young man, children, male, women etc., with reach to domestic consumer carry out deepization,
Fining analysis, probes into the purpose of household internal Behavior law.In the present embodiment, judge during family structure mainly with
For whether having old man's type in analysis domestic consumer, whether there is old man's type in detailed analysis domestic consumer.Other families at
Member's type such as judges whether to have young man's type/children's type procedure in domestic consumer and judges whether domestic consumer has old man
The process of type is similar, will not add to repeat in following embodiment.
S130: it is drawn a portrait according to the user that kinsfolk's structure constructs domestic consumer.
After judging kinsfolk's structure type of domestic consumer, in this step, family is constructed according to kinsfolk's structure
The user of front yard user draws a portrait.
For example, predicting kinsfolk's structure of a certain domestic consumer by step S120 are as follows: the elderly and young man, and
The class for being collected into the attribute information of the domestic consumer by step S110, liking viewing multimedia period and multimedia programming
The data such as type can construct the more complete user's portrait of the domestic consumer in conjunction with kinsfolk's structure.
User's portrait construction method provided in this embodiment towards domestic consumer, by more in acquisition preset time period
Media play data, further according to kinsfolk's structure of multimedia data prediction domestic consumer, finally according to kinsfolk
Structure constructs user's portrait of domestic consumer.The user's portrait constructed in this way can probe into the composition of household internal member, can be right
Domestic consumer carries out deepization, fining analysis.Platform can also be provided according to kinsfolk's structure for family more fine in turn
Service, improve home user experience degree.
As shown in Fig. 2, Fig. 2 is the process for predicting kinsfolk's structure towards domestic consumer that one embodiment of the application provides
Schematic diagram.
According to multimedia data, before the kinsfolk's structure for predicting domestic consumer, further includes:
S210: according to the multimedia programming of preset kind, tape label sample is constructed.
In the present embodiment, some multimedia programmings with mark meaning are preset, and formulate respective rule, for identifying house
Whether there is special member in the user of front yard.For example, by taking old man is special member as an example.
I.e. in the present embodiment, it is believed that as long as having viewed certain preset kind programs, include preset kind in kinsfolk
The corresponding special member of program.
Whether for example, to judge in domestic consumer containing old man's type, the present embodiment according to picking out in advance
Preset kind program is judged, and formulates respective rule, and marking to have in old man and family in becoming a Buddhist monk or nun does not have the case where old man.The rule
Then it is used to identify that most probable has the domestic consumer of old man and most unlikely has the domestic consumer of old man.
When specific implementation, preset kind program can be divided into two kinds of situations: (1) most identifying the program of meaning；(2) compare
Relatively there is the program of mark meaning.If watching the program for most identifying meaning, then it is assumed that there is special member in the user family.If
The program for comparing and having mark meaning was not watched, then it is assumed that does not have special member in the user family.
In the present embodiment, for judging whether to have old man in domestic consumer, definition most identifies the program packet of meaning
Containing following program: " sunset ", " health 365 ", " nine divisions of China in remote antiquity play garden " etc..If had in the multimedia data being collected into such
The broadcasting of program records, then can be determined that there is old man in the user family.
It includes following program: " showing sword ", " Avenue of Stars ", " Lao Liang tells stories " that the program for having mark meaning is compared in definition
Deng.If do not recorded with the broadcasting of such program in the multimedia data being collected into, determine there is no old man in family.
It should be noted that it is provided in this embodiment with mark meaning program and compare with mark meaning program
It is not limitation with the embodiment of above-mentioned offer, specifically with the program of actual coincidence the elderly, young man or children's judgment criteria
Subject to.In the present embodiment, the program of meaning will be most identified by viewing and compare the program with mark meaning and judged
Kinsfolk's structure be known as tape label sample.
S220: the characteristic information in tape label sample with mark meaning is extracted.
Tape label sample is constructed by above-mentioned rule, and the family structure member type for including in the tape label sample is
Through determination, further, the characteristic information for having determined family structure member type is extracted.
Judge domestic consumer with and without the feature in this type of old man with mark meaning for example, extraction is above-mentioned
Information.
