JP2002099866A - Information processor simulating propagation of information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an effective advertising method in place of conventional ones since recent consumers are accustomed to advertisement by mass marketing and hardly influenced by it. SOLUTION: A method is provided focusing attention on the fact that the recent consumers are easily influenced by judgment of people around them, specifying the people having a large influence on others by information processing, and specifying the people who should be prime targets of a promotion.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing apparatus for simulating the propagation of information for advertisement and the like.

[0002]

BACKGROUND OF THE INVENTION Conventionally, the following methods have been known as advertising methods. That is, as the four media
Newspapers, magazines, radios, televisions, and sales promotion methods include direct mail, insert advertisements, signboard advertisements, neon lights, and the like, as well as traffic advertisements, store promotion materials, telephone directory advertisements, exhibitions, and expositions. In recent years, so-called new media advertisements (CATV, satellite broadcasting, text broadcasting) and Internet advertisements (banners, etc.) have been increasing. However, word-of-mouth advertising has not been established.

[0003]

2. Description of the Related Art For example, Gerard Weibuch, Dietrich
Using a square model such as Stauffer (Santa Fe Rese
arch Institute, Working Paper 00-07-036), using the percolation technique to simulate the propagation of information. However, this method can only express the relationship between adjacent cells. However, actual human relationships vary widely, and it has been difficult to express diverse human communities using this method of expression. In the real world, communities exist at various scales, and there is no method for systematically analyzing them.

[0004]

In view of the recent trend in the market, it can be seen that people are greatly affected by rules different from those of conventional large-scale mass marketing advertising. That is, modern city dwellers have become accustomed to the overwhelming amount of publicity, and are less likely to be affected by the publicity information by the above-described method. In contrast, it takes a lot of effort to extract and judge the necessary information from a large amount of information, and is therefore more susceptible to the judgment of the people around him. As a result, it is considered that this is a factor in the composition of recent “single win”. As a result, we have lost effective advertising techniques.

[0005]

Summary of the Invention Malcolm Gladwell, a recent Washington Post reporter, gave an opinion on the mechanisms of such spontaneous epidemics in his book, Tipping Pont. In other words, he analyzed that the word-of-mouth explosion plays an important role in people with a special personality.

On the basis of this hypothesis, we conclude that the present inventors can identify the people who play such important roles and focus on promotion activities to effectively create an epidemic. I came to. That is, according to the present invention, it is possible to specify a person who plays an important role by quantifying and expressing the influence of a person and simulating the trend or the propagation of information using the computer.

According to the present invention, the influence is expressed by dividing a value unique to each person and a value determined by the relationship with each person. That is, in a set P = [P _1, P ₂ ,..., P _n ] composed of a finite number of elements, the relationship between each element of P is a matrix.

[0008]

(Equation 3)

[0009] is represented by, and P _m P a is P _m consists unique value set ^{_{A Pm = {a 1 Pm,}} a 2 Pm, ..., a l Pm} threshold for calculating the presence or absence of, and influence T characterized by having a ^Pm, for each element P _m, involved row or column vector R ^Pm unique value a ^Pl and P _m of P _l is at least P _l P
By comparing the value calculated by using with the threshold value T ^Pm ,
By using an information processing device that determines the state of P _m to a digit, various human relationships can be described. Also the set P is expressed as P = P ^n, the set ^{_{P n = {P 1 n-}} 1, P 2 n-1, ..., P m n-1} of m elements when expressed as, P ⁿ By adding the same condition as the set P ⁿ to the element P _l ^n-1 of, various communities can be expressed hierarchically, and scalability has been successfully achieved in the simulation of information propagation. Further, for each element P _{m of the} set P composed of n elements, the influence amount E ^pm of P _m after 0 to t ₁ time is calculated using A _Pl and vector R ^Pm .

[0010]

(Equation 4)

[0011] It is possible to observe the change of the state with respect to time. Further, by substituting discrete time from the above equation and repeating the calculation, it is possible to observe the change over time of the state in the set P.

