CN110264364A - A kind of recommended method of investor - Google Patents
A kind of recommended method of investor Download PDFInfo
- Publication number
- CN110264364A CN110264364A CN201910359691.7A CN201910359691A CN110264364A CN 110264364 A CN110264364 A CN 110264364A CN 201910359691 A CN201910359691 A CN 201910359691A CN 110264364 A CN110264364 A CN 110264364A
- Authority
- CN
- China
- Prior art keywords
- node
- investor
- label
- investment
- score value
- 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.)
- Granted
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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/06—Asset management; Financial planning or analysis
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- Operations Research (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Accounting & Taxation (AREA)
- Theoretical Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Technology Law (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention discloses the recommended methods of investor a kind of, for improving the accuracy rate for being directed to start-up company and recommending investor.In the recommended method, history investment event information and label information are obtained first from history investment data library;Then event information and label information are invested according to history and generates investment network, investment network includes: M investor's node, N number of by investor's node, P label node etc.;Successive ignition executes following diffusion process, diffusion process includes: that the first kind node score value obtained in current diffusing step invested in network is distributed to the first neighbor node being connected with first kind node, and obtained score value is used in next diffusing step by the first neighbor node;When having executed diffusing step and having reached optimal diffusion step number, terminate diffusion process;M investor's node is sorted from large to small according to the final score value being respectively provided with, recommends at least one investor's node to targeted company to be recommended according to ranking results.
Description
Technical field
The present invention relates to field of computer technology, and in particular to a kind of recommended method of investor.
Background technique
Current internet is grown rapidly, and user enters the epoch of information explosion, and the presentation of massive information is difficult people
It was found that useful information, making correct decisions.
Recommender system is the effective tool for solving problem of information overload.In the commodity such as film, books, music or service neck
Domain, recommender system have played important function, provide accurate personalized recommendation for user, and lead in financial investment
Domain, recommendation problem, which but rarely has, to be related to.
Compared to for the purchase or use data of commodity or service, the investment relation in financial investment field is more dilute
It dredges, the proposed algorithm that the prior art provides is usually applicable only to the recommendation of commodity or service, and is not suitable for financial investment neck
Investment recommendation in domain, thus the recommended models of prior art offer and algorithm are difficult to reach higher recommendation accuracy rate.
And the start-up company for finding investment, it is especially most of never to obtain for the company of investment, due to experience
Deficiency, generally requires to take a significant amount of time with energy that find may be to the interested investor in its field.It realizes to this kind of public affairs
The recommendation of department is referred to as cold start-up problem, i.e., situation unknown by investment situation in target object history in recommender system research
Under, it is difficult to realize personalized recommendation using traditional recommended method for it.
Therefore, it is badly in need of being directed to investor's recommended method of start-up company, to overcome recommendation existing for traditional proposed algorithm
The low problem of accuracy rate.
Summary of the invention
The purpose of the present invention is to provide the recommended methods of investor a kind of, recommend investment for start-up company for improving
The accuracy rate of side.
In order to achieve the above object, the present invention uses such following technical solution:
The present invention provides the recommended method of investor a kind of, comprising:
History investment event information and label information are obtained from history investment data library, the history invests event information
X-th of the investor and N number of i-th by investor being used to indicate in M investor are invested by the history between investor
Relationship, the label information include described N number of by the corresponding label of investor, N number of each quilt by investor
Investor includes at least one label, and the M, the N, the x and the i indicate positive integer, and the x is less than or equal to institute
M is stated, the i is less than or equal to the N;
Event information is invested according to the history and the label information generates investment network, and the investment network includes: M
It is a investor's node, N number of by investor's node, P label node, investor's node and by the first kind between investor's node
Type connects when the Second Type between investor's node and label node connects;Wherein, the M investor node and the M
A investor corresponds, it is described it is N number of by investor's node with it is described it is N number of corresponded by investor, the P label node
With it is described it is N number of corresponded by the corresponding all not same labels of investor, the first kind connects side and indicates x-th of investor
Node and i-th are established between investor's node investment relation, and the Second Type connects side and indicates x-th of investor's node
Establishing between label node corresponding with i-th of label included by investor's node has investment relation;
Successive ignition executes following diffusion process, and the diffusion process includes: by the first kind in the investment network
Type node score value obtained in current diffusing step distributes to the first neighbor node being connected with the first kind node,
Obtained score value is used in next diffusing step by first neighbor node, the target of recommendation to be supplied in the investment network
The corresponding label node of all labels included by company has an initial value in the diffusion process, in the investment network
The M investor node, it is described it is N number of by investor's node, the P label node except included by the targeted company
Other label nodes other than the corresponding label node of all labels do not have initial value in the diffusion process;Wherein,
When the first kind node is the label node, first neighbor node includes: to connect side by the Second Type
The investor's node being connect with the label node;When the first kind node is investor's node, described first
Neighbor node includes: to connect the label node that side is connect with investor's node by the Second Type, and pass through described
The first kind connect that side connect with investor's node by investor's node;When the first kind node is described invested
Fang Jiedian and when the label node, first neighbor node includes: to connect side by the first kind to be invested with described
Investor's node of Fang Jiedian connection, and the investor that side is connect with the label node is connected by the Second Type and is saved
Point;
Determine the diffusion process has executed whether diffusing step has reached predetermined optimal diffusion step number out,
And it is described executed diffusing step and have reached the optimal diffusion step number when, terminate the diffusion process;
Obtain final point that M investor's node at the end of the diffusion process in the investment network is respectively provided with
Value, the M investor node is sorted from large to small according to the final score value being respectively provided with, according to ranking results to the mesh
At least one investor's node is recommended by mark company.