Further, characteristic information includes following one or more:
In preset time period user be switched on number, watch multimedia programming total duration, watch programme televised live it is total when
The duration of multimedia programming that is long, watching preset kind, watch preset kind multimedia programming number.
There is the following to need to illustrate it should be noted that carrying out feature extraction in the present embodiment:
It may include said one or multiple characteristic informations under every kind of feature classification.Above-mentioned preset time period can not be counted
Program total duration is watched in interior program request, because it is watched the total duration of multimedia programming with feature and watches the total duration of programme televised live
Between there are linear relationships.During whether statistics watched some program, if the duration for watching some program reaches specified
Long, just the program was watched in determination, to reduce the influence accidentally broadcast.
S230: characteristic information is weighted by time decay factor, and does normalized, obtains tape label sample
This.
Firstly, the characteristic information and domestic consumer to extraction construct two-dimensional matrix, characteristic value is calculated, each characteristic value is passed through
Time decay factor is weighted, the formula of time decay factor are as follows:
In formula, t is the size for watching the time gap current time of this event generation of some program.
Due to the linear module and disunity of the sample of extracted characteristic information, for example, viewing time length is with minute
Or hour is unit statistics, and the booting number of user is illustrated tens times or is for several hundred times what unit was counted.Therefore, right
Above extracted characteristic information is normalized, and extracted characteristic information is fallen between (0,1), with probability mould
Type comes out.
Each characteristic value is normalized based on max-min standardized method, formula is as follows:
In formula, x indicates current sample data, and min is the maximum value of sample data, and max is the minimum value of sample data.
Further, characteristic information is weighted by time decay factor, and does normalized, obtain the band
After marker samples, further includes:
Tape label sample obtained above is divided are as follows: training set and test set.Wherein partitioning standards can will obtain
80 the percent of tape label sample is used as training set, and 20 percent is used as test set, the tape label sample that can also will be obtained
This 70 percent is used as training set, and 30 percent is used as test set, and specific partitioning standards are according to actual algorithm demand
Depending on, herein with no restrictions.
Wherein, kinsfolk structure of the training set based on preset algorithm prediction domestic consumer.
Above-mentioned prediction algorithm predicts the kinsfolk of domestic consumer by logic-based regression algorithm in the present embodiment
Structure.Whether contain certain types of kinsfolk in the algorithm prediction domestic consumer that logic-based returns, passes through logistic regression
Algorithm, will in prediction domestic consumer the problem of whether having certain types of kinsfolk's structure be converted into probabilistic model.
It is assumed that in the above-mentioned training sample of building, D={ xi,yi, i=1,2 ..., N, xi∈ RD,yi∈ { 0,1 },
In, x indicates feature set, and y indicates 0 or 1, and i indicates i-th of training sample in training set.Formula is as follows:
Further, above formula is converted are as follows:
In formula, w indicates that the weighted value of feature, b indicate adjustment parameter.Wherein, there is old man to be denoted as positive class in domestic consumer, do not have
There is old man to be denoted as negative class.P (y=1 | x) is a possibility that x is positive class, and P (y=0 | x) it is a possibility that x is negative class.Logistic regression
W and b is estimated by maximum-likelihood method.
On the basis of the above embodiments, using logic-based regression algorithm predict kinsfolk's structure, by family at
It whether include the corresponding special member of preset kind program in kinsfolk's structure of member's structure to judge domestic consumer.By family
Whether having old man, young man, the label converting of children in user is three two-value classification problems.But the present embodiment is not to use
Logistic regression algorithm predict kinsfolk's structure be limitation, correspondingly, can also apply other algorithms, as NB Algorithm,
Nearest neighbor algorithm, algorithm of support vector machine etc..
Further, test set is used to assess the performance of kinsfolk's structure of domestic consumer.
It is assessed using Accuracy (accuracy rate), Precision (accurate rate), Recall (recall rate) and F-measure
Disaggregated model.The present embodiment is used to assess logic-based regression algorithm and predicts in domestic consumer whether contain this type of old man
Accuracy.Each Performance Evaluation index can be calculated by the confusion matrix of table 1.
