CN106548375B - Method and apparatus for constructing product portrait - Google Patents
Method and apparatus for constructing product portrait Download PDFInfo
- Publication number
- CN106548375B CN106548375B CN201610964911.5A CN201610964911A CN106548375B CN 106548375 B CN106548375 B CN 106548375B CN 201610964911 A CN201610964911 A CN 201610964911A CN 106548375 B CN106548375 B CN 106548375B
- Authority
- CN
- China
- Prior art keywords
- characteristic
- user
- weight
- dimension
- consumption user
- 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.)
- Active
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
- G06Q30/0203—Market surveys; Market polls
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)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present disclosure discloses a kind of method and apparatus for constructing product portrait.The described method includes: the characteristic of the consumption user of target product is classified according to multiple dimensions；The weight of each dimension is determined according to the characteristic of the consumption user；According to the weight of the characteristic of the consumption user and identified each dimension, the scoring of each consumption user in the consumption user is determined；The feature user in the consumption user is determined according to the scoring；According to the characteristic of the feature user, the product portrait of the target product is determined.In this way, constructing more accurate product portrait on the basis of comprehensively considering the weight of each dimension of user characteristic data, being conducive to the accurate positionin to consumption user and the improvement of product, increase product benefit.
Description
Technical field
This disclosure relates to computer field, and in particular, to a kind of method and apparatus for constructing product portrait.
Background technique
Product portrait is the positioning of a kind of pair of product, may include portrait and the group of product user of product self attributes
Portrait.It is drawn a portrait by product, the characteristic of the user group of product in product promotion, can be analyzed, excavate potential customers
Group carries out targeted product improvement, reaches the demand of the high speed development steady in a long-term of enterprise.
The currently used method for constructing product portrait includes simple filtration method and artificial point system.
In simple filtration method, if as soon as user has a condition not meet, it is filtered very much.For example, in recruitment industry,
Usually user is filtered by inputting the attribute of user.Such as company's recruitment con dition are as follows: profession is software, and educational background is master, work
Make experience greater than 5 years, and the keyword of resume includes big data.The user group's range filtered out in this way is small, Hen Duoyong
Only wherein a rule does not meet (for example, educational background is undergraduate course) at family.For this resume, pass through simple filtration method just quilt completely
It filters out.
In the method manually to score, manually given a mark to each dimension of user, for example, region, the age, keyword,
Each dimension such as interest is given a mark.This method is given a mark with artificial experience, and the product portrait of building is often inaccurate.
Summary of the invention
Purpose of this disclosure is to provide a kind of simple and easy method and apparatus for constructing product portrait.
To achieve the goals above, the disclosure provides a kind of method for constructing product portrait.The described method includes: will
The characteristic of the consumption user of target product is classified according to multiple dimensions；Characteristic according to the consumption user is true
The weight of fixed each dimension；According to the weight of the characteristic of the consumption user and identified each dimension, determine each
The scoring of consumption user；The feature user in the consumption user is determined according to the scoring；According to the spy of the feature user
Data are levied, determine the product portrait of the target product.
Optionally, the step of characteristic according to the consumption user determines the weight of each dimension includes: point
Do not calculate include in its characteristic characteristic corresponding with each dimension consumption user number and the consumption user
Total number ratio, obtain the data frequency of corresponding dimension；It calculates separately in total number and its characteristic of whole user
The logarithm of the ratio of the number of whole user including characteristic corresponding with each dimension obtains the reverse frequency of corresponding dimension
Rate, wherein the entirety user includes the consumption user and non-consumption user；Respectively by the data frequency of each dimension multiplied by
Reverse frequency obtains the weight of corresponding dimension.
Optionally, the step of characteristic according to the consumption user determines the weight of each dimension includes: week
Interval weight of each dimension in current period is determined to phase property according to the characteristic of the consumption user；According to each dimension
The history interval weight and current interval weight of degree, determine the weight of each dimension.
Optionally, described according to the characteristic of the consumption user and the weight of identified each dimension, it determines every
The step of scoring of a consumption user includes: the weight that each characteristic is determined according to the characteristic of the consumption user；
According to the weight of the weight of each characteristic and identified each dimension, the scoring of each consumption user is determined.
Optionally, the step of characteristic according to the consumption user determines the weight of each characteristic packet
It includes: calculating the ratio of the number of the consumption user in its characteristic including fisrt feature data and the total number of the consumption user
Value, obtains the data frequency of the fisrt feature data；It includes described for calculating in the total number and its characteristic of whole user
The logarithm of the ratio of the number of the whole user of fisrt feature data obtains the reverse frequency of the fisrt feature data, wherein
The entirety user includes the consumption user and non-consumption user；By the data frequency of the fisrt feature data multiplied by described
The reverse frequency of fisrt feature data obtains the weight of the fisrt feature data.