After adopting the above technical scheme, technical solution provided by the invention will have the following advantages:
Investment network is generated in the embodiment of the present application, which includes: M investor's node, N number of invested
Fang Jiedian, P label nodes, investor's node and by the first kind between investor's node connect side and investor's node with
Second Type between label node connects side.Diffusion process can be executed with successive ignition in the embodiment of the present invention, so that label
The initial value that node has in diffusion process can constantly be spread in investment network, because three in investment network
The company side of partial node is connected according to real connection, and investor's node is ined succession (being invested by investor's node of investing
Company, abbreviation company) and the label node invested.Therefore, score value is spread along even side, will be the section for having indirect association
Point is all diffused into.For example, initial value on the label node contained by targeted company, in first step diffusion, invested these
Investor's node of label can all receive certain score value, and investing the more investor of these labels just will receive more points
Value.Then again value diffusing to company's node and label node, the company's node for obtaining score value is these investor's nodes
The company that the investor of label contained by targeted company invested was invested, the label node for obtaining score value is to invest target public affairs
Take charge of the label that the investor of contained label invested.That is, directly generating the investor of relationship with indirect and these labels
Node, company's node, label node can all be successively received score value, and then these score values are concentrated in investor's node set.
Therefore this weak relationship is constantly reinforced by diffusion, is finally reflected in the score value of investor's node, this score value is bigger, just says
A possibility that relationship of the bright investor and targeted company is stronger, and investment relation is generated between investor and targeted company is bigger,
The high investor of score value is recommended into targeted company.Therefore the accuracy rate for recommending investor for start-up company can be improved.
Detailed description of the invention
Fig. 1 provides a kind of process blocks schematic diagram of the recommended method of investor for the embodiment of the present invention;
Fig. 2 is the composed structure schematic diagram of investment network provided in an embodiment of the present invention;
Fig. 3 is connection relationship diagram between the node provided in an embodiment of the present invention invested in network;
Fig. 4 a is the schematic diagram of the initial value setting in investment network provided in an embodiment of the present invention;
Fig. 4 b is the schematic diagram of the first step diffusion in investment network provided in an embodiment of the present invention;
Fig. 4 c is the schematic diagram of the second step diffusion in investment network provided in an embodiment of the present invention;
Fig. 4 d is the schematic diagram of the third step diffusion in investment network provided in an embodiment of the present invention.
Specific embodiment
The embodiment of the invention provides the recommended methods of investor a kind of, recommend investor for start-up company for improving
Accuracy rate.
It is described in detail separately below.
One embodiment of the recommended method of investor of the present invention can be applied to recommend and the start-up company to start-up company
The high investor of matching degree.In subsequent embodiment of the invention, investor's node may be simply referred to as investor, by investor's node
It can be referred to as invested company (or referred to as company), target to be recommended can be referred to as target public affairs by investor's node
Department.It, can the relationship of investor and targeted company is strong and weak, generation by the diffusion process of subsequent descriptions in the embodiment of the present invention
A possibility that investment relation, size embodied by the score value that investor obtains, and the high investor of score value is finally recommended mesh
Mark company.Therefore the accuracy rate for recommending investor for start-up company can be improved.
Refering to Figure 1, the recommended method of investor provided by the invention, may include steps of:
10, history investment event information and label information are obtained from history investment data library, history invests event information
X-th of the investor and N number of i-th by investor being used to indicate in M investor are invested by the history between investor
Relationship.Label information includes N number of by the corresponding label of investor, and N number of by each of investor includes extremely by investor
A few label, M, N, x and i indicate positive integer, and x is less than or equal to M, and i is less than or equal to N.
It is pre-configured with history investment data library in the embodiment of the present invention, history investment is stored in the history investment data library
Event information and label information, therefore history investment event information and label letter can be got from the history investment data library
Breath.History investment event information is referred to as history investment relation data in the embodiment of the present invention, to indicate in M investor
X-th of investor and N number of i-th by investor by the history investment relation between investor.The value of M and N can be with
It is determined according to concrete scene.Label information be used to indicate it is N number of by the corresponding label of investor, it is each to include by investor
At least one label, for the content protion of label, herein without limitation.In addition, not limiting, the embodiment of the present invention
Middle M, N, x and i indicate positive integer, and x is less than or equal to M, and i is less than or equal to N.
In some embodiments of the invention, history investment event information includes: the mark of x-th of investor, i-th of quilt
Mark, the investment record data of investor.
Be illustrated below, obtain a period of time in history investment relation data, including all investment event informations with
And the label information of invested company.Wherein, investment event information includes the investor and invested company, throwing that the event is related to
Providing time, investment amount, the information such as equity accounting, such as format can be investor ID, invested company ID, investment time,
Investment amount, other investing tips such as equity accounting, the category information can be obtained from investment data provider, such as can be from IT
Orange, CVSource made a basket and obtained at the investment datas providers such as data, clear section's data, and some investors or company can also disclose
Its recent investment situation.
In the embodiment of the present invention, label is made of some keywords, such as the format of label can be company ID: label 1,
Label 2 ... ... label n.Label reflects industry field and the main business of company, and the label of company can be from above data service
The inquiry of the channel search such as quotient obtains, and can also voluntarily construct tag library, according to the basic introduction of company, product information, business row
For contents extractions key messages such as, news dynamic and Public Appraisals, label is set for it.
20, event information is invested according to history and label information generates investment network, investment network includes: M investor
Node, it is N number of by investor's node, P label node, investor's node and by the first kind between investor's node connect side,
Second Type between investor's node and label node connects side.
Wherein, M investor's node is corresponded with M investor, N number of by investor's node and N number of by investor one
One is corresponding, and with N number of by the corresponding all not same labels one-to-one correspondence of investor, the first kind connects side and indicates P label node
X-th of investor's node and i-th are established between investor's node investment relation, and Second Type connects side and indicates x-th of throwing
Establishing between capital's node label node corresponding with i-th of label included by investor's node has investment relation.
In some embodiments of the invention, after getting history investment event information and label information, Ke Yigen
The data content recorded in event information and label information, which is invested, according to the history extracts investor, invested company, label etc.
Information, with a node in each investor's corresponding network, with a node in each invested company's corresponding network, with
A node in each label corresponding network.
As shown in Fig. 2, for a kind of exemplary construction of investment network.If it is determined that M investor's node, N number of by investor
Node, P label node, then with M investor's node, it is N number of investment network is generated by investor's node, P label node,
It invests in addition to including above-mentioned three kinds of nodes in network, further includes the connection relationship between node.Specifically, M investor's node
Centre one in investment network arranges, N number of to be located at M investor's node two by investor's node, P label node
Side.For example, N number of left side for being located at M investor's node by investor's node, P label node is located at M investor's node
Right side.It further include investor's node and by the between investor's node in addition to including above-mentioned three kinds of nodes in investment network
One type connects when the Second Type between investor's node and label node connects, which specific investor's node and which
There is even side by establishing between investor's node, establishing between which investor's node and which label node has even side, specifically needs
It to be determined according to the data extracted from history investment data library.