Table 1
In table 1, TP: actual sample is positive class, and prediction result is positive class；FP: sample is negative class, and prediction result is negative class；
TN: sample is negative class, and prediction result is negative class；FN: sample is positive class, and prediction result is negative class.
Wherein, what Accuracy (accuracy rate) was calculated is for given test data set, the correct classification samples of classifier
Number accounts for the ratio between total number of samples.Its calculation formula is:
What Precision (accurate rate) was calculated is in all projects being retrieved, it should shared by the project being retrieved
Ratio.Its calculation formula is:
What Recall (recall rate) was calculated is the ratio that the project that be retrieved is accounted in all projects retrieved.Its
Calculation formula are as follows:
F-measure assesses the harmonic-mean that disaggregated model is Precision (accurate rate) and Recall (recall rate),
Its calculation formula is:
Further, after predicting kinsfolk's structure, kinsfolk can also be predicted according to kinsfolk's structure
Annual distribution can determine which section time is ownership old man/young man/children in one day of one family according to Annual distribution,
Can be by the time by day granularity refinement to period granularity, building can help to put down towards the more detailed user's portrait of domestic consumer
Platform carries out more accurate marketing and recommendation, different recommendation results can be generated not having to the period, to meet different type man
The hobby of front yard member.
Optionally, according to kinsfolk's structure of domestic consumer, the Annual distribution of the kinsfolk of domestic consumer is predicted.
Specifically, referring to figure 3., Fig. 3 is provided in this embodiment a kind of towards domestic consumer's prediction kinsfolk's time
The flow diagram of distribution.
S310: the score of domestic consumer different type kinsfolk within a certain period of time is calculated.
Such as: calculate separately old man's score, young man's score and children's score.
The present embodiment is for calculating old man's score: calculating domestic consumer in old man's score of some period, is listed in
Whether the viewing detail in the period has characteristic coefficient obtained in old man's model as weight for what logic-based returned,
With in the period corresponding characteristic value (duration for watching preset kind program) weighted sum.Young man's score and children's score
And so on.
S320: analyzing the score of different type kinsfolk, obtains analysis result.
Score by the elderly, young man, children in some period is analyzed, and analysis result is as follows: if compared
In young man's score and children's score, old man's score is very high, then the period of the domestic consumer is allocated to old man；If three
A point spread is little, then the period of the user is divided into " undetermined (uncertain) ", i.e., can not judged
Which kinsfolk's preference period.
S330: the Annual distribution feature of different type kinsfolk is determined based on the analysis results.
If in previous step according to analysis meter calculate every afternoon 5:00-7:00 old man score be higher by young man and youngster
The period is then divided into old man by child.Platform will push TV related with old man between 5:00-7:00 every afternoon
Program or advertisement.
Optionally, on the basis of the above embodiments further include: the multidimensional towards domestic consumer is constructed according to domestic consumer
Spend tag library.
In the method provided in this embodiment for constructing user's portrait towards domestic consumer, it can also construct towards domestic consumer's
Various dimensions tag library more fully constructs user's portrait of description domestic consumer with various dimensions label.
Further, various dimensions tag library includes: level-one label and second level label.
Level-one label includes: domestic consumer's label, kinsfolk's label.Wherein, domestic consumer's label is basic with family
Unit is described, kinsfolk's label is basic description unit with the special member in family.
Second level label specifically includes that family structure label, Annual distribution label, user's essential attribute label, interest preference
Label, viewing period preference label, home audience diversity label, user's active state label and user are worth label.
Family structure label: the family structure of the domestic consumer is described.Three-level label under it has included whether always
Whether people has young man and whether has children three.Whether the label can make the clear domestic consumer's of platform to include specific family
Front yard member, to accurately be recommended and be marketed according to specific kinsfolk.
Three-level label under user's essential attribute label mainly includes place provinces and cities, device model, the time of opening an account, uses
Client.
Viewing period preference label describes the period that the domestic consumer often watches program as a whole.The mark
Label can infer to obtain by counting the viewing duration of each period.
Home audience diversity label describes the categorical measure that domestic consumer watches program.
Three-level label under user's active state label includes new user, any active ues, the user that is sunk into sleep, is lost user.
