CN106067139A - A kind of mthods, systems and devices of construction feature database - Google Patents
A kind of mthods, systems and devices of construction feature database Download PDFInfo
- Publication number
- CN106067139A CN106067139A CN201610374915.8A CN201610374915A CN106067139A CN 106067139 A CN106067139 A CN 106067139A CN 201610374915 A CN201610374915 A CN 201610374915A CN 106067139 A CN106067139 A CN 106067139A
- Authority
- CN
- China
- Prior art keywords
- information
- party
- big
- network
- strongly connected
- 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
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
-
- 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
Abstract
The present invention provides a kind of method of construction feature information bank, comprising: the network information issued in obtaining third party's accounts information and Preset Time；Analyze the interactive information between third party's account, generate the information transfer network based on described interactive information；Obtain based on all maximal independent sets in all very big strongly connected components of described information transfer network and each very big strongly connected components；In all maximal independent sets from each very big strongly connected components, select containing minimum maximal independent set；After the some duplicate removal that comprise minimum maximal independent sets all in all very big strongly connected componentses, merge into a vertex set；Merge third party's accounts information and vertex set information, generate characteristic information storehouse.Additionally provide corresponding system and device with the method.The present invention can support the marketing of items of different types or activity neatly, had both saved marketing resource, reduces again the construction cost of Marketing Model, also selects to be very helpful to the new user of operation aspect simultaneously.
Description
Technical field
The present invention relates to communication technical field, particularly relate to a kind of identification, the method analyzing social networks, system and dress
Put.
Background technology
In amusement performance industry, especially in internet ticket selling industry, find target marketing customers, be usually and determining item
On the basis of mesh or determination activity scene, the mode of relevant item or similar item has been seen to enter row filter based on the conception of history
's.This mode does not consider the finiteness of ticket amount resource, does not considers the migration of user interest yet, cannot obtain accurately simultaneously yet
Take new user, ultimately cause marketing effectiveness difference, the problem such as wasting of resources.
Existing technical scheme, such as application number 2015104847492, a kind of build and analyze the method for social networks and be
In system, disclose the technical scheme building social networks based on stable contacts, but message can not be carried out effectively by the program
Propagate, such as:
User A and B, for working principle, often has the contact of phone, note, but in real life, if A is found that one
The preferential bag of individual flow, and B does not knows, often because be between user different region, relation intimate not or do not have with
The reasons such as one interest, A can't actively inform this preferential policy of B, and therefore stable contacts can not constitute the basis having effect spread,
Therefore the mode often poor effect of marketing is propagated.
Secondly, the technical scheme based on complete subgraph searching algorithm of above-mentioned patent application publication is based on simply undirected
Figure, is all stable contacts relation between any two user in complete subgraph requirement figure own, and the condition of this requirement own limited
Greatly, in actual applications, major part is all 2 points, 3 complete graphs, causes the customer group of marketing not have on practical significance
Reduce, do not reach the purpose effectively utilizing limited marketing resource.
Again, the application of AHP step analysis algorithm is limited to user's interworking case of field of telecommunications by above-mentioned patent application,
Therefore the high-impact user screening is high-value user, itself is not high propagated user, i.e. itself considers
Factor and the factor that considers of stable contacts without actual variance.
Content of the invention
In order to solve above-mentioned technical problem, present invention is primarily targeted at and a kind of social network analysis and structure are provided
Mthods, systems and devices, utilize the huge social networks that third party's Account Logon customer group is formed, and identify have high-impact
Customer group is seed marketing user, is the precision marketing of each scene, establishes solid user data basis.
For reaching above-mentioned purpose, one aspect of the present invention provides a kind of method of construction feature information bank, comprising:
The network information issued in step 1, acquisition third party's accounts information and Preset Time；
Step 2, the interactive information analyzed between third party's account, generate the information transfer network based on described interactive information；
Step 3, acquisition divide based on all very big strongly connected components and each very big strong continune of described information transfer network
All maximal independent sets in Zhi；
In step 4, all maximal independent sets from each very big strongly connected components, select containing minimum very big independence
Collection；
After step 5, the some duplicate removal that minimum maximal independent sets all in all very big strongly connected componentses are comprised, merge into one
Individual vertex set；
Step 6, merging third party's accounts information and vertex set information, generate characteristic information storehouse.
Further, in described step 1, third party's accounts information and the interior network information issued of Preset Time, bag are obtained
Include:
The Back ground Information of third party's account is obtained and network that third party's account is issued in Preset Time during associating login
Information.
Further, in described step 2, analyze the interactive information between third party's account, generate based on described interactive information
Information transfer network, comprising:
By third party's accounts information of acquisition in step one and the network information of issue in Preset Time, extract third party
Interactive information between all third party's accounts in the network information that account is issued in this Preset Time, based on described interactive information
Direction of transfer, determine oriented information transfer network.