In embodiments of the present invention, connect side in Fig. 2 with the two types that dotted line comes in the schematically illustrate embodiment of the present invention.Its
In, the first kind connects side and indicates x-th of investor's node and established between investor's node for i-th have investment relation, and second
Type connects side and indicates between x-th of investor's node label node corresponding with i-th of label included by investor's node
Foundation has investment relation.For example, the first kind, which connects side, refers to investor's node and by the investment relation of investor's node, if throwing
Capital's node and there are investment relations by investor's node is then established the first kind between two nodes of investment network and is connected
Side, if investor's node and by investor's node be not present investment relation, investment network two nodes between do not establish
The first kind connects side.Similarly, Second Type, which connects to establish between side instruction investor's node and label node, investment relation, if
Investor's node invested the company containing certain label, then established Second Type between two nodes of investment network and connect side, if
Investor's node did not invest the company containing some label, then did not established Second Type between two nodes of investment network
Lian Bian.
In some embodiments of the present application, it includes: the mark of x-th of investor, i-th of quilt that history, which invests event information,
Mark, the investment record data of investor;
I-th by least one label that investor's node includes for reflect i-th of industry field by investor with
Main business.
Wherein, tag reactant is accordingly by the industry field of investor's node and main business.For example, certain company labels
Are as follows: tourism, tourism synthesis service, hotel, visa service, trip ticketing service etc..
In some embodiments of the present application, step 20 invests event information according to history and label information generates investment net
After network, method provided by the embodiments of the present application further includes following steps:
Determine that the first kind connects the weight on side according to investor's node and by the investment record data between investor's node;
According to investor's node and by between investor's node investment relation, by the institute between investor's node and label
Category relationship, investor's node and by between investor's node investment record data determine that Second Type connects the weight on side.
Wherein, investment record data may include investment amount, other investing tips such as equity accounting, according to the investor
Node and by between investor's node investment record data determine that the first kind connects the weight on side, similarly, can also establish
Second Type connects the weight on side.When the first kind while and in the case that when Second Type, all has weight, be based on subsequent
Score value when investment network is diffused requires to be diffused according to the size of the weight, subsequent by detailed illustration.
Available three kinds of data acquisition systems in embodiments of the present invention: investor set I, invested company set C and mark
Sign set T and two kinds of data relationships: the belonging relation between investment relation and company-label between investor-company.
In proof analysis, investor has apparent investment preference to specific area and label, therefore the embodiment of the present invention is to be thrown
Money company connects investor and label as bridge, to use this preference information in recommendation.
Fig. 3 is connection relationship diagram between the node provided in an embodiment of the present invention invested in network.According to acquisition
Data construct company-investor-label three parts figure G (C, I, T), building rule are as follows: each investor, invested company, mark
Label indicate that investment relation is indicated with even side with node, and on the both sides of three parts figure, they are invested with corresponding for company and label
Fang Xianglian, but the not company of foundation side between company and label, if investor IxInvested company Ci, then in IxAnd CiBetween build
Li Lianbian;If company C simultaneouslyiInclude label TaAnd Tb, then in IxAnd Ta、TbBetween the company of foundation side, gather between internal node
Do not connect side, side is not connected between Ji Ge company, side is not connected between each label, does not also connect side between each investor.Company-
The weight for connecting side between investor is quantified by actual numerical values such as investment amount, equity accountings, and the power on side is connected between investor-label
Weight is invested the numerical value such as company's number containing the label and investment amount, the equity accounting of each investment by investor and is determined.
For example, respectively I1, I2, I3, I4, invested company, which gathers, 6 if investor's set includes 4 investors
A company C1~C6, tag set include 4 labels, are T1~T4;I1 invested C2, C4, and I2 invested C3, C5, I3 investment
C4, C5, C6 are crossed, I4 invested C5;The label of C1 is T1, T2, and the label of C2 is T1, T2, T4, and the label of C3 is T1, T2, T4,
The label of C4 is T3, T4, and the label of C5 is T4, and the label of C6 is T4, and the embodiment of the present invention uses a simple weight here
Rule is as an example, all regard the side right weight between investor and company as 1, by company's side right weight between investor and label
It is set as investor and invests the number containing the label company, its company-throwing as shown in Figure 3 can be established according to investment data as above
Capital-label three parts figure, the digital representation in figure on side are great small.Since targeted company to be recommended does not have any thrown
Money experience, so it will not be connected with any investor in the three parts figure of building, such as the company C1 in Fig. 3.
30, successive ignition executes following diffusion process, and diffusion process includes: that will invest the first kind node in network
The score value obtained in current diffusing step distributes to the first neighbor node being connected with first kind node, the first neighbours section
Obtained score value is used in next diffusing step by point.
Wherein, the corresponding label node of all labels included by the targeted company of recommendation to be supplied is invested in network to expand
There is initial value during dissipating, invest M investor's node in network, N number of removed by investor's node, P label node
Other label nodes other than the corresponding label node of all labels included by targeted company do not have just in diffusion process
Beginning score value;Wherein, when first kind node is label node, the first neighbor node includes: to connect Bian Yubiao by Second Type
Sign investor's node of node connection;When first kind node is investor's node, the first neighbor node includes: by second
The label node of type Lian Bianyu investor's node connection, and thrown by what first kind Lian Bianyu investor's node connected
Capital's node;When first kind node is by investor's node and label node, the first neighbor node includes: to pass through the first kind
Type connects investor's node that side is connect with by investor's node, and the investment connected by Second Type Lian Bianyu label node
Fang Jiedian.