It can judge that user's enlivens shape by analyzing viewing duration, nearly one month viewing duration and the last viewing time daily
State.
Three-level label under the value label of user includes high-value user, middle value user and low value user.It can
According to the same day booting frequency, same day viewing number of programs, same day rating duration, of that month whether once consumption pay content, of that month use
Family spending limit, the judgement of nearly spending limit in March.
Annual distribution label is basic description unit with kinsfolk, describes the period of certain kinsfolk's preference.
Interest preference label is basic description unit with kinsfolk, describes old man/young man/children frequency respectively
Road preference, program category preference, director's preference, performer's preference, show time preference, regional preference, program duration preference and section
Mesh subject matter preferences.
Wherein, Annual distribution label and interest preference label belong to kinsfolk's label, special in domestic consumer for describing
The Behavior law for certain period of determining member in one day.
It should be noted that the calculation of different weights is taken in the setting in user tag library.Label weight is mainly declined
Subtracting coefficient and behavioural information weights influence.
It should be noted that the weight of label can indicate the temperature of label, weight is bigger, and temperature is higher, and the time declines
Subtracting coefficient embodies the process that the temperature of label is gradually cooled down with the time.
Such as: the label of interest preference is arranged, the calculation and time decay factor and viewing mode of weight
It is related.Compared to the watching behavior before for a long time, nearest behavior should give biggish weight.It is watched compared to live streaming, Searching point
Greater weight should be given by broadcasting viewing mode.
Determine that the family of domestic consumer ties by predicting the family structure feature of domestic consumer, and according to family structure feature
Structure label and the essential attribute information for being collected into domestic consumer determine essential attribute label, are often watched according to domestic consumer
The period of multimedia programming determines that period preference label and the multi-medium data being collected into determine home audience diversity mark
Label, root pass through the daily viewing multimedia duration of analysis domestic consumer, nearly one month viewing duration and the last viewing time
Judge that the active state label of user, be switched on the frequency according to the same day, the same day watches number of programs, and same day rating duration, whether is this month
Once pay content, of that month customer consumption amount are consumed, the information such as nearly spending limit in March judge that the user of domestic consumer is worth mark
Label the Annual distribution label of domestic consumer's different types of structure can be obtained according to kinsfolk's structural analysis, according to different home
Channel preferences, program category preference, director's preference, performer's preference, show time preference, the regional preference, program of subtype member
The interest preference label of duration preference and program subject matter preference building domestic consumer's different types of structure.
By constructing above-mentioned label, user's portrait towards domestic consumer can be established out.It is provided in this embodiment towards family
The user of front yard user draws a portrait, and analyzes kinsfolk's structure of domestic consumer, and according to kinsfolk's structure prediction family
Time granularity is sub-divided into the period by the Annual distribution of user, in conjunction with the interest of different time sections different type kinsfolk
Preference can carry out deepization, fining analysis to domestic consumer, platform domestic consumer can be made to provide finer service,
Improve the Experience Degree of domestic consumer.
Fig. 4 is the structural schematic diagram that user's portrait device is constructed towards domestic consumer that one embodiment of the application provides.
The device that user's portrait is constructed towards domestic consumer that the present embodiment also provides, comprising: acquisition module 410, prediction
Module 420 and building module 430, in which: acquisition module 410, the multimedia for acquiring domestic consumer in preset time period are broadcast
Put data.Prediction module 420, for predicting kinsfolk's structure of domestic consumer according to multimedia data.Construct module
430, the user for constructing domestic consumer according to kinsfolk's structure draws a portrait.
Further, module 430 is constructed, the multimedia programming according to preset kind is also used to, constructs tape label sample.It should
Device further include: extraction module 440 and processing module 450.Wherein, extraction module 440 have for extracting in tape label sample
Identify the characteristic information of meaning.Processing module 450 for being weighted to characteristic information by time decay factor, and is done and is returned
One change processing, obtains tape label sample.
Further, tape label sample is divided are as follows: training set and test set.Prediction module 420 is also used to training set base
Kinsfolk's structure of the domestic consumer is predicted in preset algorithm.
The device further include: evaluation module 460, for assessing the property of domestic consumer kinsfolk's structure according to training set
Energy.