Further, in described step 3, obtain based on all very big strongly connected components of described information transfer network and every
All maximal independent sets in individual very big strongly connected components, specifically,
The application Tarjan algorithm whole information transfer network of traversal, obtain based on described information transfer network all greatly
Strongly connected components；
Theoretical in conjunction with vertex coloring, obtain all maximal independent sets in each very big strongly connected components.
Further, the whole information transfer network of described application Tarjan algorithm traversal, obtains based on the transmission of described information
The all very big strongly connected components of network, comprising:
Step 3.11, elder generation all initialize white the color of all of node, and stack are emptied；
Step 3.12, find a white node；
Step 3.13, give one timestamp of node, in node stacking, and node is labeled as grey, makes Low [i]
=DFN [i]；
Step 3.14, traversing nodes i each edge (i, j), if color [j] be white, just to node i repetition step 3.2
～3.5, and make Low [i]=min (Low [j], low [i]) if. color [j] is grey, make Low [i]=min (Low [i],
DFN[j])；If black is left intact.
Step 3.15, the color of node is changed into black, if Low [i]=DFN [i], just from stack top to node i
Element eject；
Step 3.16, repeat step 3.2, to there is no White vertex；
Wherein, color [i] is the state for indicating node i: white refers to also be less than search, grey just searched,
Black is disposed.
Further, described combination vertex coloring is theoretical, obtains all very big independence in each very big strongly connected components
Collection, comprising:
Digraph in very big connected component is processed into non-directed graph；
The limit filtering any point-to-point transmission in described non-directed graph is a limit.
Further, described combination vertex coloring is theoretical, obtains all very big independence in each very big strongly connected components
Collection, specifically includes:
Step 3.21: the adjacency matrix A of input figure G, initializes size scale of A, independent sets S=H.
Step 3.22: if scale≤1, output S and last vertex v ', would recover last G and A, update scale；
Otherwise jump procedure 3；
Step 3.33: one by one to all v ∈ V (G), S=S+{v}, if the point in v and all G is adjacent, jump procedure 4；
Otherwise jump procedure 5；
Step 3.34: judge whether figure G is complete graph, if complete graph, exports S, S=S-{v}, jump procedure 3；As
Fruit is not complete graph, S=S-{v}, jump procedure 3.
Step 3.35:G '=G-NG [v], A ' is the adjacency matrix of G ', makes G=G ', A=A ', updates scale, redirects step
Rapid 2；
Further, described select containing minimum maximal independent set, comprising:
Select the minimum maximal independent set containing minimum some number.
Further, described third party's accounts information, comprising:
The basic business attribute of third party's account, microblogging attribute and/or own user tag.
Another aspect of the present invention additionally provides the system of a kind of construction feature information bank, comprising:
Acquisition module, for obtaining third party's accounts information and the interior network information issued of Preset Time；
Analyzing module, for analyzing the interactive information between third party's account, the information based on described interactive information that generates passes
Pass network；
Acquisition module, for obtaining based on all very big strongly connected components of described information transfer network and each is greatly strong
All maximal independent sets in connected component；
Screening module, for, in all maximal independent sets from each very big strongly connected components, selecting containing minimum pole
Big independent sets；
Merge module, after being used for the some duplicate removal that minimum maximal independent sets all in all very big strongly connected componentses are comprised,
Merge into a vertex set；
Generation module, is used for merging third party's accounts information and vertex set information, generates characteristic information storehouse.
Further, described acquisition module, comprising:
Acquiring unit, is used for when combining login obtaining the Back ground Information of third party's account and third party's account when default
The network information of interior issue.
Further, described analysis module, comprising:
Extraction unit, in the information that described acquisition module gathers, extracts third party's account in this Preset Time
Interactive information between all third party's accounts in the network information issued, based on the direction of transfer of described interactive information, determination has
To information transfer network.
Further, described acquisition module, comprising:
First acquiring unit, is used for applying the whole information transfer network of Tarjan algorithm traversal, obtains based on described information
The all very big strongly connected components of transmission network；
Second acquisition unit, is used for combining vertex coloring theoretical, obtain in each very big strongly connected components all greatly
Independent sets.
Further, described second acquisition unit, comprising:
Processing unit, for being processed into non-directed graph by the digraph in very big connected component；
Filter element, is a limit for filtering the limit of any point-to-point transmission in the non-directed graph in described processing unit.
Further, described screening module, comprising:
Screening unit, for selecting the minimum maximal independent set containing minimum some number.
Another aspect of the present invention additionally provides the device of a kind of construction feature information bank, including described system.
The characteristic information storehouse that the present invention builds, can support the marketing of items of different types or activity, both more neatly
Save marketing resource, reduced again the construction cost of Marketing Model, can reach and once build the application of many scenes, the i.e. weight of model
Multiple utilization rate is high, is also very helpful the new user that draws over to one's side of operation aspect simultaneously, is finally reached maximum resource utilization, throwing
Enter the maximized purpose of output ratio.