Further, in a set P composed of a finite number of elements, a set of unique values of each element P _m is a vector A ^Pm =
{A ^Pm , B ^Pm , ...}, where A ^Pm = ｛a ₁ ^Pm , a ₂ ^Pm , ..., a
_l ^Pm ｝, B ^Pm = {b ₁ ^Pm , b ₂ ^Pm , ..., b _l ^Pm ｝, ..., (
Vector variables are denoted by “vector” before the variables to distinguish them from scalar variables), and a set of unique constants represented by an n-dimensional vector, and a threshold vector T ^Pm = ^{_{{T a Pm, T a Pm}} , ...} and characterized by having the results obtained by vector operation using a vector R ^Pm as defined in claim 1 and a vector a ^Pm for each element P _m By expressing the state of P _{m in} a digit or as a set of digit notations by comparing it with a threshold vector T ^Pm , it is possible to determine the state of P _m based on a plurality of criteria, It has become possible to bring the cause closer to the natural judgment of humans used for evaluation. As a specific example, parameter setting and potential notation are shown below.
Of course, the present invention is not limited to these.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS As an example, an influence curve of each individual is formulated by a function of an influence coefficient A and a characteristic coefficient b. Also, let the intimacy be distance r and the affected threshold be T
And The setting of various variables as an example is described below. Variable I _Pi : influence potential: influence potential curve of individual Pi A _Pi : influence coefficient: determined from the qualities of individual Pi. Expresses trust, good feelings, etc. b _Pi : characteristic coefficient: expresses the behavior characteristic, extroversion, etc. of the individual Pi. r: Human relationship distance: Represents intimacy. It is related to conversation time and relationships. τ: Unit time: System unit time T _pi : Influence threshold: Influence E _(Pi) (τ) applied to the individual Pi exceeds this value. E _(x) : Influence amount: Integral of influence over time Potential curve It is assumed that the influence of each person is expressed as I _Pi = A _Pi | r | ^(-bpi) .

[0014]

[Potential notation of influence] FIG. 1 shows one-dimensional coordinates X =
The position of 0 indicates a state undergoing existent influence P _1. Now, when P ₂ exists at a distance of coordinate x = −3,
Since the human relationship distance r = 3, when t = τ ₀ (that is, τ ₁
The transmission amount E _(p2) (τ ₀ ) received just before the change to the above can be expressed as follows. E _P2 (τ ₀ ) = A _P1 · 3 ^(-bpi) Equation 1

Next, let T _P1 be the threshold at which P ₂ is affected,
If the threshold is exceeded for τ ₁ , that is, E _(P1) (τ ₁ )> T _P1 , the individual P ₂ is affected, and a potential I _{P2 of} P ₂ is generated at the coordinate X = −3. At this time, the potential of the entire system can be represented by a linear sum of the potentials of the individuals P ₁ and P ₂ (FIG. 2). I (x) = I _P1 (x) + I _P2 (x) Equation 2 FIG. 2: Potential curve when P _P1 (0) and P _P2 (-3) are affected. In this case, for example, coordinates The potential at X = −4 is τ ₀ : I _P1 (−4) = 0.5 → τ ₁ : I _P1 (−4) + I _P2 (−
4) = 1.5. That if there are if x = -4 to individuals P _3, will be exposed to three times the influence by changing the τ ₀ → τ _1.

E _(P3) (τ ₁ ) = E _(P3) (τ ₀ ) + E _(P3) (τ ₁ ) = I _P1 (−4) + (I _P1 (−4) + I _P2 (−4) )) Equation 4 Consider a state where n people are present in the same space. After jτ, that is, the influence amount of Pi at τ _j is obtained by adding the sum of all potentials of the activated persons among n persons at time t to t = τ ₀
To t = τ _j .

[0017]

(Equation 5)

I _{k *} ^t : Potential consisting of elements affected at time t By using the above equation, calculating the amount of influence of each individual at an arbitrary time τ _j in an arbitrary n-person space Becomes possible. By comparing the amount of influence obtained in this way with the threshold value of each individual, the propagation of influence can be expressed.

[0019]

[When there are multiple people] (Example 1) Human relationship density is low
If on the coordinates of the one-dimensional human relationship distance the same unique as listed in Table 1
The person with the value, P₁, P _Two, P_Three, P_Four, P_FiveOf the five people
X = −10, −5,0,5,10
You. Among them, P₁(X = 0) is τ₀Influence with
Consider the state of receiving.