Further, prediction module 420 are also used to kinsfolk's structure according to domestic consumer, predict domestic consumer's
The Annual distribution of kinsfolk.
Further, module 430 is constructed, is also used to construct the various dimensions label towards domestic consumer according to domestic consumer
Library.
Further, various dimensions tag library includes: level-one label and second level label；
Level-one label includes: domestic consumer's label, kinsfolk's label；
Second level label include: family structure label, Annual distribution label, user's essential attribute label, interest preference label,
It watches period preference label, home audience diversity label, user's active state label and user and is worth label.
The method that above-mentioned apparatus is used to execute previous embodiment offer, it is similar that the realization principle and technical effect are similar, herein not
It repeats again.
The above module can be arranged to implement one or more integrated circuits of above method, such as: one
Or multiple specific integrated circuits (Application Specific Integrated Circuit, abbreviation ASIC), or, one
Or multi-microprocessor (digital singnal processor, abbreviation DSP), or, one or more field programmable gate
Array (Field Programmable Gate Array, abbreviation FPGA) etc..For another example, when some above module passes through processing elements
When the form of part scheduler program code is realized, which can be general processor, such as central processing unit (Central
Processing Unit, abbreviation CPU) or it is other can be with the processor of caller code.For another example, these modules can integrate
Together, it is realized in the form of system on chip (system-on-a-chip, abbreviation SOC).
Fig. 5 is the structural schematic diagram that user's portrait device is constructed towards domestic consumer that another embodiment of the application provides,
The device can integrate the chip in terminal device or terminal device, which can be the calculating for having image processing function
Equipment.
The device includes: processor 510, memory 520.
Memory 520 is for storing program, the program that processor 510 calls memory 520 to store, to execute the above method
Embodiment.Specific implementation is similar with technical effect, and which is not described herein again.
It should be noted that calling Hive (Tool for Data Warehouse), Hbase when processor handles corresponding data
Data in (distributed memory system), Spark (computing engines) establish the data warehouse based on Hive, realize to magnanimity number
According to various dimensions, varigrained inquiry and analysis, lay the foundation for user's portrait.For excavating the life of class user portrait label
At using the machine learning library of Spark, i.e. Spark MLlib.(MLlib is the distribution constructed on Spark to MLlib at present
Machine learning library) it supported classification, returned, cluster, collaborative filtering scheduling algorithm.The big wide table of the label of all user's portraits, that is, use
All kinds of labels in family, are stored in HBase, to realize batch real-time query.
Optionally, the present invention also provides a kind of program product, such as computer readable storage medium, including program, the journeys
Sequence is when being executed by processor for executing above method embodiment.
In several embodiments provided by the present invention, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or logical of device or unit
Letter connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server or the network equipment etc.) or processor (English: processor) execute this hair
The part steps of bright each embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory
(English: Read-Only Memory, abbreviation: ROM), random access memory (English: Random Access Memory, letter
Claim: RAM), the various media that can store program code such as magnetic or disk.
Claims (10)
- The construction method 1. a kind of user towards domestic consumer draws a portrait characterized by comprisingAcquire the multimedia data of domestic consumer in preset time period；According to the multimedia data, kinsfolk's structure of the domestic consumer is predicted；User's portrait of the domestic consumer is constructed according to kinsfolk's structure.
- 2. the method according to claim 1, wherein predicting the family according to the multimedia data Before kinsfolk's structure of user, further includes:According to the multimedia programming of preset kind, tape label sample is constructed；Extract the characteristic information in the tape label sample with mark meaning；The characteristic information is weighted by time decay factor, and does normalized, obtains the tape label sample.
- 3. according to the method described in claim 2, it is characterized in that, being added to the characteristic information by time decay factor Power, and does normalized, after obtaining the tape label sample, further includes:The tape label sample is divided are as follows: training set and test set；The training set predicts kinsfolk's structure of the domestic consumer based on preset algorithm；The test set assesses the performance of domestic consumer kinsfolk's structure.