Optionally, the step of the characteristic of the consumption user by target product is classified according to multiple dimensions
Before, the method also includes: determine similar with target product like product；By the consumption user of the like product
It is determined as the consumption user of the target product.
The disclosure also provides a kind of for constructing the device of product portrait.Described device includes: categorization module, is used for mesh
The characteristic for marking the consumption user of product is classified according to multiple dimensions；Weight determination module, for according to the consumption
The characteristic of user determines the weight of each dimension；Score determining module, for the characteristic according to the consumption user
With the weight of identified each dimension, the scoring of each consumption user is determined；Feature user's determining module, for according to
Scoring determines the feature user in the consumption user；Product portrait determining module, for the feature according to the feature user
Data determine the product portrait of the target product.
Through the above technical solutions, the weight of each dimension is determined based on the characteristic of consumption user itself, so that
The weight of dimension more accurately embodies the feature of consumption user.Therefore, the disclosure is comprehensively considering each of user characteristic data
On the basis of the weight of a dimension, more accurate product portrait is constructed, the accurate positionin and production to consumption user are conducive to
The improvement of product increases product benefit.
Other feature and advantage of the disclosure will the following detailed description will be given in the detailed implementation section.
Detailed description of the invention
Attached drawing is and to constitute part of specification for providing further understanding of the disclosure, with following tool
Body embodiment is used to explain the disclosure together, but does not constitute the limitation to the disclosure.In the accompanying drawings:
Fig. 1 is the flow chart for the method for constructing product portrait that an exemplary embodiment provides；
Fig. 2 is the flow chart of the weight for each dimension of determination that an exemplary embodiment provides；
Fig. 3 is the flow chart of the weight for each dimension of determination that another exemplary embodiment provides；
Fig. 4 is the flow chart for the method for constructing product portrait that another exemplary embodiment provides；
Fig. 5 is the block diagram for the device for constructing product portrait that an exemplary embodiment provides.
Specific embodiment
It is described in detail below in conjunction with specific embodiment of the attached drawing to the disclosure.It should be understood that this place is retouched
The specific embodiment stated is only used for describing and explaining the disclosure, is not limited to the disclosure.
As described above, when using simple filter method, though in the characteristic of user only wherein one do not meet
Preset rules are also filtered completely.But in fact, do not meet for wherein one, but other aspects are particularly pertinent
Resume, many companies are also to think favourably of.In view of the above problems, inventor expects, when constructing product portrait, Ke Yicong
The characteristic of consumption user itself is set out, to determine the weight of each dimension that characteristic is divided into, to make according to power
Determining feature user is more acurrate again, finally makes product portrait more acurrate.
Fig. 1 is the flow chart for the method for constructing product portrait that an exemplary embodiment provides.As shown in Figure 1, institute
The method of stating may comprise steps of.
In step s 11, the characteristic of the consumption user of target product is classified according to multiple dimensions.
Wherein, consumption user can be the user that (for example, having bought target product) is consumed to target product.
The characteristic of consumption user may include the characteristic of multiple dimensions (or type).For example, may include for describing people
Mouthful attribute, behavior, hobby, business etc.: the age, occupation, the dimensions such as region, collects, thumbs up at gender.
In step s 12, the weight of each dimension is determined according to the characteristic of consumption user.
In the characteristic of a consumption user, not necessarily have in each dimension.The feature of certain dimension
Data occur more in consumption user, show that the dimension is more important for the description of feature user, the weight of the dimension is just
It is higher.
Simply, can calculate separately in its characteristic include characteristic corresponding with each dimension consumption user
Number and consumption user total number ratio, obtain the weight of corresponding dimension.For example, the total number of consumption user is 100,
The number in its characteristic including the consumption user of male or female's (characteristic corresponding with gender dimension) is 20, then gender is tieed up
The weight of degree is 0.2 (20/100).
In step s 13, according to the weight of the characteristic of consumption user and identified each dimension, determination each disappears
The scoring at expense family.
Simply, each characteristic included by each dimension can be assigned to scheduled value, the scheduled value is for example
It can be set, can also be set according to experiment or experience according to the tendentiousness of itself wish by user.It is used to a consumption
When family is scored, the value of characteristic of the consumption user in each dimension can be weighted (place dimension
Weight) summation, obtain the scoring of the consumption user.What the scoring embodied the feature of consumption user and target product is associated with journey
Degree.