[0020]

[Table 1]

The initial value was set to P ₃ , and the subsequent change was calculated by using Equation 5.

[0022]

[Table 2]

At τ ₃ , P ₂ and P ₃ are affected beyond the cumulative influence threshold. The potential curve of the system at this time is shown in FIG. (Example 2) When the human relationship density is high Next, as initial conditions, P ₁ , P ₂ , P ₃ , P ₄ , and P ₅ are respectively X = −2, −1, 0, 1, and 2. Consider the state that exists at a location. Other conditions are the same. At this time, unlike Example 1, P ₂ and P ₄ were inverted after 2τ, and all were inverted after 3τ (FIG. 4). As described above, when X is set as the human relationship distance, it has been found that the density of the human relationship greatly affects the propagation of the influence.

(Example 3) In the case of coarse and dense Next, the life time (TTL = time to live) of the affected individual is 3
Consider τ, and consider a state with density. That is, P₁, P_Two, P_Three,
P_Four, P_FiveAre located at X = -9, 0, 1, 1.5, 6 respectively.
Are there. At this time, τ₀In P_ThreeThat is, the person with X = 1
A simulation was first attempted for the affected case.
τ_TwoIn P_TwoAnd P_FourIs affected (FIG. 5), and τ_ThreeIn addition
P₁And P_FiveAlso affected P_ThreeIs TTL (Time to Live) = 3
At the end of his life, his potential has disappeared. And τ_FiveThen P_Two,
P_FourHas expired and its potential has disappeared (Fig.
7). The above process is expressed in black and white. ○ is influential
And ● indicates that there is an effect, τ ₀→ τ_FiveIs shown in the table below.
Will be revealed.

[0025]

[Table 3]

Next, under the same environment as in Example 3, referred to the case where P ₅ as an initial value is affected in a similar manner (Table 4, Figure 8).

[0027]

[Table 4]

From the above, it has been found that how to select the initial value greatly affects the subsequent situation.

[0029]

[Effect of Parameter A] The parameter A affects the magnitude of the absolute value of the influence potential. Specifically, the person ’s personality, especially intellect, confidence, kindness,
It is expressed as a composite of elements that are trusted by people around. (Example 4) The parameters are set as follows in the same manner as in the parameter example 1 of P _{1 to} P ₅ .

[0030]

[Table 5]

In this state, when P ₃ is set to the initial value, the change is

[0032]

[Table 6]

On the other hand, when P ₂ is set to the initial value,

[0034]

[Table 7]

As a result, no change occurs. Similarly, the value of b greatly affects the entire system. As seen above, the influence propagation mechanism is simulated by expressing the influence in terms of potential and capturing changes with time. Further, at this time, there is an extreme result that almost the entire system is affected or not affected, and an intermediate value is not obtained.

[0036]

[Human relationship distance r] The human relationship distance expresses the opportunity density of influence. This value represents the relationship of each element in the "group" as a sociological variable pointed out by Lazarsfeld et al., For example, the probability of existence of a conversational space per day, the relationship with each other due to work, etc. Can be expressed as a function of the observable quantity of Human relationship distance = f (probability of existence of conversational space, mutual relationship) Of course, each element can belong to a plurality of sets depending on its life activity pattern.

Next, consider the size of the set. When you look at the influence of others in real life, the scope is almost limited to the communities in which you participate. As an example, consider the environment that are of Mr. P _1. P _1's in the office who are the employees of about 16 people, are working. Holiday to his wife, spend at P ₃₄₈ and the house. Let's express this situation in a human relationship distance table. P ₃₄₈ and other employees of Company A (P ₂ to P ₁₆ )
For example, if the above function is considered to be adaptable, it can be said that there is almost no conversational space existence probability and mutual relationship. Therefore, the chances of influencing derived from these parameters may be considered as zero, and there is no problem. That is,
Regarded the apartment and the A's workplace P ₁ as an independent set, better to representation is divided into two kinds of human relations distance table is reasonable.

[0038]

[Table 8]

[0039]

[Table 9]

In this way, each community can be made independent and its human relationship can be expressed. _Suppose further that P ₃₄₈ worked for another company B. Then, P ₃₄₈ participates in the company B's community and is represented in the company B's relationship distance table. As a result, A company and B company will have an associated through home community P _1.