- 4. according to the method described in claim 2, it is characterized in that, the characteristic information includes following one or more:In the preset time period user be switched on number, watch multimedia programming total duration, watch programme televised live it is total when The duration of the multimedia programming that is long, watching the preset kind, the multimedia programming for watching the preset kind Number.
- 5. the method according to claim 1, wherein further include:According to kinsfolk's structure of the domestic consumer, the Annual distribution of the kinsfolk of the domestic consumer is predicted.
- 6. the method according to claim 1, wherein further include:The various dimensions tag library towards the domestic consumer is constructed according to the domestic consumer.
- 7. according to the method described in claim 6, it is characterized in that, the various dimensions tag library includes: level-one label and second level Label；The level-one label includes: domestic consumer's label, kinsfolk's label；The second level label include: family structure label, Annual distribution label, user's essential attribute label, interest preference label, It watches period preference label, home audience diversity label, user's active state label and user and is worth label.
- The construction device 8. a kind of user towards domestic consumer draws a portrait characterized by comprisingAcquisition module, for acquiring the multimedia data of domestic consumer in preset time period；Prediction module, for predicting kinsfolk's structure of the domestic consumer according to the multimedia data；Module is constructed, the user for constructing the domestic consumer according to kinsfolk's structure draws a portrait.
- 9. device according to claim 8, which is characterized in that further include:Module is constructed, the multimedia programming according to preset kind is also used to, constructs tape label sample；Extraction module, for extracting the characteristic information in the tape label sample with mark meaning；Processing module for being weighted to the characteristic information by time decay factor, and does normalized, obtains institute State tape label sample.
- 10. device according to claim 9, which is characterized in that divide the tape label sample are as follows: training set and test Collection；Further include:Prediction module is also used to kinsfolk's structure that the training set predicts the domestic consumer based on preset algorithm；Evaluation module assesses the performance of domestic consumer kinsfolk's structure for the test set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811568282.XA CN109636481A (en) | 2018-12-19 | 2018-12-19 | User's portrait construction method and device towards domestic consumer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811568282.XA CN109636481A (en) | 2018-12-19 | 2018-12-19 | User's portrait construction method and device towards domestic consumer |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109636481A true CN109636481A (en) | 2019-04-16 |
Family
ID=66076136
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811568282.XA Pending CN109636481A (en) | 2018-12-19 | 2018-12-19 | User's portrait construction method and device towards domestic consumer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109636481A (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110442761A (en) * | 2019-06-21 | 2019-11-12 | 深圳中琛源科技股份有限公司 | A kind of user draws a portrait construction method, electronic equipment and storage medium |
| CN111475718A (en) * | 2020-03-30 | 2020-07-31 | 清华大学 | User activity model construction method and system based on preference diversity |
| CN111556369A (en) * | 2020-05-21 | 2020-08-18 | 四川省有线广播电视网络股份有限公司 | Television-based family classification method |
| CN111586348A (en) * | 2020-04-15 | 2020-08-25 | 福建星网视易信息系统有限公司 | Video background image acquisition method, storage medium, video matting method and storage device |
| CN111601166A (en) * | 2020-05-21 | 2020-08-28 | 广州欢网科技有限责任公司 | Method, device, storage medium and server for determining family member composition |
| CN111601169A (en) * | 2020-05-21 | 2020-08-28 | 广州欢网科技有限责任公司 | Method, device, storage medium and server for determining age stage of family member |