Brief description
The flow chart of method one embodiment for a kind of construction feature database according to the present invention for the Fig. 1；
The structured flowchart of system one embodiment for a kind of construction feature database according to the present invention for the Fig. 2.
Detailed description of the invention
The method of a kind of construction feature database that the embodiment of the present invention is provided by 1-2, system and dress below in conjunction with the accompanying drawings
Put and be described in detail.
The inventive concept of the embodiment of the present invention is step 1, obtains third party's accounts information and the interior net issued of Preset Time
Network information；Step 2, the interactive information analyzed between third party's account, generate the information transfer network based on described interactive information；Step
Rapid 3rd, obtain based in all very big strongly connected components of described information transfer network and each very big strongly connected components all
Maximal independent set；In step 4, all maximal independent sets from each very big strongly connected components, select containing minimum greatly solely
Vertical collection；After step 5, the some duplicate removal that minimum maximal independent sets all in all very big strongly connected componentses are comprised, merge into one
Vertex set；Step 6, merging third party's accounts information and vertex set information, generate characteristic information storehouse.The one of present invention offer
Planting the mthods, systems and devices of construction feature information bank, the Filtering system that can make target marketing customers is old based on inside
The accurate label of user, separately wards off the angle logging in from third party in strange footpath, and simply screening is converted to high propagated use
The identification at family comes up, and is both screened old user accurately, is also propagated new user, has saved marketing resource,
Reach the precision marketing of project or activity, also reached to draw new effect in operation aspect simultaneously.
Below with reference to accompanying drawing, the present invention is described in detail.
Embodiment one
With reference to Fig. 1, Fig. 1 shows the flow chart of an embodiment of the method for the present invention.
The embodiment of the present invention one provides a kind of method of construction feature information bank, comprising:
The network information issued in step 1, acquisition third party's accounts information and Preset Time.
Concrete, obtain the Back ground Information of third party's account when combining login and third party's account is issued in Preset Time
The network information.Described third party's accounts information, comprising: the basic business attribute of third party's account, microblogging attribute and/or own
User tag.
Step 2, the interactive information analyzed between third party's account, generate the information transfer network based on described interactive information.
By third party's accounts information of acquisition in step one and the network information of issue in Preset Time, extract third party
Interactive information between all third party's accounts in the network information that account is issued in this Preset Time, based on described interactive information
Direction of transfer, determine oriented information transfer network.
Concrete, by analyzing concern, comment, forwarding between all third party's accounts, putting situations such as praising, determine that information can
The standard (as user A has paid close attention to user B, then can be defined as B and can pass information to A) reaching；Generate third party's account
Information transfer network (if user B can pass information to A, then generate a directed edge from user B to user A；If user
A can pass information to B, then also generate a directed edge from A to user B；Ultimately generate huge oriented information transmission
Figure).
Step 3, acquisition divide based on all very big strongly connected components and each very big strong continune of described information transfer network
All maximal independent sets in Zhi.
The application Tarjan algorithm whole information transfer network of traversal, obtain based on described information transfer network all greatly
Strongly connected components；Theoretical in conjunction with vertex coloring, obtain all maximal independent sets in each very big strongly connected components.
In step 4, all maximal independent sets from each very big strongly connected components, select containing minimum very big independence
Collection.
After step 5, the some duplicate removal that minimum maximal independent sets all in all very big strongly connected componentses are comprised, merge into one
Individual vertex set.
Step 6, merging third party's accounts information and vertex set information, generate characteristic information storehouse.
Concrete, merge the basic business attribute of third party's account, microblogging attribute, own user tag etc. with minimum greatly
Vertex set information (user's aggregate information) after independent sets merging, generates user's marketing message storehouse.
The whole information transfer network of described application Tarjan algorithm traversal, obtains owning based on described information transfer network
Very big strongly connected components, comprising:
Step 3.11, elder generation all initialize white the color of all of node, and stack are emptied；
Step 3.12, find a white node；
Step 3.13, give one timestamp of node, in node stacking, and node is labeled as grey, makes Low [i]
=DFN [i]；
Step 3.14, traversing nodes i each edge (i, j), if color [j] be white, just to node i repetition step 3.2
～3.5, and make Low [i]=min (Low [j], low [i]) if. color [j] is grey, make Low [i]=min (Low [i],
DFN[j])；If black is left intact.
Step 3.15, the color of node is changed into black, if Low [i]=DFN [i], just from stack top to node i
Element eject；
Step 3.16, repeat step 3.2, to there is no White vertex.
Wherein, digraph: if giving one direction of each edge regulation of figure, then the figure obtaining is referred to as digraph.
Unilateral connection: letting d be digraph, between any pair of node, at least a node is up to another node, then claim
This digraph is unilaterably connected.
Strong continune: let d be digraph, is mutually accessibility between any pair of node, then claim this digraph to be to connect by force
Lead to.