In embodiments of the present invention, after generating investment network by step 20, following expansion can be executed with successive ignition
The process of dissipating includes following two processes in each diffusion process: the first kind node invested in network is expanded currently
Score value obtained in stroll is rapid distributes to the first neighbor node being connected with first kind node, and the first neighbor node will obtain
Score value in next diffusing step.After i.e. the first neighbor node obtains score value, using first neighbor node as first
Type node re-starts diffusion, that is, re-executes and obtain the first kind node invested in network in current diffusing step
Score value distribute to the first neighbor node being connected with first kind node.
It should be noted that in the embodiment of the present invention, when first kind node is by investor's node and label node,
First neighbor node includes: to connect investor's node that side is connect with by investor's node by the first kind, and pass through second
Investor's node of type Lian Bianyu label node connection.Both sides can be by even side toward intermediate diffusion, and investor can be with
Left end perhaps right end connection or is all connected with left and right ends to be connected with the left side simultaneously and can obtain the score value that the left side is spread,
The score value that the right diffusion comes can be obtained by being connected with the right.
In embodiments of the present invention, when the first kind while and in the case that when Second Type, all has weight, step
The first kind node score value obtained in current diffusing step invested in network being distributed to and first kind section in 30
The first neighbor node that point is connected, comprising:
First kind node score value obtained in current diffusing step in network will be invested, connects side according to the first kind
Weight and Second Type connect the weight on side, be averagely allocated to the first neighbor node.
Wherein, the first kind while and in the case that when Second Type, all has weight, carried out based on investment network
Score value when diffusion requires to be diffused according to the size of the weight, and the first kind node invested in network is expanded currently
Take a walk it is rapid obtained in score value, according to the first kind connect while weight and Second Type connect while weight, be averagely allocated to first
Neighbor node.
In some embodiments of the invention, the diffusion process in step 30, can specifically include following steps:
Step a: it in the initial step before starting diffusion, is wrapped by the targeted company of recommendation to be supplied in investment network
Initial value is arranged in the corresponding label node of all labels included, in investment network M investor's node, N number of invested
Other label nodes in addition to the corresponding label node of Fang Jiedian, P label nodes all labels included by the targeted company
It is 0 that score value, which is arranged,;
Step b: when first kind node is label node, the initial value of label node is connected into side by Second Type
Investor's node is distributed to according to weighted average;After step b execution, successively iteration executes following step c, step d, directly
Stop iteration when reaching optimal diffusion step number;
Step c: when first kind node is investor's node, investor's node is led to by the score value that step b is obtained
It crosses Second Type and connects when, the first kind connects and distribute to label node according to weighted average and by investor's node;
Step d: when first kind node is by investor's node and label node, step will be passed through by investor's node
The score value that the score value and label node that c is obtained are obtained by step c is connected when, Second Type connects by the first kind according to power
It is averagely allocated to investor's node again, investor's node will connect point respectively obtained when, Second Type connects by the first kind
Value is superimposed, and the score value triggering obtained using investor's node by step d executes step c.
Wherein, it in abovementioned steps c and step d, needs to be executed according to multiplicating iteration.In order to recommend to targeted company
Eligible investment side, in investment network (company-investor-label three parts figure), the embodiment of the present invention assigns tag set
In the certain initial value of each label, such as the label that is connected with targeted company obtains the initial value of 1 unit, other marks
Label initial value is set as 0.Next, these score values will be diffused in three parts figure by even side.In the diffusion of every step,
Each its score value is distributed to connected neighbours according to weighted average and saved by the node containing score value (investor, company or label)
The score value of acquisition is added to obtain the new score value of oneself by point, each node.It is spread by odd number step, score value will fall in investor's collection
In conjunction.Score value calculating after every step diffusion is shown below:
It should be noted that above-mentioned diffusion process can be it is two-way, in the diffusion of every step, the node containing score value will point
Value distributes to connected neighbor node, and after dispensing, itself score value is reset, then all nodes by oneself it is new receive (from
What other node was assigned to) score value be added and be set as oneself new score value.
For example, in the initial stage, initial value is in label node set, so containing score value in first step diffusion
Node just there was only label node contained by targeted company, the score value of oneself distributed to connected investment by these label nodes
Fang Jiedian;Followed by second step spread, at this time the node containing score value is several investor's nodes, they by oneself
Score value distributes to connected company's node, while can also distribute to connected label node;Then third step is spread, and contains score value
Node be several company's nodes and label node, the score value of oneself is distributed to connected investor's node by company's node, mark
The score value of oneself is also allocated to connected investor's node by label node;Following 4th step diffusion, at this time containing score value
Node is several investor's nodes, and the score value of oneself is distributed to connected company's node and label node by they, and so on.
So when score value is dispensed from investor's node (intermediate node set), be spread simultaneously to both sides, the reason is that
Investor's node was not only connected with company node, but also was connected with label node.And score value subsidiary company node or label node diffuse out
Uni-directional diffusion when going because company's node only with investor's node have even side, and label node also only and investor
Node has even side.
It is illustrated below, f (Iy) it is node IyThrough the score value before step diffusion, f ' (Iy) it is score value after diffusion, A is
Investor-company's adjacency matrix, if investor IyInvested company Cx, then A (Iy,Cx)=Wyx, WyxAccording to investment amount, equity
The quantization of the actual numerical values such as accounting, is otherwise 0;A ' is investor-label adjacency matrix, if investor IyInvested T containing labelz's
Company, then A ' (Iy,Tz)=Wyz, WyzInvestment amount of number and each investment by investor's investment containing the label company,
The numerical value such as equity accounting determine, are otherwise 0;kIy→CIndicate investor IyWith the sum of the weight of all company Lian Bian, kIy→TInvestment
Square IyConnect the sum of the weight on side, k with all labelsCxAnd kTzRespectively indicate company CxWith label TzConnect the power on side with all investors
The sum of weight;λ is adjustment parameter, and the score value ratio spread to the left and right sides for adjusting investor's node, λ can be from historical data
Training obtains optimal value.
It as shown in fig. 4 a, is the schematic diagram of the initial value setting in investment network provided in an embodiment of the present invention, Fig. 4 b
For the schematic diagram of the first step diffusion in investment network provided in an embodiment of the present invention, Fig. 4 c is throwing provided in an embodiment of the present invention
The schematic diagram of the second step diffusion in network is provided, Fig. 4 d is the third step diffusion in investment network provided in an embodiment of the present invention
Schematic diagram.
Targeted company is C1 at this time, is the initial value that 1 unit is arranged in label T1 and T2 contained by C1, as shown in fig. 4 a, so
After start diffusion process.In first step diffusion, the score value of oneself is averagely allocated to connected throwing according to weight by each label
The score value of acquisition is added by capital, investor, obtains the score value of oneself, as shown in Figure 4 b;In second step diffusion, each investment
Whole score values that oneself is obtained are given connected company and label according to weighted average by side, wherein gathering distribution total score to company
λ times of value, to (the 1- λ) times of tag set distribution total score, company and label have obtained corresponding score value at this time, such as Fig. 4 c institute
Show;In third step diffusion, each company and label are respectively distributed to the score value that oneself is obtained by weighted average again to be connected
Investor, investor are added the score value that both sides distribution comes to obtain new score value, as shown in figure 4d.According to same rule, the present invention
Embodiment can continue multistep diffusion, and score value is taken to fall in the case where investor gathers (spreading by odd number step) to expand
Dissipating bind beam spot.Optimal diffusion step number can be trained from historical data and be obtained.
In some embodiments of the invention, λ times of the score value of investor's node is used to distribute to and be connected by the first kind
Side connect with investor's node by investor's node, (the 1- λ) of the score value of investor's node is times for distributing to by second
The label node of type Lian Bianyu investor's node connection, λ are the numerical value greater than 0 and less than 1, and λ is predetermined optimal double
To diffusion specific gravity.
Assuming that the value of each investor's node has " 1 part ", this " 1 part " score value, which can demarcate, to be come, and is distributed respectively to the left and right.Than
As distributed " 0.3 part " toward the left side, toward the right distribution " 0.7 part ".It here is not direct average mark (left " 0.5 part ", right " 0.5 part ")
The reason of allow for, in three parts figure, the ratio recanalization toward the right and left diffusion can have an impact to recommendation effect, therefore draw
Enter parameter lambda.
Detailed, in first step diffusion, the score value of oneself is averagely allocated to connected by each label node according to weight
Investor's node, the score value of acquisition is added by investor's node, obtain the score value of oneself, as shown in Figure 4 b:
T1 spread toScore value, T2 spread toScore value;
T1 spread toScore value, T2 spread toScore value.
In second step diffusion, whole score values that each investor's node obtains oneself are given connected according to weighted average
Company's node and label node, wherein distributing the 1- λ of total score to label node to company's node distribution λ times of total score
Times, company's node and label node have obtained corresponding score value at this time, as illustrated in fig. 4 c:
I1 spread toScore value;
I2 spread toScore value;
I1 spread toScore value;
I2 spread toScore value;
I1 spread toScore value, I2 spread toScore value;
I1 spread toScore value, I2 spread to
Score value;
I1 spread toScore value;
I1 spread toScore value, I2 spread to
Score value.
In third step diffusion, the score value that oneself is obtained is pressed weighted average respectively by each company's node and label node again
Connected investor is distributed to, investor's node is added the score value that both sides distribution comes to obtain new score value, as shown in figure 4d:
I1 is leftC2 spread toScore value, C4 spread toScore value;
I2 is left:C3 spread toScore value, C5 spread toScore value;
I3 is left:C4 spread toScore value, C5 spread toScore value;
I4 is left:C5 spread toScore value;
I1 is right:T1 spread toScore value, T2 spread toScore value, T3 spread toScore value, T4 spread toScore value.
I2 is right:T1 spread toScore value, T2
Spread toScore value, T4 spread toScore value.
I3 is right:T3 spread toScore value, T4 are spread to I3Score value.
I4 is right:T4 spread toScore value.
40, determine diffusion process has executed whether diffusing step has reached predetermined optimal diffusion step number out,
And when having executed diffusing step and having reached optimal diffusion step number, terminate diffusion process.
In embodiments of the present invention, optimal diffusion step number can be obtained by way of training in advance, such as optimal expansion
Stroll number can wait odd numbers value for 1 or 3 or 5, spread step number in step 30 and reach predetermined optimal diffusion step number out
When, terminate above-mentioned diffusion process, triggering executes following step 50.
50, it obtains and invests the final score value that M investor's node in network is respectively provided at the end of diffusion process, it will
M investor's node is sorted from large to small according to the final score value being respectively provided with, and is saved to target by investor according to ranking results
Point recommends at least one investor's node.
In embodiments of the present invention, after diffusion, the score value that each investor's node obtains reflects the investor couple
In the recommendation degree of targeted company, score value is ranked up by sequence from big to small, according to number requirement is recommended, selects ranking
Forward investor is that targeted company generates personalized recommendation list.Because the company side of three parts figure is connected according to real connection
, investor ins succession the company invested and the label invested.Therefore, score value is spread along even side, will be indirect having
The node of connection is all diffused into.For example, initial value on the label node contained by targeted company, in first step diffusion, is thrown
The investor's node for providing these labels can all receive certain score value, and investing the more investor's node of these labels will
Receive more score values.Then these investor's nodes obtain score value again value diffusing to company's node and label node
Company's node is to invest the company that the investor of label contained by targeted company invested, and the label node for obtaining score value is
Invested the label that the investor of label contained by targeted company invested.That is, directly with indirectly being generated with these labels
Investor's node, company's node, the label node of relationship can all be successively received score value, and then these score values are concentrated in investor
In node set, wherein odd number step can be the diffusion of the step numbers such as 1,3,5.That is, this weak relationship is continuous by diffusion
Reinforce, is finally reflected in the score value of investor's node, this score value is bigger, just illustrates the relationship of the investor and targeted company
Stronger, the embodiment of the present invention judges that a possibility that investment relation is generated between them is bigger, i.e., recommends the high investor of score value
To targeted company.
In some embodiments of the invention, at least one investor is recommended to include: to targeted company according to ranking results
By the preceding S investor nodes recommendations in ranking results to targeted company, S is the positive integer less than M.
Wherein, the value of S can be 1, i.e., will recommend it with the highest investor of targeted company's matching degree to be recommended,
Perhaps the value of S can be 2 perhaps 3 or more multivalue, recommend multiple investors to realize to targeted company, convenient for its progress
Subsequent further investor's selection.