In step S14, according to the feature user determined in consumption user that scores.
As described above, consumption user can be the user consumed to target product.In consumption user, it can wrap
Include feature user and other users.Wherein, feature user can be the user of its feature Yu target product highlights correlations, or
It says, is the easy consumer groups of target product.And the other users except feature user can be with the target product degree of association less
High user is in other words the accidental consumer group of target product.The group is only because accidental cause has consumed target production
Therefore product when carrying out product portrait, can be excluded.It is, filtering out from consumption user and easily disappearing in the disclosure
Take crowd, the analysis of characteristic is then carried out by the easy consumer groups to the target product, obtains the product of target product
Portrait.
For example, the scoring the high, the easier consumption target product of the consumption user is indicated, scoring can be higher than predetermined
Scoring threshold value consumption user determination be characterized user.
In step S15, according to the characteristic of feature user, the product portrait of target product is determined.
After determining feature user, the characteristic of feature user can be analyzed using a variety of methods.Simply, may be used
Using the method for statistics accounting.It is, the consumption in characteristic including a certain characteristic is used in each dimension
The accounting in amount mesh and its characteristic including the number of the consumption user of any characteristic in the dimension is more than predetermined threshold
Value or more than other characteristics accounting, so that it may a part that this feature data are drawn a portrait as product.
For example, having 80 consumption users includes the data characteristics of age dimension in all 100 consumption users.Wherein,
Consumption user in its characteristic including 20-30 years old data characteristics has 50 people, includes 30-40 age in characteristic according to spy
The consumption user of sign has 20 people, includes that the consumption user of 40-50 years old data characteristics has 10 people in characteristic.Then its characteristic
It include any data feature in age dimension in consumption user and its characteristic in including 20-30 years old data characteristics
The accounting (50/80) of the number of consumption user is greater than 30-40 years old corresponding accounting (20/80), and corresponding greater than 40-50 years old
Accounting (10/80), the then a part that 20-30 years old data characteristics can be drawn a portrait as product.
Through the above technical solutions, the weight of each dimension is determined based on the characteristic of consumption user itself, so that
The weight of dimension more accurately embodies the feature of consumption user.Therefore, the disclosure is comprehensively considering each of user characteristic data
On the basis of the weight of a dimension, more accurate product portrait is constructed, the accurate positionin and production to consumption user are conducive to
The improvement of product increases product benefit.
The reverse document-frequency (Term Frequency-Inverse Document Frequency, TF-IDF) of word frequency-is calculated
Method is a kind of weighting technique that can be used for information retrieval and data mining.In one embodiment of the disclosure, the calculation can be applied
The weight of method conceived to determine each dimension.
Specifically, Fig. 2 is the flow chart of the weight for each dimension of determination that an exemplary embodiment provides.As shown in Fig. 2,
The step of determining the weight of each dimension according to the characteristic of consumption user (step S12) may comprise steps of.
In step S121, calculate separately include in its characteristic characteristic corresponding with each dimension consumption use
The ratio of the total number of the number and consumption user at family obtains the data frequency of corresponding dimension.
Wherein, the data value in characteristic corresponding with each dimension namely this dimension, it is corresponding with each dimension
Characteristic may include multiple characteristics.For example, characteristic corresponding with gender dimension may include male and female, with
The corresponding characteristic of age dimension may include 20-30 years old, 30-40 years old and 40-50 years old.
In the characteristic of a consumption user, not necessarily have in each dimension.The feature of certain dimension
Data occur more in consumption user, show that the dimension is more important for the description of feature user, the weight of the dimension is just
It is higher.
For example, the total number of consumption user is 100, it include characteristic corresponding with gender dimension in characteristic
The number of the consumption user of (male or female) is 20, then the data frequency of gender dimension is 0.2 (20/100).
In step S122, the total number for calculating separately whole user is corresponding including each dimension with its characteristic
The logarithm (for example, being bottom with ten) of the ratio of the number of the whole user of characteristic obtains the reverse frequency of corresponding dimension.
Wherein, whole user may include consumption user and non-consumption user (user of non-post-consumer target product).It can
Include with understanding, in characteristic the whole user of characteristic corresponding with dimension number it is more, then the dimension
The difference degree of degree is lower, and corresponding weight is lower.
For example, the total number of whole user is 10,000,000, it include spy corresponding with gender dimension in characteristic
The number for levying the whole user of data (male or female) is 10,000, then the reverse frequency of gender dimension is
In step S123, respectively by the data frequency of each dimension multiplied by reverse frequency, the weight of corresponding dimension is obtained.