Strongly connected components: let d be digraph, if V is a point set of D, in V, any point-to-point transmission is all mutually can to reach
, then the induced subgraph of one-tenth V is the strongly connected components of D.
Very big strongly connected components: letting d be digraph, G is a strongly connected components of D, if any one is not in G in D
Point add in G after, G can become non-strongly connected components, then claim the very big strongly connected components that G is D.
Timestamp: DFN [i] refers to the time that node i is traversed.
The ancestors the earliest that Low [i]: refer in search tree, node i and its descendants can have access to, Low [i]=Min
(DFN [i], DFN [j], Low [k]) wherein j is ancestors' (we link the limit of ancestors backward dege descendants) of i, and k is the son of i
Female.
The color of node: color [i] is the state for indicating node i: white refers to also be less than search, and grey is
Searched, black is disposed.Represent white, grey, black with-1,0,1 respectively in practical operation.
Further, described combination vertex coloring is theoretical, obtains all very big independence in each very big strongly connected components
Collection, comprising:
Digraph in very big connected component is processed into non-directed graph；
The limit filtering any point-to-point transmission in described non-directed graph is a limit.
Wherein, vertex coloring is theoretical, also cries summit double-colored theory, refers in a non-directed graph, selects a point set,
One of described point set point is dyed a kind of color, and left point dyes another color, any adjacent point of final guarantee
Color be all different；Now many collection that comprises in one of them color to count are collectively referred to as maximal independent set.
Further, described combination vertex coloring is theoretical, obtains all very big independence in each very big strongly connected components
Collection, specifically includes:
Step 3.21: the adjacency matrix A of input figure G, initializes size scale of A, independent sets S=H.
Step 3.22: if scale≤1, output S and last vertex v ', would recover last G and A, update scale；
Otherwise jump procedure 3；
Step 3.33: one by one to all v ∈ V (G), S=S+{v}, if the point in v and all G is adjacent, jump procedure 4；
Otherwise jump procedure 5；
Step 3.34: judge whether figure G is complete graph, if complete graph, exports S, S=S-{v}, jump procedure 3；As
Fruit is not complete graph, S=S-{v}, jump procedure 3.
Step 3.35:G '=G-NG [v], A ' is the adjacency matrix of G ', makes G=G ', A=A ', updates scale, redirects step
Rapid 3.22.
Further, described select containing minimum maximal independent set, comprising:
Select the minimum maximal independent set containing minimum some number.
The method of a kind of construction feature information bank that the present embodiment one provides, by third party's login user (such as microblogging)
Between interactive situation (as concern, comment, point praise, forwarding etc.) analysis, determine whether can reach between user (as user A closes
Note user B, then can be defined as B and can pass information to A), and then make the 3rd under a certain channel according to this standard
Side's login user forms a huge social networks (this network is complicated digraph), during application drawing is theoretical on that network
Tarjan algorithm, searches all very big strongly connected componentses in this digraph, and the information that simultaneously takes account of information in communication process declines
Subtract fast characteristic, theoretical in conjunction with vertex coloring, obtain the minimum maximal independent set in each very big strongly connected components, the most at last institute
There is the point in the minimum maximal independent set in very big strongly connected components to be merged into a set, be seed marketing customer group.
While technical scheme disclosed in the present embodiment one solves three problems of prior art, it is also considered that during propagation
Decay Rate:
1st, whether up to concept, utilization is that the mutual-action behavior of third party's social platform is weighed, because social flat
Whether playscript with stage directions body contains the theory of propagation, therefore up to can weigh whether message has effect spread between user.
2nd, Tarjan algorithm launches on digraph, and only require between user be indirectly up to, greatly reduce
Requirement to graph structure, has very strong practicality, therefore when customer group is screened, can effectively reduce original marketing customer group
Magnitude.
3rd, present patent application creative use dyeing theory, carries out specific aim battalion to the link being likely to result in information attenuation
Pin, makes information transmission be consolidated and strengthen, reaches message widely dissemination goal.
Embodiment two
The embodiment two that the present invention provides is with the difference of embodiment one, and the method step also includes:
Step 7, marketing personnel are according to concrete marketing scene (activity or project), and screening meets specifies number of users vertex set
Close composition target marketing customers to market.
Step 8, tracking marketing effectiveness and user feedback, constantly adjust variable, the power of parameter introducing model at each node
Weight, training, refreshing model so that the seed user screening of marketing is more and more accurate.
The method that the present embodiment two provides, outside reaching the accessible beneficial effect of embodiment one, is logging in as third party
When customer volume is very big, the concurrent magnitude that digraph calculates can be extremely huge, and therefore model refreshes category and should suitably reduce (such as one
Within individual month, refresh once).
Very big strongly connected components out after, the attenuation that information is transmitted must be considered, therefore need between spaced points
Signal to be arranged " booster ", i.e. must coordinate dyeing theory to carry out spaced points marketing, finally combine the basic business of user
Attribute, own user tag system, generation user's marketing message storehouse, and by existing precision marketing system is transformed,
Preferably complete data-interface docking, finally realize that precision marketing automates, reduce the artificial cost that triggers, lifting user's conversion
Rate, reaches profit maximization.