Next by taking an actual implementation process as an example, this method realizes the specific steps recommended are as follows:
Step S01, history investment relation data and associated tag information are obtained, the investment including every history investment event
Label contained by the information such as side, invested company, investment amount, equity accounting and company.
S02, it is trained by history investment data and obtains optimal step number T*With optimal λ*。
Wherein, history investment relation data refer to the investment event between several investors and company, such as certain investment
Invested certain company for side.And history investment data includes the label information of history investment relation and corresponding company.
Following detailed diffusion process is executed in S02: history investment data is divided into two parts in chronological order, first
Part is training data, and second part is test data.Company-investor-is established according to training data and associated tag information
Label three parts figure, the company's of determination side right weight.There is no investment relation in training data for each, thrown in test data
The company of money regards it as test company, and initial value 1 is arranged in label contained by it, according to company's side right weight in three parts figure, meter
Point counting value after odd number step (i.e. 1,3,5,7 etc. step) diffusion as a result, obtaining the recommendation score value of all investors.Investor is pressed
Recommend score value to sort from large to small, sequence point is calculated according to investor's ranking of the test company of actual investment in test dataWherein r is the ranking of the investor of actual investment the said firm in test set, and L is to participate in sequence
Investor's total number.
Constantly change λ and diffusion step number, when the average sequence point minimum, i.e. accuracy rate highest of all test companies, obtains
To optimal step number T*And λ*.Sequence point is a common recruitment evaluation index of recommender system, exceptionally except sequence, here can also be with
Use recall rate (Recall), other accuracy rate measurement indexs such as AUC.It is higher that general 3 step or 5 steps can be obtained by accuracy rate
As a result.
S03, company-investor-label three parts are established according to all history investment relation data and associated tag information
Figure, the company's of determination side right weight set up initial value for label contained by targeted company to be recommended, by optimized parameter λ*It substitutes into, calculates
Spread T*The recommendation score value of investor after step obtains each investor between target user according to the height of recommendation score value
Recommendation degree selects the higher investor of score value to generate the recommendation list of targeted company.
The measurement of recommendation effect can also be assessed with sequence point, that is, utilize the investor of these targeted companies of actual investment
Measure recommend accuracy rate, certainly, the embodiment of the present invention can also manually choose some companies, it is assumed that invest its company with
And investment event after this occurs is unknown, and diffusion process above-mentioned is carried out to it, carries out accuracy rate after obtaining recommendation results
Test.
By illustration above-mentioned it is found that generating investment network in the embodiment of the present application, which includes:
It is M investor's node, N number of by investor's node, P label node, investor's node and by first between investor's node
Type connects when the Second Type between investor's node and label node connects.It can also repeatedly change in the embodiment of the present invention
Substitute performance diffusion process, the initial value that label node is had in diffusion process are constantly expanded in investment network
It dissipates, because the company side of three parts node in investment network is connected according to real connection, investor ins succession the quilt invested
Invest label node (i.e. invested company, abbreviation company) and invested.Therefore, score value is spread along even side, will be
There is the node of indirect association to be all diffused into.For example, initial value on the label node contained by targeted company, is spread in the first step
In, the investor's node for investing these labels can all receive certain score value, invest the more investor's section of these labels
Point just will receive more score values.Then these investor's nodes obtain again value diffusing to company's node and label node
Company's node of score value is to invest the company that the investor of label contained by targeted company invested, and obtains the label section of score value
Point is to invest the label that the investor of label contained by targeted company invested.That is, directly with indirectly being marked with these
Investor's node, company's node, the label node of label generation relationship can all be successively received score value, and then these score values are concentrated in
In investor's node set.Therefore this weak relationship is constantly reinforced by diffusion, is finally reflected in the score value of investor's node,
This score value is bigger, just illustrates that the relationship of the investor and targeted company is stronger, and investment is generated between investor and targeted company
A possibility that relationship, is bigger, i.e., the high investor of score value is recommended targeted company.Therefore it can be improved and pushed away for start-up company
Recommend the accuracy rate of investor.
Above-described embodiment is to of the invention for example, not as a limitation of the invention, based on the above embodiment also
With equivalent variations or more embodiments can be replaced out, no longer explained one by one herein.
Claims (7)
1. a kind of recommended method of investor characterized by comprising
History investment event information and label information are obtained from history investment data library, the history investment event information is used for
Indicate x-th of investor in M investor and N number of i-th by investor by the history investment relation between investor,
The label information include it is described N number of by the corresponding label of investor, it is described it is N number of by each of investor by investor
Comprising at least one label, the M, the N, the x and the i indicate positive integer, and the x is less than or equal to the M, institute
I is stated less than or equal to the N;
Event information is invested according to the history and the label information generates investment network, and the investment network includes: M throwing
Capital's node, it is N number of by investor's node, P label node, investor's node and by between investor's node the first kind connect
When the Second Type between investor's node and label node connects;Wherein, the M investor node and described M throwing
Capital correspond, it is described it is N number of by investor's node with it is described it is N number of corresponded by investor, the P label node and institute
State it is N number of corresponded by the corresponding all not same labels of investor, the first kind connects side and indicates x-th of investor's node
Being established between investor's node with i-th has an investment relation, and the Second Type connects side and indicates x-th of investor's node and the
Establishing between the i corresponding label nodes of the label included by investor's node has investment relation;
Successive ignition executes following diffusion process, and the diffusion process includes: by the first kind section in the investment network
Point score value obtained in current diffusing step distributes to the first neighbor node being connected with the first kind node, described
Obtained score value is used in next diffusing step by the first neighbor node, and the target of recommendation to be supplied is thrown in the investment network
The corresponding label node of all labels included by capital's node has initial value, the investment net in the diffusion process
The M investor node in network, it is described it is N number of by investor's node, the P label node except the target is by investor
Other label nodes other than the corresponding label node of all labels included by node do not have just in the diffusion process
Beginning score value;Wherein, when the first kind node is the label node, first neighbor node includes: by described
Second Type connects investor's node that side is connect with the label node;When the first kind node is investor's node
When, first neighbor node includes: to connect the label node that side is connect with investor's node by the Second Type, with
And by the first kind connect that side connect with investor's node by investor's node;When the first kind node is
It is described by investor's node and the label node when, first neighbor node includes: by the first kind Lian Bianyu
The investor's node connected by investor's node, and connect what side was connect with the label node by the Second Type
Investor's node;
Determine the diffusion process executed diffusing step whether have reached predefine optimal diffusion step number, and
It is described when having executed diffusing step and having reached the optimal diffusion step number, terminate the diffusion process;
The final score value that M investor's node at the end of the diffusion process in the investment network is respectively provided with is obtained,
The M investor node is sorted from large to small according to the final score value being respectively provided with, according to ranking results to the target
By at least one investor's node of investor's nodes recommendations.