In the disclosure, data frequency and reverse frequency be respectively equivalent to word frequency in the reverse document-frequency algorithm of word frequency-and
Reverse document-frequency.According to the design of the algorithm, the weight of each dimension can be the data frequency of the dimension multiplied by reverse frequency
Rate.On the basis of above-mentioned example, the weight of gender dimension can be 0.2 × 3=0.6.
So far, for each dimension, its corresponding weight can be calculated.In the embodiment, according to the reverse text of word frequency-
Part frequency algorithm calculates the weight of each dimension, accurately embodies each dimension for the importance of feature user, simply
Easy, availability is good.
Over time, a large amount of accumulations of data volume may impact the accuracy of result.Also, it is special
The fixed period, which is likely to the relevance between the characteristic and target product to some users, larger impact.For example,
During Europe Cup football match, there are a large amount of pseudo- football fans, consumed a large amount of football products.And after competing, these pseudo- football fans for
The interest of the football product of post-consumer can greatly weaken before, or even disappear.In consideration of it, in the another embodiment of the disclosure
In, it may be considered that the historical development of weight in each dimension, to determine its weight.
Fig. 3 is the flow chart of the weight for each dimension of determination that another exemplary embodiment provides.As shown in figure 3, according to
The characteristic of consumption user determines that the step of weight of each dimension (step S12) may comprise steps of.
In step S124, periodically determine each dimension in current period according to the characteristic of consumption user
Interval weight.
It is, each dimension can be determined in this week according to the consumption user in each period (for example, one month)
Weight in phase, i.e. interval weight.Interval weight is obtained by the data in the period, embodies the dimension and target in specific period
The correlation degree of product.
In step s 125, according to the history interval weight of each dimension and current interval weight, each dimension is determined
Weight.
The step of after determining the interval weight of dimension, can obtaining the weight of the dimension accordingly, being applied to later.It examines
Consider in general, time closer data can more embody the trend of current development, such as can be to history interval weight
Scheduled smaller and biggish weight is assigned respectively with current interval weight, by both history interval weight and current interval weight
Weighted sum obtains the weight of the dimension.
For example, it can be obtained by one month nearest characteristic, the corresponding interval weight of gender dimension is 0.6
(current interval weight).And it is available according to the characteristic before one month, the weight of gender dimension is 0.4 (history section
Weight).Scheduled 0.8 and 0.2 weight can be assigned respectively to current interval weight and history interval weight, then gender dimension
Weight can be 0.6 × 0.8+0.4 × 0.2=0.56.
For another example, it can be obtained by the characteristic during one month of Europe Cup football match, the corresponding section power of gender dimension
Weight is 0.6 (current interval weight).And it is available according to the characteristic before one month, the weight of gender dimension (is gone through for 0.9
History interval weight).In view of during football match, since propaganda strength is strong, phenomenon of following the wind is serious, a large amount of pseudo- balls are emerged
Fan.So the weight of current interval weight can be set to smaller, the weight of history interval weight can be set to larger.Example
Such as, scheduled 0.1 and 0.9 weight can be assigned respectively to current interval weight and history interval weight, then gender dimension
Weight can be 0.6 × 0.1+0.9 × 0.9=0.87.
In the embodiment, it is contemplated that the historical development of dimension weight determines the weight of dimension in a manner of increment, so that
The weight of identified dimension more meets current practice, to keep product portrait more accurate.
As previously mentioned, the value of each characteristic for giving a mark to consumption user (is also the power of characteristic below
Weight), it can be rule of thumb or test determination.The weight of characteristic can also be according to the characteristic itself of user come really
It is fixed.In an embodiment of the disclosure, according to the weight of the characteristic of consumption user and identified each dimension, determine every
The step of scoring of a consumption user (step S13) may include step S131 and step S132.
In step S131, the weight of each characteristic is determined according to the characteristic of the consumption user.
The step can be implemented according to the design of the reverse document-frequency algorithm of word frequency-, specifically, can be according to following step
It is rapid to implement.
(1) number of the consumption user in its characteristic including fisrt feature data and the total number of consumption user are calculated
Ratio, obtain the data frequency of fisrt feature data.
Wherein, fisrt feature data can be any feature data of consumption user.A certain characteristic is in consumption user
In occur more, show that this feature data are more important for the description of feature user, the weight of this feature data is higher.
For example, the total number of consumption user is 100, the number in characteristic including the consumption user of male is 80, then
The data frequency of masculinity data is 0.8 (80/100).