| CN111601171A (en) * | 2020-05-21 | 2020-08-28 | 广州欢网科技有限责任公司 | Method, device and controller for positioning family members of smart television |
| CN111626774A (en) * | 2020-05-21 | 2020-09-04 | 广州欢网科技有限责任公司 | Advertisement delivery system, method and readable storage medium |
| CN111861550A (en) * | 2020-07-08 | 2020-10-30 | 上海视九信息科技有限公司 | OTT (over the Top) equipment-based family portrait construction method and system |
| CN112235644A (en) * | 2020-09-29 | 2021-01-15 | 深圳市鼎盛光电有限公司 | Program information recommendation method and device, digital television and storage medium |
| CN112506063A (en) * | 2020-11-25 | 2021-03-16 | 中移（杭州）信息技术有限公司 | Data analysis method, system, electronic device and storage medium |
| CN112559586A (en) * | 2020-12-03 | 2021-03-26 | 中国联合网络通信集团有限公司 | Method and system for acquiring home user characteristic data |
| CN113553105A (en) * | 2020-04-23 | 2021-10-26 | 百度在线网络技术（北京）有限公司 | Method and device for generating guide page |
| CN114095786A (en) * | 2021-11-17 | 2022-02-25 | 四川长虹电器股份有限公司 | Smart television user family member identification method based on community discovery algorithm |
| CN114268838A (en) * | 2021-12-15 | 2022-04-01 | 深圳市酷开网络科技股份有限公司 | Method and device for processing family member portrait based on OTT user portrait |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103440242A (en) * | 2013-06-26 | 2013-12-11 | 北京亿赞普网络技术有限公司 | User search behavior-based personalized recommendation method and system |
| CN105045859A (en) * | 2015-07-10 | 2015-11-11 | 海信集团有限公司 | User feature analysis method and apparatus for intelligent device |
| CN105430504A (en) * | 2015-11-27 | 2016-03-23 | 中国科学院深圳先进技术研究院 | Family member mix identification method and system based on television watching log mining |
| CN105516810A (en) * | 2015-12-04 | 2016-04-20 | 山东大学 | Television user family member analysis method based on LDA (Latent Dirichlet Allocation) model |
| CN106910136A (en) * | 2017-02-23 | 2017-06-30 | 北京小米移动软件有限公司 | It is method and device, the system of family's portrait |
| CN107124653A (en) * | 2017-05-16 | 2017-09-01 | 四川长虹电器股份有限公司 | The construction method of TV user portrait |
| CN107360465A (en) * | 2017-08-22 | 2017-11-17 | 四川长虹电器股份有限公司 | A kind of method that Intelligent television terminal is drawn a portrait using vocal print generation user |
-
2018
- 2018-12-19 CN CN201811568282.XA patent/CN109636481A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103440242A (en) * | 2013-06-26 | 2013-12-11 | 北京亿赞普网络技术有限公司 | User search behavior-based personalized recommendation method and system |
| CN105045859A (en) * | 2015-07-10 | 2015-11-11 | 海信集团有限公司 | User feature analysis method and apparatus for intelligent device |
| CN105430504A (en) * | 2015-11-27 | 2016-03-23 | 中国科学院深圳先进技术研究院 | Family member mix identification method and system based on television watching log mining |
| CN105516810A (en) * | 2015-12-04 | 2016-04-20 | 山东大学 | Television user family member analysis method based on LDA (Latent Dirichlet Allocation) model |
| CN106910136A (en) * | 2017-02-23 | 2017-06-30 | 北京小米移动软件有限公司 | It is method and device, the system of family's portrait |
| CN107124653A (en) * | 2017-05-16 | 2017-09-01 | 四川长虹电器股份有限公司 | The construction method of TV user portrait |
| CN107360465A (en) * | 2017-08-22 | 2017-11-17 | 四川长虹电器股份有限公司 | A kind of method that Intelligent television terminal is drawn a portrait using vocal print generation user |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110442761A (en) * | 2019-06-21 | 2019-11-12 | 深圳中琛源科技股份有限公司 | A kind of user draws a portrait construction method, electronic equipment and storage medium |
| CN111475718A (en) * | 2020-03-30 | 2020-07-31 | 清华大学 | User activity model construction method and system based on preference diversity |
| CN111586348A (en) * | 2020-04-15 | 2020-08-25 | 福建星网视易信息系统有限公司 | Video background image acquisition method, storage medium, video matting method and storage device |
| CN113553105A (en) * | 2020-04-23 | 2021-10-26 | 百度在线网络技术（北京）有限公司 | Method and device for generating guide page |
| CN111601166A (en) * | 2020-05-21 | 2020-08-28 | 广州欢网科技有限责任公司 | Method, device, storage medium and server for determining family member composition |
| CN111601171A (en) * | 2020-05-21 | 2020-08-28 | 广州欢网科技有限责任公司 | Method, device and controller for positioning family members of smart television |