[0041]

[Hierarchical representation of communities] From the examples so far, in a community represented by a certain human relation distance table, almost all elements are affected when the criticality is exceeded,
If it does not exceed, it converges. This is nothing but the fact that the community itself has a threshold and a digitized representation of whether it is affected or unaffected is applicable. This means that the set of individuals, P
Also have a relationship with other sets, and the above-mentioned notation method can be applied.

[0042] (example) work to the company's head office of community relations distance Table P ₁ located in Tokyo, Tokyo branch office on the same floor, but also there is a branch in Yokohama and Kawasaki branch office to another place. P ₁ has been working in the company A Kawasaki branch office, located on the 6th floor of the K building. In addition, B software development, which is the outsourcing destination, is located on the 5th and 4th floors of the building, and unrelated stores exist on the 1st to 3rd floors. This state can be expressed as a community relation distance table in the same manner as described above.

[0043]

[Table 10]

[0044]

[Table 11]

The community relationships expressed in this manner have a hierarchical structure, and the scale can be properly used according to the purpose.

[0046]

[Calculation of parameters] Two types of parameters for calculating the influence potential curve are set as examples. One is the influence coefficient A, which affects the absolute value of the potential, and is calculated from the reliability, liking, and the like. Next, the characteristic coefficient b affects the shape of the potential curve, and is calculated from the person's extraversion and behavioral characteristics.

[0047]

[Table 12]

As a specific calculation method, Table 12 shows an SPI (Synthetic Personal Computer) often used for personnel measurement and the like.
ity Inventory) An example in which A and b of each person were calculated from the personality test was listed.

[0049]

[Calculation of Threshold] The threshold of an individual for a certain event differs depending on the object of interest of the individual. Table 13 shows, as an example, a calculation table of a threshold value regarding a purchasing behavior when a mobile strap using a new character is sold for 1000 yen.

[0050]

[Table 13]

In addition, a threshold value can be calculated from necessary information for various types of events (purchasing behavior of a trendy item, information transmission, etc.). For the calculation of these threshold values, a more reliable value can be obtained by repeating the operation of testing the existing data for an appropriate relationship by, for example, covariance structure analysis.

[0052]

[TTL (Time To Live)] Generally, information transmission has a life. That is, it can be considered that the information similar to the epidemic targeted in this study loses its propagating power when its freshness is lost. TTL (Time)
to Live) and can be expressed in multiples of the unit time. It can also be expressed as a function of time or a value unique to each element.

[0053]

[Calculation of potential] Let's calculate the potential of the community of Kawasaki branch office. The profile of each member is as follows.
Table 1 is used. From these values, the state of change when P ₆ having the largest influence coefficient was set as the initial value was calculated and shown in Table 14. At this time, τ ₂ affected the entire system, and all converged by τ ₅ .

[0054]

[Table 14]

[0055] initial value and then at a system similar to example when taking a P ₁₁ shown in Table 15.

[0056]

[Table 15]

[0057] converges with tau _7, all of P ₁ to P ₁₆ is activated so far. In the table, the unaffected state is represented as 0, and the affected state is represented as 1.

[0058]

When the information processing apparatus according to the present invention is used,
It becomes possible to simulate the spread of the influence of word-of-mouth, and to carry out public relations activities effectively.
The use of the Internet makes it easier to collect information for determining parameters. In addition, it is possible to promote specific people using the Internet based on the obtained information.

[Brief description of the drawings]

FIG. 1 is a diagram showing potential notation at X = 0.

FIG. 2 is a diagram showing a potential curve when P _P1 (0) and P _P2 (-3) are affected.

FIG. 3 is a diagram showing P ₂ , P ₃ , and P ₄ after τ ₃ and a total potential curve.

FIG. 4 is a diagram showing a potential curve at τ ₃ .

FIG. 5 is a diagram showing a potential at τ ₂ .

FIG. 6 is a diagram showing a potential at τ ₃ .

FIG. 7 is a diagram showing a potential at τ ₅ .

FIG. 8 is a diagram showing a potential from τ ₀ to τ ₂ when P ₅ is set to an initial value.