(2) number of the whole user including fisrt feature data in the total number and characteristic of whole user is calculated
The logarithm of ratio obtains the reverse frequency of fisrt feature data.Wherein, whole user includes consumption user and non-consumption user.
It is understood that the number of the whole user in characteristic including fisrt feature data is more, then this feature
The difference degree of data is lower, and corresponding weight is lower.
For example, the total number of whole user is 10,000,000, it include the number of the whole user of male in characteristic
It is 10,000, then reverse frequency is
(3) data frequency of fisrt feature data is obtained into fisrt feature number multiplied by the reverse frequency of fisrt feature data
According to weight.
As described above, the weight of each characteristic is this feature number according to the design of the reverse document-frequency algorithm of word frequency-
According to data frequency multiplied by reverse frequency.On the basis of above-mentioned example, the weight of masculinity data can be 0.8 × 3=
2.4。
Herein, the weight of characteristic is equivalent to the value of characteristic as described in the examples shown in FIG. 1, with Fig. 1's
Embodiment is compared, and in the embodiment, determines characteristic according to characteristic itself and the reverse document-frequency algorithm of word frequency-
Weight (i.e. the value of characteristic), accurately embodies the feature of feature user, and simple and easy, availability is good.
In step S132, according to the weight of the weight of each characteristic and identified each dimension, determine each
The scoring of consumption user.
Specifically, the weight of characteristic of the consumption user in each dimension can be weighted (the power of dimension
Weight) summation, obtain the scoring of the consumption user.The scoring embodies the feature of consumption user and the correlation degree of target product.
Similarly with the embodiment in above-mentioned Fig. 3, for the weight of characteristic, feature can also be periodically determined
The interval weight of data considers the historical development of weight, and the weight of characteristic is determined in a manner of increment, so that determining
The weight of characteristic more meet current practice, to keep product portrait more accurate.The embodiment will not be described in great detail.
In practice, each product is since new listing.In view of when initial, target product does not have also
There is the data volume of consumption user or consumption user less, like product similar with its can be found, at this time to imitate new product
Consumption user, further obtain new product product portrait.
Fig. 4 is the flow chart for the method for constructing product portrait that another exemplary embodiment provides.As shown in figure 4,
On the basis of Fig. 1, in the step of characteristic of the consumption user of target product is classified according to multiple dimensions (step
11) before, the method can also include the following steps.
In step s 110, like product similar with target product is determined.
Like product can be rule of thumb found, or can determine like product using some algorithms.It is real one
It applies in example, step S110 may comprise steps of:
According to any one or the more persons in Jie Kade Coefficient Algorithm and Pearson's similarity algorithm, determine target product and
The similarity of other products；Like product similar with target product is determined according to identified similarity.
For example, in the embodiment according to Pearson's similarity algorithm similarity can be calculated using following formula:
Wherein, r indicates the similarity of target product and another product, and X and Y respectively indicate target product and another product
The weight of i-th dimension degree, n indicate dimension number.Wherein, the weight of the dimension of target product can for example be obtained by experience.
It for another example, can be first according to Jie Kade according in Jie Kade Coefficient Algorithm and the embodiment of Pearson's similarity algorithm
Coefficient Algorithm is screened, and in the multiple products obtained after screening, Pearson's similarity algorithm is recycled to obtain similarity.
Next, similarity can be greater than predetermined after determining like product similar with target product according to similarity
The product of similarity threshold be determined as like product.
In step S111, the consumption user of like product is determined as to the consumption user of target product.
In this embodiment, it can estimate to obtain by like product in the insufficient situation of data volume of consumption user
The product of target product is drawn a portrait, to solve the problems, such as product " cold start-up " data deficiencies.
Fig. 5 is the block diagram for the device for constructing product portrait that an exemplary embodiment provides.As shown in figure 5, described
Device 10 for constructing product portrait may include categorization module 11, weight determination module 12, scoring determining module 13, feature
User's determining module 14, product portrait determining module 15.
Categorization module 11 is for classifying the characteristic of the consumption user of target product according to multiple dimensions.
Weight determination module 12 is used to determine the weight of each dimension according to the characteristic of the consumption user.
The determining module 13 that scores is used for according to the characteristic of the consumption user and the weight of identified each dimension,
Determine the scoring of each consumption user.
Feature user determining module 14 is used to determine the feature user in the consumption user according to the scoring.
Product draws a portrait determining module 15 for the characteristic according to the feature user, determines the production of the target product
Product portrait.
Optionally, the weight determination module 12 may include the first data frequency computational submodule, the first reverse frequency
Computational submodule and the first weight determine submodule.