| CN111626774A (en) * | 2020-05-21 | 2020-09-04 | 广州欢网科技有限责任公司 | Advertisement delivery system, method and readable storage medium |
| CN111601169A (en) * | 2020-05-21 | 2020-08-28 | 广州欢网科技有限责任公司 | Method, device, storage medium and server for determining age stage of family member |
| CN111556369A (en) * | 2020-05-21 | 2020-08-18 | 四川省有线广播电视网络股份有限公司 | Television-based family classification method |
| CN111861550B (en) * | 2020-07-08 | 2023-09-08 | 上海视九信息科技有限公司 | Family portrait construction method and system based on OTT equipment |
| CN111861550A (en) * | 2020-07-08 | 2020-10-30 | 上海视九信息科技有限公司 | OTT (over the Top) equipment-based family portrait construction method and system |
| CN112235644A (en) * | 2020-09-29 | 2021-01-15 | 深圳市鼎盛光电有限公司 | Program information recommendation method and device, digital television and storage medium |
| CN112506063A (en) * | 2020-11-25 | 2021-03-16 | 中移（杭州）信息技术有限公司 | Data analysis method, system, electronic device and storage medium |
| CN112506063B (en) * | 2020-11-25 | 2024-05-07 | 中移（杭州）信息技术有限公司 | Data analysis method, system, electronic device and storage medium |
| CN112559586A (en) * | 2020-12-03 | 2021-03-26 | 中国联合网络通信集团有限公司 | Method and system for acquiring home user characteristic data |
| CN112559586B (en) * | 2020-12-03 | 2023-06-09 | 中国联合网络通信集团有限公司 | Method and system for acquiring home user characteristic data |
| CN114095786A (en) * | 2021-11-17 | 2022-02-25 | 四川长虹电器股份有限公司 | Smart television user family member identification method based on community discovery algorithm |
| CN114268838A (en) * | 2021-12-15 | 2022-04-01 | 深圳市酷开网络科技股份有限公司 | Method and device for processing family member portrait based on OTT user portrait |
| CN114268838B (en) * | 2021-12-15 | 2023-12-26 | 深圳市酷开网络科技股份有限公司 | Family member portrait processing method and device based on OTT user portrait |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109636481A (en) | User's portrait construction method and device towards domestic consumer | |
| CN104935970B (en) | Carry out the method and Television clients of television content recommendation | |
| Yin et al. | LCARS: A spatial item recommender system | |
| CN106028071A (en) | Video recommendation method and system | |
| Košir et al. | Database for contextual personalization | |
| CN103686237B (en) | Recommend the method and system of video resource | |
| Chen et al. | On information coverage for location category based point-of-interest recommendation | |
| CN103634687B (en) | The method and system of video search result are provided in intelligent television | |
| CN110430471A (en) | It is a kind of based on the television recommendations method and system instantaneously calculated | |
| CN109189951A (en) | A kind of multimedia resource recommended method, equipment and storage medium | |
| CN105282565A (en) | Video recommendation method and device | |
| CN105281925B (en) | The method and apparatus that network service groups of users divides | |
| CN110543598A (en) | information recommendation method and device and terminal | |
| CN109511015A (en) | Multimedia resource recommended method, device, storage medium and equipment | |
| KR20130119246A (en) | Apparatus and method for recommending contents based sensibility | |
| JP2008542870A (en) | Method and apparatus for estimating the overall interest of a group of users for content | |
| US20120124089A1 (en) | User interest pattern modeling server and method for modeling user interest pattern | |
| CN111191133B (en) | Service search processing method, device and equipment | |
| CN106951471A (en) | A kind of construction method of the label prediction of the development trend model based on SVM | |
| CN107846611A (en) | A kind of TV programme method for pushing and system based on age bracket | |
| CN103997662A (en) | Program pushing method and system | |
| CN105843876A (en) | Multimedia resource quality assessment method and apparatus | |
| CN111177559A (en) | Text travel service recommendation method and device, electronic equipment and storage medium | |
| CN114004654A (en) | Internet card user portrait construction method and system based on big data analysis | |
| CN109635192A (en) | Magnanimity information temperature seniority among brothers and sisters update method and platform towards micro services |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190416 |