[Procedure amendment]

[Submission date] January 16, 2001 (2001.1.1)
6)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] Claim 1

[Correction method] Change

[Correction contents]

(Equation 1) In the expression, and P _m ∈ set ^{_{A Pm = {a 1 Pm,}} a 2 Pm, ..., a l Pm} of P at a P _m consists specific value threshold T ^Pm to calculate the presence of, and effects has, for each element P _m, at least P _l ∈ P a is P _l unique value a ^Pl and P _m involved row or column vector R threshold value calculated using ^Pm T of the
By comparison with ^Pm, the information processing apparatus to simulate the propagation of information and determining the state of the P _m to digit.

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0009

[Correction method] Change

[Correction contents]

And P_m ∈P which is P_mIs unique
Set A consisting of the values of^Pm= ｛A₁ ^Pm, a_Two ^Pm,…, A_l ^Pm及, and
Threshold T for calculating the presence or absence of influence^PmCharacterized by having
And each element P_mFor at least P_l ∈P which is P_l
Unique value A of^PlAnd P_mThe row or column vector R involved
^PmIs calculated using the threshold T^PmBy comparing with
, P_mUsing an information processing device that determines the state of the
Various human relationships.
Noh. Also, the set P is P = PⁿAnd m elements
A set P consisting ofⁿ= ｛P₁ ^n-1, P_Two ^n-1,…, P_m ^n-1｝ And expression
When you doⁿElement P of _l ^n-1Set PⁿWith the same conditions as
To provide a hierarchical representation of the various communities.
To simulate information propagation.
We succeeded in giving us mobility. N more elements
Each element P of the set P consisting of primes_m From 0 to t₁time
Later P_mInfluence amount E^pmA_PlAnd the vector R^PmUsing
hand,

Claims (7)

[Claims] A set P = [P ₁ , P ₂ ,
.. _Pn ], the relationship between the elements of P is a matrix In the expression, and P _m P a is set A ^Pm P _m consists unique value ^{_{= {a 1 Pm, a 2}} Pm, ..., a l Pm} have a threshold T ^Pm to calculate the presence of, and effects and, that for each element P _m, is compared with a threshold value T ^Pm the calculated value with the involved row or column vector R ^Pm unique value a ^Pl and P _m of P _l is at least P _l P An information processing apparatus that simulates the propagation of information, wherein the state of P _m is determined to be a digit. 2. The set P is defined as P = PⁿAnd m
A set of primes is P ⁿ= ｛P₁ ^n-1, P_Two ^n-1,…, P_m ^n-1}When
When expressed, PⁿElement P of_l ^n-1Is the set PⁿSame conditions as
2. The propagation of information according to claim 1, wherein
Information processing device to simulate. 3. Each element P _{m of a} set P consisting of n elements
When performing the calculation using the A _Pl and the vector R ^Pm , the influence amount E ^pm of P _m after time t ₁ from 0 is expressed as follows: Expressed as an information processing apparatus to simulate the propagation of claim 1, wherein the information and determines this to a digit of the state of the P _m by comparing the threshold value T ^Pm. 4. The information processing apparatus for simulating information propagation according to claim 3, wherein a time-dependent change of the set P is observed by repeatedly determining time as a variable. 5. In the set P, each element P_mHave
The set of unique values is vector A^Pm= ｛A^Pm, B^Pm,…｝,
But A^Pm= ｛A₁ ^Pm, a_Two ^Pm,…, A_l ^Pm｝, B^Pm= ｛B₁ ^Pm,
b_Two ^Pm,…, B_l ^Pmで, ..., n-dimensional vector
Can be represented by a set of unique constants and a vector of the same dimension
Threshold vector T^Pm= ｛T_A ^Pm, T_A ^Pm,…｝, Each element
P_mFor the vector A^PmAnd the vector R^PmUsing
Calculates the result obtained by vector operation as a threshold vector
Le T^PmBy comparing with _mTo the digit
Or expressed as a set of digital notations
The information for simulating the propagation of information according to claim 1.
Information processing device. 6. An information processing apparatus for simulating the propagation of information, wherein the influence amount after time t is expressed by the expression according to claim 3 using the vector according to claim 5. 7. The information processing apparatus for simulating information propagation according to claim 6, wherein a change over time of the set P is observed by repeatedly determining the time using the time as a variable.