Embodiment three
With reference to Fig. 2, Fig. 2 shows the structured flowchart of an embodiment of the system of the present invention.
Acquisition module 21, for obtaining third party's accounts information and the interior network information issued of Preset Time；Implement
Function and processing mode see step 1.
Described acquisition module, comprising: acquiring unit 211, for obtaining the basis letter of third party's account when combining login
The network information that breath and third party's account are issued in Preset Time.Described third party's accounts information, comprising: third party's account
Basic business attribute, microblogging attribute and/or own user tag.
Analyze module 22, for analyzing the interactive information between third party's account, generate the information based on described interactive information
Transmission network；The function and the processing mode that implement see step 2.
Described analysis module, comprising: extraction unit 221, in the information that described acquisition module gathers, extracts the 3rd
Interactive information between all third party's accounts in the network information that side's account is issued in this Preset Time, based on described interactive letter
The direction of transfer of breath, determines oriented information transfer network.
Concrete, by analyzing concern, comment, forwarding between all third party's accounts, putting situations such as praising, determine that information can
The standard (as user A has paid close attention to user B, then can be defined as B and can pass information to A) reaching；Generate third party's account
Information transfer network (if user B can pass information to A, then generate a directed edge from user B to user A；If user
A can pass information to B, then also generate a directed edge from A to user B；Ultimately generate huge oriented information transmission
Figure).
Acquisition module 23, for obtaining based on all very big strongly connected components of described information transfer network and each is very big
All maximal independent sets in strongly connected components；The function and the processing mode that implement see step 3.
Described acquisition module 23, comprising: the first acquiring unit 231, is used for applying the whole information of Tarjan algorithm traversal to pass
Pass network, obtain all very big strongly connected components based on described information transfer network；Second acquisition unit 232, is used for combining
Vertex coloring is theoretical, obtains all maximal independent sets in each very big strongly connected components.
Described second acquisition unit 232, comprising: processing unit 2321, for by the digraph in very big connected component
Manage into non-directed graph；Filter element 2322, is one for filtering the limit of any point-to-point transmission in the non-directed graph in described processing unit
Limit.
Screening module 24, in all maximal independent sets from each very big strongly connected components, selects containing minimum
Maximal independent set；The function and the processing mode that implement see step 4.
Described screening module, comprising: screening unit 241, for selecting the minimum maximal independent set containing minimum some number.
Merge module 25, for the some duplicate removal that minimum maximal independent sets all in all very big strongly connected componentses are comprised
After, merge into a vertex set；The function and the processing mode that implement see step 5.
Generation module 26, is used for merging third party's accounts information and vertex set information, generates characteristic information storehouse.Concrete real
Existing function and processing mode see step 6.
The process being realized by the system of the present embodiment and function essentially correspond to the enforcement of the method shown in earlier figures 1
Example, principle and example, therefore not detailed part in the description of the present embodiment, may refer to the related description in previous embodiment,
This does not repeats.
The system that the embodiment of the present invention three provides, by the interactive situation between third party's login user (such as microblogging)
(as concern, comment, point praise, forwarding etc.) analysis, determine whether can reach between user and (as user A has paid close attention to user B, then may be used
A can be passed information to be defined as B), and then according to this standard, the third party's login user under a certain channel is formed
One huge social networks (this network is complicated digraph), the Tarjan algorithm during application drawing is theoretical on that network, look into
Look for all very big strongly connected componentses in this digraph, the fast characteristic of the information that simultaneously takes account of information attenuation in communication process, knot
Close vertex coloring theoretical, obtain the minimum maximal independent set in each very big strongly connected components, the most all very big strong continune
The point in minimum maximal independent set in branch is merged into a set, is seed marketing customer group.
While technical scheme disclosed in the present embodiment one solves three problems of prior art, it is also considered that during propagation
Decay Rate:
1st, whether up to concept, utilization is that the mutual-action behavior of third party's social platform is weighed, because social flat
Whether playscript with stage directions body contains the theory of propagation, therefore up to can weigh whether message has effect spread between user.
2nd, Tarjan algorithm launches on digraph, and only require between user be indirectly up to, greatly reduce
Requirement to graph structure, has very strong practicality, therefore when customer group is screened, can effectively reduce original marketing customer group
Magnitude.
3rd, present patent application creative use dyeing theory, carries out specific aim battalion to the link being likely to result in information attenuation
Pin, makes information transmission be consolidated and strengthen, reaches message widely dissemination goal.
Wherein, native system is when concrete application, and marketing personnel are according to concrete marketing scene (activity or project), and screening is full
Toe determine number of users vertex set constitute target marketing customers market.
System follows the tracks of marketing effectiveness and user feedback, constantly adjusts variable, the weight of parameter introducing model at each node,
Training, refreshing model so that the seed user screening of marketing is more and more accurate.