First data frequency computational submodule includes spy corresponding with each dimension for calculating separately in its characteristic
The ratio for levying the number of the consumption user of data and the total number of the consumption user obtains the data frequency of corresponding dimension.
First reverse frequency computational submodule includes in the total number and its characteristic for calculating separately whole user
The logarithm of the ratio of the number of the whole user of characteristic corresponding with each dimension obtains the reverse frequency of corresponding dimension,
Wherein, the whole user includes the consumption user and non-consumption user.
First weight determines submodule, for being corresponded to respectively by the data frequency of each dimension multiplied by reverse frequency
The weight of dimension.
Optionally, the weight determination module 12 may include that interval weight determines that submodule and the second weight determine submodule
Block.
Interval weight determines submodule for periodically determining each dimension according to the characteristic of the consumption user
Interval weight in current period.
Second weight determines that submodule for the history interval weight and current interval weight according to each dimension, determines every
The weight of a dimension.
Optionally, the scoring determining module 13 may include that data weighting determines submodule and the determining submodule that scores.
Data weighting determines submodule for determining the power of each characteristic according to the characteristic of the consumption user
Weight.
It scores and determines that submodule is used to determine according to the weight of each characteristic and the weight of identified each dimension
The scoring of each consumption user.
Optionally, the data weighting determines that submodule includes the second data frequency computational submodule, the second reverse frequency
Computational submodule and third weight determine submodule.
Second data frequency computational submodule is used to calculate the consumption user in its characteristic including fisrt feature data
Number and the consumption user total number ratio, obtain the data frequency of the fisrt feature data.
It includes described that second reverse frequency computational submodule, which is used to calculate in the total number and its characteristic of whole user,
The logarithm of the ratio of the number of the whole user of fisrt feature data obtains the reverse frequency of the fisrt feature data, wherein
The entirety user includes the consumption user and non-consumption user.
Third weight determines that submodule is used for the data frequency of the fisrt feature data multiplied by the fisrt feature number
According to reverse frequency, obtain the weight of the fisrt feature data.
Optionally, described device 10 can also include like product determining module and consumption user determining module.
Like product determining module is for determining like product similar with the target product.
Consumption user determining module is used to for the consumption user of the like product being determined as the consumption of the target product
User.
About the device in above-described embodiment, wherein modules execute the concrete mode of operation in related this method
Embodiment in be described in detail, no detailed explanation will be given here.
Through the above technical solutions, the weight of each dimension is determined based on the characteristic of consumption user itself, so that
The weight of dimension more accurately embodies the feature of consumption user.Therefore, the disclosure is comprehensively considering each of user characteristic data
On the basis of the weight of a dimension, more accurate product portrait is constructed, the accurate positionin and production to consumption user are conducive to
The improvement of product increases product benefit.
The preferred embodiment of the disclosure is described in detail in conjunction with attached drawing above, still, the disclosure is not limited to above-mentioned reality
The detail in mode is applied, in the range of the technology design of the disclosure, a variety of letters can be carried out to the technical solution of the disclosure
Monotropic type, these simple variants belong to the protection scope of the disclosure.
It is further to note that specific technical features described in the above specific embodiments, in not lance
In the case where shield, it can be combined in any appropriate way.In order to avoid unnecessary repetition, the disclosure to it is various can
No further explanation will be given for the combination of energy.
In addition, any combination can also be carried out between a variety of different embodiments of the disclosure, as long as it is without prejudice to originally
Disclosed thought equally should be considered as disclosure disclosure of that.
Claims (8)
1. a kind of method for constructing product portrait, which is characterized in that the described method includes:
The characteristic of the consumption user of target product is classified according to multiple dimensions；
The weight of each dimension is determined according to the characteristic of the consumption user；
According to the weight of the characteristic of the consumption user and identified each dimension, commenting for each consumption user is determined
Point；
Wherein, described according to the characteristic of the consumption user and the weight of identified each dimension, determine each consumption
The step of scoring of user includes:
The weight of each characteristic is determined according to the characteristic of the consumption user；
According to the weight of the weight of each characteristic and identified each dimension, the scoring of each consumption user is determined；
The feature user in the consumption user is determined according to the scoring；
According to the characteristic of the feature user, the product portrait of the target product is determined.