2. the method according to claim 1, wherein history investment event information includes: x-th of investor
Mark, i-th by the mark of investor, investment record data;
It is used to reflect that described i-th industry by investor to be led by least one label that investor's node includes described i-th
Domain and main business.
3. according to the method described in claim 2, it is characterized in that, described invest event information and the mark according to the history
After signing information generation investment network, the method also includes:
Determine that the first kind connects the weight on side according to investor's node and by the investment record data between investor's node;
According to investor's node and by between investor's node investment relation, by the affiliated pass between investor's node and label
System, investor's node and by between investor's node investment record data determine that the Second Type connects the weight on side.
4. according to the method described in claim 3, it is characterized in that, the first kind node by the investment network exists
Score value obtained in current diffusing step distributes to the first neighbor node being connected with the first kind node, comprising:
By first kind node score value obtained in current diffusing step in the investment network, according to the first kind
Even while weight and the Second Type connect while weight, be averagely allocated to first neighbor node.
5. the method according to claim 1, wherein the diffusion process, specifically comprises the following steps:
Step a: in the initial step before starting diffusion, for recommendation to be supplied in the investment network target by investor
Initial value is arranged in the corresponding label node of all labels included by node, invests for described M in the investment network
Fang Jiedian, it is described it is N number of by investor's node, the P label node except the target included by investor's node own
Other label nodes setting score value other than the corresponding label node of label is 0;
Step b: when the first kind node is the label node, the initial value of the label node is passed through described
Second Type connects side and distributes to investor's node according to weighted average;After step b execution, successively iteration executes following step
Rapid c, step d, stop iteration when reaching the optimal diffusion step number;
Step c: when the first kind node is investor's node, investor's node is obtained by step b
Score value connected by the Second Type and distribute to label node according to weighted average when, the first kind connects and invested
Fang Jiedian;
Step d: when the first kind node be it is described by investor's node and the label node when, by described by investor
The score value that node is obtained by the obtained score value of step c and the label node by step c, is connected by the first kind
Investor's node is distributed to according to weighted average when, the Second Type connects, investor's node will pass through the first kind
Lian Bian, the Second Type connect the score value that side respectively obtains and are superimposed, point obtained using investor's node by step d
Value triggering executes step c.
6. according to the method described in claim 5, it is characterized in that, λ times of the score value of investor's node is used to distribute to
By the first kind connect that side connect with investor's node by investor's node, the score value of investor's node
1- λ times, which is used to distribute to, connects the label node that side is connect with investor's node by the Second Type, the λ be greater than
0 and the numerical value less than 1, the λ are predetermined optimal two-way diffusion ratio weight.
7. the method according to claim 1, wherein described saved to the target by investor according to ranking results
Point recommend at least one investor's node include:
By the preceding S investor nodes recommendations in the ranking results to the target by investor's node, the S is less than institute
State the positive integer of M.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910359691.7A CN110264364B (en) | 2019-04-30 | 2019-04-30 | Recommendation method for investor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910359691.7A CN110264364B (en) | 2019-04-30 | 2019-04-30 | Recommendation method for investor |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110264364A true CN110264364A (en) | 2019-09-20 |
CN110264364B CN110264364B (en) | 2022-08-23 |
Family
ID=67913949
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910359691.7A Active CN110264364B (en) | 2019-04-30 | 2019-04-30 | Recommendation method for investor |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110264364B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110750653A (en) * | 2019-10-22 | 2020-02-04 | 中国工商银行股份有限公司 | Information processing method, information processing apparatus, electronic device, and medium |
CN112488842A (en) * | 2020-12-14 | 2021-03-12 | 天津北晟企业服务有限公司 | Investment institution recommendation method and device |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120054200A1 (en) * | 2010-08-26 | 2012-03-01 | International Business Machines Corporation | Selecting a data element in a network |
CN103544632A (en) * | 2013-07-22 | 2014-01-29 | 杭州师范大学 | Method and system for individually recommending network commodities |
US8832091B1 (en) * | 2012-10-08 | 2014-09-09 | Amazon Technologies, Inc. | Graph-based semantic analysis of items |
CN104899763A (en) * | 2015-05-07 | 2015-09-09 | 西安电子科技大学 | Personalized recommendation method based on bilateral diffusion of bipartite network |
CN106326367A (en) * | 2016-08-11 | 2017-01-11 | 华南师范大学 | Mixed collaborative recommendation algorithm based on WNBI and RSVD |
US20180032930A1 (en) * | 2015-10-07 | 2018-02-01 | 0934781 B.C. Ltd | System and method to Generate Queries for a Business Database |
CN107704632A (en) * | 2017-10-31 | 2018-02-16 | 河海大学 | Modification method is recommended based on the Chinese label of synonym and antonym |
CN108038746A (en) * | 2017-11-24 | 2018-05-15 | 河海大学 | Method is recommended based on the bigraph (bipartite graph) of key user and time context |
CN108256958A (en) * | 2017-12-22 | 2018-07-06 | 华南师范大学 | A kind of mixing Collaborative Recommendation algorithm based on WUDiff and RMF |
CN108269117A (en) * | 2017-01-03 | 2018-07-10 | 阿里巴巴集团控股有限公司 | The push of data determines method and device, terminal |
CN108805642A (en) * | 2017-05-02 | 2018-11-13 | 合信息技术(北京)有限公司 | Recommend method and device |
CN109165847A (en) * | 2018-08-24 | 2019-01-08 | 广东工业大学 | A kind of item recommendation method based on recommender system, device and equipment |
-
2019
- 2019-04-30 CN CN201910359691.7A patent/CN110264364B/en active Active
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120054200A1 (en) * | 2010-08-26 | 2012-03-01 | International Business Machines Corporation | Selecting a data element in a network |
US8832091B1 (en) * | 2012-10-08 | 2014-09-09 | Amazon Technologies, Inc. | Graph-based semantic analysis of items |
CN103544632A (en) * | 2013-07-22 | 2014-01-29 | 杭州师范大学 | Method and system for individually recommending network commodities |
CN104899763A (en) * | 2015-05-07 | 2015-09-09 | 西安电子科技大学 | Personalized recommendation method based on bilateral diffusion of bipartite network |
US20180032930A1 (en) * | 2015-10-07 | 2018-02-01 | 0934781 B.C. Ltd | System and method to Generate Queries for a Business Database |
CN106326367A (en) * | 2016-08-11 | 2017-01-11 | 华南师范大学 | Mixed collaborative recommendation algorithm based on WNBI and RSVD |
CN108269117A (en) * | 2017-01-03 | 2018-07-10 | 阿里巴巴集团控股有限公司 | The push of data determines method and device, terminal |
CN108805642A (en) * | 2017-05-02 | 2018-11-13 | 合信息技术(北京)有限公司 | Recommend method and device |
CN107704632A (en) * | 2017-10-31 | 2018-02-16 | 河海大学 | Modification method is recommended based on the Chinese label of synonym and antonym |
CN108038746A (en) * | 2017-11-24 | 2018-05-15 | 河海大学 | Method is recommended based on the bigraph (bipartite graph) of key user and time context |
CN108256958A (en) * | 2017-12-22 | 2018-07-06 | 华南师范大学 | A kind of mixing Collaborative Recommendation algorithm based on WUDiff and RMF |
CN109165847A (en) * | 2018-08-24 | 2019-01-08 | 广东工业大学 | A kind of item recommendation method based on recommender system, device and equipment |
Non-Patent Citations (7)
Title |
---|
JIN MAO 等: "Profiling users with tag networks in diffusion-based personalized recommendation", 《JOURNAL OF INFORMATION SCIENCE》 * |
M. SHANG 等: "Collaborative filtering with diffusion-based similarity on tripartite graphs", 《PHYSICA A: STATISTICAL MECHANICS AND ITS APPLICATIONS》 * |
Z.-K. ZHANG 等: "Solving the cold-start problem in recommender systems with social tags", 《EUROPHYSICS LETTERS》 * |
ZI-KE ZHANG 等: "Personalized recommendation via integrated diffusion on user-item-tag tripartite graphs", 《PHYSICA A: STATISTICAL MECHANICS AND ITS APPLICATIONS》 * |
戴瑾如: "基于加权三部图模型的推荐算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
梁艺 等: "一种基于用户信任网络的物质扩散推荐算法", 《计算机与现代化》 * |
王冉 等: "结合三部图综合扩散的Slope One推荐算法", 《微电子学与计算机》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110750653A (en) * | 2019-10-22 | 2020-02-04 | 中国工商银行股份有限公司 | Information processing method, information processing apparatus, electronic device, and medium |
CN110750653B (en) * | 2019-10-22 | 2023-01-06 | 中国工商银行股份有限公司 | Information processing method, information processing apparatus, electronic device, and medium |
CN112488842A (en) * | 2020-12-14 | 2021-03-12 | 天津北晟企业服务有限公司 | Investment institution recommendation method and device |
Also Published As
Publication number | Publication date |
---|---|
CN110264364B (en) | 2022-08-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6993496B2 (en) | Method and system for determining market demand based on consumer contributions | |
Johari | Efficiency loss in market mechanisms for resource allocation | |
US7509274B2 (en) | Internet-based system for identification, measurement and ranking of investment portfolio management, and operation of a fund supermarket, including “best investor” managed funds | |
Dilger et al. | Upcrowding energy co-operatives–Evaluating the potential of crowdfunding for business model innovation of energy co-operatives | |
US7505926B2 (en) | Methods, apparatus, and computer readable medium for processing bids in an auction | |
CN105574147A (en) | Information processing method and server | |
CN1842815A (en) | Method and apparatus for determining a minimum price per click for a term in an auction based internet search | |
CN105960656A (en) | Methods and systems for creating data-driven attribution model for assigning attribution credit to multiple events | |
JP6971413B1 (en) | Asset processing methods and systems for converting assets into digital assets | |
CN103034963B (en) | A kind of service selection system and system of selection based on correlation | |
CN110264364A (en) | A kind of recommended method of investor | |
Jovanovic et al. | Equilibrium bid-price dispersion | |
Kujawski | 4.7. 3 Multi‐Criteria Decision Analysis: Limitations, Pitfalls, and Practical Difficulties | |
AU2001257079B2 (en) | Internet-based system for identification, measurement and ranking of investment portfolio management, and operation of a fund supermarket, including "best investor" managed funds | |
CN110175883A (en) | A kind of sort method, device, electronic equipment and non-volatile memory medium | |
CN109767333A (en) | Select based method, device, electronic equipment and computer readable storage medium | |
Hyde et al. | Investigating concurrency in online auctions through visualization | |
Arsad et al. | Selection input output by restriction using DEA models based on a fuzzy Delphi approach and expert information | |
AU2001257079A1 (en) | Internet-based system for identification, measurement and ranking of investment portfolio management, and operation of a fund supermarket, including "best investor" managed funds | |
McCarthy et al. | A live-user evaluation of incremental dynamic critiquing | |
Zawadzki et al. | Stock markets’ reactions to the announcement of the hosts. An event study in the analysis of large sporting events in the years 1976–2032 | |
Bai et al. | A fair incentive mechanism in federated learning | |
JP2004164374A (en) | Data processor, data processing program and data processing method | |
CN107146159A (en) | Generate the method and system of index of securities | |
Nanduri | Application of reinforcement learning-based algorithms in CO 2 allowance and electricity markets |
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 |