Wherein, when third party's login user amount is very big, the concurrent magnitude that digraph calculates can be extremely huge, therefore
The category that refreshes model should suitably reduce (as one month refreshes once).
Very big strongly connected components out after, the attenuation that information is transmitted must be considered, therefore need between spaced points
Signal to be arranged " booster ", i.e. must coordinate dyeing theory to carry out spaced points marketing, finally combine the basic business of user
Attribute, own user tag system, generation user's marketing message storehouse, and by existing precision marketing system is transformed,
Preferably complete data-interface docking, finally realize that precision marketing automates, reduce the artificial cost that triggers, lifting user's conversion
Rate, reaches profit maximization.
Embodiment four
Another aspect of the present invention additionally provides the device of a kind of construction feature information bank, including described system.Specifically,
This device includes but is not limited to the smart machine such as mobile phone, notebook computer.
The function implementing and processing mode see the concrete steps that embodiment of the method describes.
The process and the function that are realized by the built-in system of the device of the present embodiment essentially correspond to earlier figures 1 to Fig. 2
Shown method, the embodiment of system, principle and example, therefore not detailed part in the description of the present embodiment, may refer to aforementioned
Related description in embodiment, does not repeats at this.
The characteristic information storehouse that the present invention builds, can support the marketing of items of different types or activity, both more neatly
Save marketing resource, reduced again the construction cost of Marketing Model, can reach and once build the application of many scenes, the i.e. weight of model
Multiple utilization rate is high, is also very helpful the new user that draws over to one's side of operation aspect simultaneously, is finally reached maximum resource utilization, throwing
Enter the maximized purpose of output ratio.
It may be noted that according to the needs implemented, each step/parts described in this application can be split as more multistep
Suddenly/parts, it is possible to the part operation of two or more step/parts or step/parts is combined into new step/parts,
To realize the purpose of the present invention.
Above-mentioned the method according to the invention can realize in hardware, firmware, or is implemented as being storable in recording medium
Software in (such as CD ROM, RAM, floppy disk, hard disk or magneto-optic disk) or computer code, or it is implemented through network download
Original storage and the meter in local recording medium will be stored in long-range record medium or nonvolatile machine readable media
Calculation machine code, thus method described here can be stored in use all-purpose computer, application specific processor or able to programme or special
With the such software process on the record medium of hardware (such as ASIC or FPGA).It is appreciated that computer, processor, micro-
Processor controller or programmable hardware include can storing or receive software or computer code storage assembly (for example, RAM,
ROM, flash memory etc.), when described software or computer code are by computer, processor or hardware access and execution, it is achieved at this
The processing method describing.Additionally, when all-purpose computer accesses the code of the process being used for realizing being shown in which, the execution of code
All-purpose computer is converted to the special-purpose computer for performing the process being shown in which.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and any
Those familiar with the art, in the technical scope that the invention discloses, can readily occur in change or replace, should contain
Cover within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with described scope of the claims.
Claims (16)
1. the method for a construction feature information bank, it is characterised in that include:
The network information issued in step 1, acquisition third party's accounts information and Preset Time；
Step 2, the interactive information analyzed between third party's account, generate the information transfer network based on described interactive information；
Step 3, obtain based in all very big strongly connected components of described information transfer network and each very big strongly connected components
All maximal independent sets；
In step 4, all maximal independent sets from each very big strongly connected components, select containing minimum maximal independent set；
After step 5, the some duplicate removal that minimum maximal independent sets all in all very big strongly connected componentses are comprised, merge into a top
Point set；
Step 6, merging third party's accounts information and vertex set information.Generate characteristic information storehouse.
2. the method for claim 1, it is characterised in that in described step 1, obtain third party's accounts information and default when
The network information of interior issue, comprising:
The Back ground Information of third party's account is obtained and the network information that third party's account is issued in Preset Time during associating login.
3. method as claimed in claim 1 or 2, it is characterised in that in described step 2, analyze the interaction between third party's account
Information, generates the information transfer network based on described interactive information, comprising:
By third party's accounts information of acquisition in step one and the network information of issue in Preset Time, extract third party's account
Interactive information between all third party's accounts in the network information issued in this Preset Time, based on the biography of described interactive information
Pass direction, determine oriented information transfer network.
4. the method as described in one of claim 1-3, it is characterised in that in described step 3, obtains based on the transmission of described information
All maximal independent sets in all very big strongly connected components of network and each very big strongly connected components, specifically,
The whole information transfer network of application Tarjan algorithm traversal, obtains and greatly connects by force based on all of described information transfer network
The reduction of fractions to a common denominator is propped up；
Theoretical in conjunction with vertex coloring, obtain all maximal independent sets in each very big strongly connected components.
5. method as claimed in claim 4, it is characterised in that the whole information delivery network of described application Tarjan algorithm traversal
Network, obtains all very big strongly connected components based on described information transfer network, comprising:
Step 3.11, elder generation all initialize white the color of all of node, and stack are emptied；
Step 3.12, find a white node；
Step 3.13, give one timestamp of node, in node stacking, and node is labeled as grey, makes Low [i]=DFN
[i]；
Step 3.14, traversing nodes i each edge (i, j), if color [j] be white, just to node i repetition step 3.2～
3.5, and make Low [i]=min (Low [j], low [i]) if. color [j] is grey, make Low [i]=min (Low [i],
DFN[j])；If black is left intact.
Step 3.15, the color of node is changed into black, if Low [i]=DFN [i], just the unit from stack top to node i
Element ejects；
Step 3.16, repeat step 3.2, to there is no White vertex；
Wherein, color [i] is the state for indicating node i: white refers to also be less than search, grey just searched, black
It is disposed.
6. the method as described in claim 4 or 5, it is characterised in that described combination vertex coloring is theoretical, obtains each greatly strong
All maximal independent sets in connected component, comprising:
Digraph in very big connected component is processed into non-directed graph；
Process any point-to-point transmission in above-mentioned non-directed graph and there is polygon situation, retain a limit, remaining edge contract.
7. method as claimed in claim 6, it is characterised in that described combination vertex coloring is theoretical, obtains each and greatly connects by force
The reduction of fractions to a common denominator prop up in all maximal independent sets, specifically include:
Step 3.21: the adjacency matrix A of input figure G, initializes size scale of A, independent sets S=H.
Step 3.22: if scale≤1, output S and last vertex v ', would recover last G and A, update scale；Otherwise
Jump procedure 3；
Step 3.33: one by one to all v ∈ V (G), S=S+{v}, if the point in v and all G is adjacent, jump procedure 4；Otherwise
Jump procedure 5；
Step 3.34: judge whether figure G is complete graph, if complete graph, exports S, S=S-{ v}, jump procedure 3；If
It not complete graph, S=S-{v}, jump procedure 3.
Step 3.35:G '=G-NG [v], A ' is the adjacency matrix of G ', makes G=G ', A=A ', updates scale, jump procedure 2.
8. the method for claim 1, it is characterised in that described select containing minimum maximal independent set, comprising:
Select the minimum maximal independent set containing minimum some number.
9. method as claimed in claim 6, it is characterised in that described third party's accounts information, comprising:
The basic business attribute of third party's account, microblogging attribute and/or own user tag.
10. the system of a construction feature information bank, it is characterised in that include:
Acquisition module, for obtaining third party's accounts information and the interior network information issued of Preset Time；
Analyze module, for analyzing the interactive information between third party's account, generate the information delivery network based on described interactive information
Network；
Acquisition module, is used for obtaining all very big strongly connected components based on described information transfer network and each very big strong continune
All maximal independent sets in branch；
Screening module, in all maximal independent sets from each very big strongly connected components, select containing minimum greatly solely
Vertical collection；
Merge module, after being used for the some duplicate removal that minimum maximal independent sets all in all very big strongly connected componentses are comprised, merge
It is a vertex set；
Generation module, is used for merging third party's accounts information and vertex set information, generates characteristic information storehouse.
11. systems as claimed in claim 10, it is characterised in that described acquisition module, comprising:
Acquiring unit, is used for when combining login obtaining the Back ground Information of third party's account and third party's account in Preset Time
The network information issued.
12. systems as described in claim 10 or 11, it is characterised in that described analysis module, comprising:
Extraction unit, in the information that described acquisition module gathers, extracts third party's account and issues in this Preset Time
The network information in interactive information between all third party's accounts, based on the direction of transfer of described interactive information, determine oriented letter
Breath transmission network.
13. systems as described in one of claim 10-12, it is characterised in that described acquisition module, comprising:
First acquiring unit, is used for applying the whole information transfer network of Tarjan algorithm traversal, obtains based on the transmission of described information
The all very big strongly connected components of network；
Second acquisition unit, is used for combining vertex coloring theoretical, obtains all very big independence in each very big strongly connected components
Collection.
14. systems as claimed in claim 13, it is characterised in that described second acquisition unit, comprising:
Processing unit, for being processed into non-directed graph by the digraph in very big connected component；
Filter element, for by containing polygon situation in non-directed graph in processing unit, being processed as a limit.
15. systems as claimed in claim 10, it is characterised in that described screening module, comprising:
Screening unit, for selecting the minimum maximal independent set containing minimum some number.
The device of 16. 1 kinds of construction feature information banks, it is characterised in that include the system as described in one of claim 10-15.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610374915.8A CN106067139A (en) | 2016-05-31 | 2016-05-31 | A kind of mthods, systems and devices of construction feature database |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610374915.8A CN106067139A (en) | 2016-05-31 | 2016-05-31 | A kind of mthods, systems and devices of construction feature database |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106067139A true CN106067139A (en) | 2016-11-02 |
Family
ID=57420999
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610374915.8A Pending CN106067139A (en) | 2016-05-31 | 2016-05-31 | A kind of mthods, systems and devices of construction feature database |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106067139A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107993156A (en) * | 2017-11-28 | 2018-05-04 | 中山大学 | A kind of community discovery method based on social networks digraph |
| CN113742538A (en) * | 2021-07-16 | 2021-12-03 | 深圳云天励飞技术股份有限公司 | Business analysis method and device based on graph hierarchy, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103825762A (en) * | 2014-02-26 | 2014-05-28 | 重庆邮电大学 | Traffic grooming and distinction importance protection method based on clustering |
| CN103945508A (en) * | 2014-02-24 | 2014-07-23 | 浙江理工大学 | Wireless-sensing-network topology construction method based on probability comparison |
| CN105095419A (en) * | 2015-07-15 | 2015-11-25 | 哈尔滨工程大学 | Method for maximizing influence of information to specific type of weibo users |
| CN105574649A (en) * | 2015-12-10 | 2016-05-11 | 西安交通大学 | Taxpayer tax evasion suspicion group detection method based on multi-stage MapReduce model |
-
2016
- 2016-05-31 CN CN201610374915.8A patent/CN106067139A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103945508A (en) * | 2014-02-24 | 2014-07-23 | 浙江理工大学 | Wireless-sensing-network topology construction method based on probability comparison |
| CN103825762A (en) * | 2014-02-26 | 2014-05-28 | 重庆邮电大学 | Traffic grooming and distinction importance protection method based on clustering |
| CN105095419A (en) * | 2015-07-15 | 2015-11-25 | 哈尔滨工程大学 | Method for maximizing influence of information to specific type of weibo users |
| CN105574649A (en) * | 2015-12-10 | 2016-05-11 | 西安交通大学 | Taxpayer tax evasion suspicion group detection method based on multi-stage MapReduce model |
Non-Patent Citations (3)
| Title |
|---|
| JI GUI-LIN 等: "An Approximation Algorithm for the Influential Nodes Selection Problem in Social Network", 《新疆大学学报(自然科学版)》 * |
| 冀桂琳: "无线传感网络中的路由选择及社会网络中影响集选择的近似算法", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 杨铀 等: "求解图的最大独立集的一种算法", 《电脑开发与应用》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107993156A (en) * | 2017-11-28 | 2018-05-04 | 中山大学 | A kind of community discovery method based on social networks digraph |
| CN107993156B (en) * | 2017-11-28 | 2021-06-22 | 中山大学 | Social network directed graph-based community discovery method |
| CN113742538A (en) * | 2021-07-16 | 2021-12-03 | 深圳云天励飞技术股份有限公司 | Business analysis method and device based on graph hierarchy, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105718490A (en) | Method and device for updating classifying model | |
| CN108288208A (en) | The displaying object of image content-based determines method, apparatus, medium and equipment | |
| CN106603690A (en) | Data analysis device, data analysis processing system and data analysis method | |
| CN107592296A (en) | The recognition methods of rubbish account and device | |
| CN104077723A (en) | Social network recommending system and social network recommending method | |
| CN109951846A (en) | Wireless network recognition methods, device, storage medium and computer equipment | |
| CN110297990A (en) | The associated detecting method and system of crowdsourcing marketing microblogging and waterborne troops | |
| CN109062807B (en) | Method and device for testing application program, storage medium and electronic device | |
| CN112100450A (en) | Graph calculation data segmentation method, terminal device and storage medium | |
| CN110232178A (en) | Report generation method and device | |
| CN113282623A (en) | Data processing method and device | |
| CN103885977B (en) | A kind of classification method of web data, device and system | |
| Hu et al. | A new algorithm CNM-Centrality of detecting communities based on node centrality | |
| CN109978575A (en) | A kind of method and device excavated customer flow and manage scene | |
| CN106067139A (en) | A kind of mthods, systems and devices of construction feature database | |
| CN106406985A (en) | A distributed computing frame and a distributed computing method | |
| CN113569162A (en) | Data processing method, device, equipment and storage medium | |
| CN110532266A (en) | A kind of method and apparatus of data processing | |
| CN110532448A (en) | Document Classification Method, device, equipment and storage medium neural network based | |
| CN114201663B (en) | Group customer mining method and device based on label propagation and electronic equipment | |
| CN106156256A (en) | A kind of user profile classification transmitting method and system | |
| CN109471935A (en) | A kind of questionnaire survey object determines method, apparatus, electronic equipment and storage medium | |
| De Silva et al. | Harnessing mobile pervasive computing to enhance livelihood processes: Farmer response to a mobile agriculture information system | |
| CN107679168A (en) | A kind of targeted website content acquisition method based on java platforms | |
| TW201828193A (en) | Method and device for dividing user group in which core user groups are first identified according to a user node diagram and then target user groups are defined among the core user groups |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161102 |
|
| RJ01 | Rejection of invention patent application after publication |