2. the method according to claim 1, wherein described determine often according to the characteristic of the consumption user
The step of weight of a dimension includes:
Calculate separately includes that the number of consumption user of characteristic corresponding with each dimension disappears with described in its characteristic
The ratio of the total number at expense family obtains the data frequency of corresponding dimension；
Calculate separately include in the total number and its characteristic of whole user characteristic corresponding with each dimension entirety
The logarithm of the ratio of the number of user obtains the reverse frequency of corresponding dimension, wherein the entirety user includes that the consumption is used
Family and non-consumption user；
Respectively by the data frequency of each dimension multiplied by reverse frequency, the weight of corresponding dimension is obtained.
3. the method according to claim 1, wherein described determine often according to the characteristic of the consumption user
The step of weight of a dimension includes:
Interval weight of each dimension in current period is periodically determined according to the characteristic of the consumption user；
According to the history interval weight and current interval weight of each dimension, the weight of each dimension is determined.
4. the method according to claim 1, wherein described determine often according to the characteristic of the consumption user
The step of weight of a characteristic includes:
Calculate the number of the consumption user in its characteristic including fisrt feature data and the total number of the consumption user
Ratio obtains the data frequency of the fisrt feature data；
Calculate the number of the whole user including the fisrt feature data in the total number and its characteristic of whole user
The logarithm of ratio obtains the reverse frequency of the fisrt feature data, wherein it is described entirety user include the consumption user and
Non-consumption user；
By the data frequency of the fisrt feature data multiplied by the reverse frequency of the fisrt feature data, it is special to obtain described first
Levy the weight of data.
5. the method according to claim 1, wherein in the characteristic of the consumption user by target product
Before the step of being classified according to multiple dimensions, the method also includes:
Determine like product similar with the target product；
The consumption user of the like product is determined as to the consumption user of the target product.
6. a kind of for constructing the device of product portrait, which is characterized in that described device includes:
Categorization module, for the characteristic of the consumption user of target product to be classified according to multiple dimensions；
Weight determination module, for determining the weight of each dimension according to the characteristic of the consumption user；
Score determining module, for determining according to the characteristic of the consumption user and the weight of identified each dimension
The scoring of each consumption user；
Wherein, the scoring determining module includes:
Data weighting determines submodule, for determining the weight of each characteristic according to the characteristic of the consumption user；
It scores and determines submodule, for determining every according to the weight of each characteristic and the weight of identified each dimension
The scoring of a consumption user；
Feature user's determining module, for determining the feature user in the consumption user according to the scoring；
Product portrait determining module determines that the product of the target product is drawn for the characteristic according to the feature user
Picture.
7. device according to claim 6, which is characterized in that the weight determination module includes:
First data frequency computational submodule includes characteristic corresponding with each dimension for calculating separately in its characteristic
According to consumption user number and the consumption user total number ratio, obtain the data frequency of corresponding dimension；
First reverse frequency computational submodule, for calculate separately whole user total number and its characteristic in include with often
The logarithm of the ratio of the number of the whole user of the corresponding characteristic of a dimension obtains the reverse frequency of corresponding dimension, wherein
The entirety user includes the consumption user and non-consumption user；
First weight determines submodule, for respectively by the data frequency of each dimension multiplied by reverse frequency, obtaining corresponding dimension
Weight.
8. device according to claim 6, which is characterized in that the weight determination module includes:
Interval weight determines submodule, for periodically determining that each dimension is being worked as according to the characteristic of the consumption user
Interval weight in the preceding period；
Second weight determines submodule, for the history interval weight and current interval weight according to each dimension, determines each
The weight of dimension.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610964911.5A CN106548375B (en) | 2016-11-04 | 2016-11-04 | Method and apparatus for constructing product portrait |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610964911.5A CN106548375B (en) | 2016-11-04 | 2016-11-04 | Method and apparatus for constructing product portrait |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106548375A CN106548375A (en) | 2017-03-29 |
| CN106548375B true CN106548375B (en) | 2019-10-08 |
Family
ID=58395295
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610964911.5A Active CN106548375B (en) | 2016-11-04 | 2016-11-04 | Method and apparatus for constructing product portrait |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106548375B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107392656A (en) * | 2017-07-07 | 2017-11-24 | 芜湖恒天易开软件科技股份有限公司 | The formulation of the industry user that hires a car portrait and marketing effectiveness tracking |
| CN107516237A (en) * | 2017-07-22 | 2017-12-26 | 长沙兔子代跑网络科技有限公司 | A kind of drawn a portrait according to user excavates the method and device of generation race client |
| CN107944481B (en) * | 2017-11-16 | 2022-02-18 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN108805383A (en) * | 2018-03-20 | 2018-11-13 | 东华大学 | A kind of user's portrait platform and application for washing shield big data based on clothes |
| CN108564262A (en) * | 2018-03-31 | 2018-09-21 | 甘肃万维信息技术有限责任公司 | Enterprise's portrait big data model system based on big data analysis |
| CN109325796B (en) * | 2018-08-13 | 2023-09-26 | 中国平安人寿保险股份有限公司 | Potential user screening method, device, computer equipment and storage medium |
| CN109409936A (en) * | 2018-09-28 | 2019-03-01 | 深圳壹账通智能科技有限公司 | Customer consumption portrait generation method, device, equipment and readable storage medium storing program for executing |
| CN110009438A (en) * | 2018-11-07 | 2019-07-12 | 爱保科技（横琴）有限公司 | Information processing method and device based on car owner's community |
| CN109753994B (en) * | 2018-12-11 | 2024-05-14 | 东软集团股份有限公司 | User image drawing method, device, computer readable storage medium and electronic equipment |
| CN111352962B (en) * | 2018-12-24 | 2024-03-29 | 网智天元科技集团股份有限公司 | Customer portrait construction method and device |
| CN109919219B (en) * | 2019-03-01 | 2021-02-26 | 北京邮电大学 | Xgboost multi-view portrait construction method based on kernel computing ML-kNN |
| CN111581296B (en) * | 2020-04-02 | 2022-08-16 | 深圳壹账通智能科技有限公司 | Data correlation analysis method and device, computer system and readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103309960A (en) * | 2013-05-29 | 2013-09-18 | 亿赞普(北京)科技有限公司 | Method and device for extracting multidimensional information of network public sentiment event |
| CN105761106A (en) * | 2016-02-07 | 2016-07-13 | 徐璇炫 | Object ordering method based on user setting and device thereof |
| CN105869001A (en) * | 2015-01-19 | 2016-08-17 | 苏宁云商集团股份有限公司 | Customized product recommendation guiding method and system |
-
2016
- 2016-11-04 CN CN201610964911.5A patent/CN106548375B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103309960A (en) * | 2013-05-29 | 2013-09-18 | 亿赞普(北京)科技有限公司 | Method and device for extracting multidimensional information of network public sentiment event |
| CN105869001A (en) * | 2015-01-19 | 2016-08-17 | 苏宁云商集团股份有限公司 | Customized product recommendation guiding method and system |
| CN105761106A (en) * | 2016-02-07 | 2016-07-13 | 徐璇炫 | Object ordering method based on user setting and device thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106548375A (en) | 2017-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106548375B (en) | Method and apparatus for constructing product portrait | |
| TWI743428B (en) | Method and device for determining target user group | |
| US8527352B2 (en) | System and method for generating optimized bids for advertisement keywords | |
| US10509791B2 (en) | Statistical feature engineering of user attributes | |
| CN106372249B (en) | A kind of clicking rate predictor method, device and electronic equipment | |
| US20170206361A1 (en) | Application recommendation method and application recommendation apparatus | |
| US9026594B2 (en) | Method and system for determining user impact on their content pools within an online social network | |
| US20120072220A1 (en) | Matching text sets | |
| US20230054747A1 (en) | Automatic Generation of Preferred Views for Personal Content Collections | |
| US20060116920A1 (en) | Methods and systems for forecasting with model-based PDF estimates | |
| CN110222975A (en) | A kind of loss customer analysis method, apparatus, electronic equipment and storage medium | |
| CN107526810B (en) | Method and device for establishing click rate estimation model and display method and device | |
| EP3279806A1 (en) | Data processing method and apparatus | |
| Rafeh et al. | An adaptive approach to dealing with unstable behaviour of users in collaborative filtering systems | |
| US20170185652A1 (en) | Bias correction in content score | |
| US20140229510A1 (en) | Method of temporal bipartite projection | |
| US20230050197A1 (en) | Systems and methods quantifying trust perceptions of entities within social media documents | |
| US20190236718A1 (en) | Skills-based characterization and comparison of entities | |
| US20190087838A1 (en) | Determining brand exclusiveness of users | |
| US9064018B2 (en) | Information providing system that provides information based on strenghth of desire and current situation | |
| CN109543940B (en) | Activity evaluation method, activity evaluation device, electronic equipment and storage medium | |
| CN106886559A (en) | The collaborative filtering method of good friend's feature and similar users feature is incorporated simultaneously | |
| KR20130134046A (en) | Cosine similarity based expert recommendation technique using hybrid collaborative filtering | |
| WO2015101161A1 (en) | Method and device for generating user page corresponding to target system | |
| CN103020141A (en) | Method and equipment for providing searching results |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |