JP2012078768A - Person matching device, method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to motivate a non-expert to improvement more efficiently.SOLUTION: A person matching device collects multiple life log data and attribute data of multiple experts from terminals ET1 to ETn of the experts and accumulates both data in an expert information database 31. The person matching device acquires life log data and attribute data of non-expert UFi from a terminal FTi of the non-expert in this state. A matching processing part 25 extracts the feature amount from the life log data and attribute data and extracts each feature amount from the stored life log data and attribute data of the experts. By using a formula for DGM (Degree of Gap Matching) each similarity of the extracted feature amounts is calculated between the non-expert UFi and the experts. An expert who has the maximum DGM is selected and the life log data and attribute data of the selected expert are transmitted to the terminal FTi of the non-expert UFi.

Description

  The present invention relates to a person matching apparatus, method, and program for selecting, for example, a skilled person who has similarities to the unskilled person and presenting the information.

  In educational institutions such as universities and various training institutions, various measures for improving the level of learners or trainees are being studied. As one example, collaborative filtering is used, for example, the degree of similarity in the number of browsing times for the learning material content between learner A and other learners B and C is obtained, and the number of times the common learning material content is browsed based on the result. A technique has been proposed in which learner B or C is discovered as a learner having a similar learning pattern, and teaching material effective for overcoming weakness is taught from learner B or C to learner A (for example, non-patent literature) 1).

  In addition, as a method for transmitting the skill of the skilled worker to the trainer, for example, there is a teaching method for detecting the position information and force information of the skilled worker's brush in the practice of calligraphy and presenting the log data to the trainer for guidance. (For example, refer nonpatent literature 2).

Yuma Tsutsume, Yuji Wada, Kenichi Toi, “Implementation of Recommendation Technology in e-Learning-Extraction of Teaching Materials Contributing to Overcoming Disabilities”, Information Science and Technology Forum (FIT2009), pp. 615-618, 2009. Satoshi Satoshi, Naoki Kawakami, and Susumu Tsuji, “Study on Teaching Techniques Utilizing Force Subjectivity”, Transactions of the Virtual Reality Society of Japan, Vol. 10, No. 3, pp. 363-369, 2005.

  However, in the method described in Non-Patent Document 1, a learner with similar browsing materials by collaborative filtering, that is, a learner having a similar learning level is found, and the learning material used by the learner is recommended. Yes. For this reason, a learner who has received a recommendation for a teaching material will receive a recommendation from another learner at the same level as himself / herself. "I can't expect it to feel."

  On the other hand, in the method described in Non-Patent Document 2, since the log data of a person with a higher skill level than the trainee is presented to the trainee, an effect that makes it seem that “I want to increase the level like that person” can be expected. . However, the trainer only knows the current level of the skilled person, and the skilled person can only feel that he is far from the trainer himself. I can't expect an effect that makes me think. As a result, it is difficult for the trainer to be motivated to take action to increase his / her level.

  The present invention has been made paying attention to the above circumstances, and an object of the present invention is to provide a person matching apparatus, method, and program that enable non-skilled persons to have a motivation to improve more effectively. There is.

In order to achieve the above object, a first aspect of the present invention comprises the following components.
That is, life log data and attribute data of a plurality of skilled workers from the past to the present are collected and stored. In this state, the current life log data and attribute data of the unskilled person are acquired from the terminal of the unskilled person, and from the acquired unskilled person life log data and the stored life log data of each skilled person The current skill level of each unskilled person and each skilled person is extracted. Further, the feature amount representing the current situation of the unskilled person is extracted from the acquired attribute data of the unskilled person, and the feature representing the past situation of each expert from the stored attribute data of each skilled person Extract the amount. Further, the difference between the extracted unskilled person and the current skill level of each skilled person, and the similarity between the extracted current feature quantity of the unskilled person and the past feature quantity of each skilled person are also obtained. And calculating a quantitative evaluation index representing the degree of similarity between the unskilled person and the plurality of skilled persons. Based on the calculated quantitative evaluation index, a predetermined condition is selected from the plurality of skilled persons. An expert who satisfies the condition is selected, and information on the selected expert is transmitted to the terminal used by the non-expert.

とすると共に、ｐを上記記憶された過去のライフログデータの集合の中で、

が最大になるときの値（ｐ_max）とし、Ｌ[m] をレベルベクトルの要素（m個）、Ｘ[n] を特徴量（n個）、Ｌ，Ｘの添え字であるＥ，Ｆをそれぞれ熟練者及び非熟練者、ｃを現在、ｐを過去とし、さらにＷ_L[m]を各レベルベクトル要素の重み付け係数、Ｗ_X[n]を各特徴量の重み付け係数（n個）とそれぞれ表すとき、定量評価指数ＤＧＭを

により計算する。 In particular,

And p in the stored set of past life log data,

Is the maximum value (p_max), L [m] is the element of the level vector (m), X [n] is the feature quantity (n), and E and F that are subscripts of L and X are Expert and non-skilled person, c as current, p as past, W _L [m] as weighting coefficient for each level vector element, W _X [n] as weighting coefficient for each feature quantity (n) When representing the quantitative evaluation index DGM

Calculate according to

  Therefore, according to the first aspect of the present invention, the feature amount reflecting both the current ability level difference between the unskilled person and the skilled person, and the current life log data and attribute data of the unskilled person And a first quantitative evaluation index between the unskilled person and the skilled person is obtained based on the similarity between the feature amount reflecting both the life log data and the attribute data of the skilled person, An expert is selected based on the first quantitative evaluation index. In other words, an expert who has a much higher skill than an unskilled person but has a similar past situation is selected by the matching process. For this reason, it becomes possible to give a non-skilled person a willingness to “increase the level like that person”. Moreover, since the information of the skilled person whose past situation is similar to the current self is presented to the unskilled person, the unskilled person can feel that the skilled person is familiar. You can expect an effect that makes you feel like you can improve your level. "

In addition, according to a second aspect of the present invention, when calculating the quantitative evaluation index, the evaluation data for the transmitted information regarding the skilled person is received from the terminal of the unskilled person, and the received evaluation data is included in the received evaluation data. Based on this, the weighting coefficient W _X [n] of each feature quantity in the above equation (1) is made variable.
In this way, the weighting coefficient for calculating the quantitative evaluation index is adjusted based on the evaluation data transmitted from the unskilled person in response to the notification of the data representing the matching result, thereby calculating the quantitative evaluation index. The expert is selected again. Therefore, information about the skilled person whose feature amount is closer to that of the unskilled person is presented, and this makes it possible to more effectively increase the non-skilled person's desire to improve.

により計算するようにしてもよい。
このようにすると、定量評価指数を計算する際に、重み付け和の計算に限らず積の形を考慮することができる。 Further, when calculating the above quantitative evaluation index, L is a level vector, X is a feature vector, the feature quantity contained in these level vector L and feature vector X is expressed by equation (1), and Sim is the degree of similarity When the quantitative evaluation index DGM is

You may make it calculate by.
In this way, when calculating the quantitative evaluation index, not only the calculation of the weighted sum but the shape of the product can be considered.

  According to a third aspect of the present invention, for each of a plurality of skilled workers, the growth rate represented by the extracted current skill level and the time taken to obtain the skill level, and the extracted past The degree of growth difficulty representing the difficulty of growth for each feature amount is calculated, and the second quantitative evaluation index, which becomes larger as the calculated growth rate and growth difficulty level are higher, Further calculations are made for each combination with the expert. Then, based on the calculated first and second quantitative evaluation indices, an expert who satisfies a predetermined condition is selected from the plurality of experts, and information on the selected expert is obtained as the non-skilled person. Is transmitted to the terminal used by the user.

とすると共に、ｐを前記記憶された過去のライフログデータの集合の中で、

が最大になるときの値（ｐ_max）とし、Ｌ[m] をレベルベクトルの要素（m個）、Ｘ[n] を特徴量（n個）、Ｌ，Ｘの添え字であるＥ，Ｆをそれぞれ熟練者及び非熟練者、ｃを現在、ｐを過去とし、さらにＷ_L[m]を各レベルベクトル要素の重み付け係数、Ｗ_X[n]を各特徴量の重み付け係数（n個）とし、さらにｔを熟練者の過去ｐから現在ｃまでの総努力時間、Ｄ[n]を特徴量ごとの熟練者の成長困難度、Ｗ_T[m]を各レベルベクトル要素ごとの成長率の重み、Ｗ_D[n]を各特徴量の成長困難度の重みとしたとき、第２の定量評価指数ＥＰを、

により計算する。 Specifically, the calculation of the second quantitative evaluation index is performed as follows.

And p in the stored set of past life log data,

Is the maximum value (p_max), L [m] is the element of the level vector (m), X [n] is the feature quantity (n), and E and F that are subscripts of L and X are Each skilled and non-skilled person, c as the present, p as the past, W _L [m] as the weighting coefficient for each level vector element, W _X [n] as the weighting coefficient for each feature quantity (n), Furthermore, t is the total effort time from the past p to the current c of the skilled person, D [n] is the difficulty level of the skilled person for each feature, W _T [m] is the weight of the growth rate for each level vector element, When W _D [n] is the weight of the growth difficulty of each feature amount, the second quantitative evaluation index EP is

Calculate according to

As a method for selecting the expert, a method using weighted addition and a combination of normalization processing and multiplication processing are conceivable.
That is, first, using weighted addition, the first quantitative evaluation index is DGM and its weighting factor is W ₁ , the second quantitative evaluation index is EP and its weighting factor is W ₂ . When
W _{1 ·} DGM + W ₂ · EP
And an expert who satisfies a predetermined condition is selected from the plurality of experts based on the calculated value.

When calculating the above W _{1 ·} DGM + W ₂ · EP, the input of the evaluation result of the unskilled person with respect to the selection result is accepted, and the weighting factors W ₁ and W ₂ are made variable based on the input evaluation result. It is good to set.
In addition, a combination of normalization processing and multiplication processing, when selecting the expert,
When the first quantitative evaluation index is DGM and the second quantitative evaluation index is EP, the first and second quantitative evaluation indices DGM and EP are normalized, and the normalized DGM and EP are normalized. Product DGM × EP
And an expert who satisfies a predetermined condition is selected from the plurality of experts based on the calculated value.

  Therefore, according to the third aspect of the present invention, the growth rate represented by the current skill level of each expert and the time required to obtain the skill level, and each of the extracted past feature values. A second quantitative evaluation index that is larger as the degree of growth difficulty representing the difficulty of growth is higher is calculated. Then, in addition to the first quantitative evaluation index described above, an expert is selected in consideration of the second quantitative evaluation index. In other words, a skilled person who has a higher growth rate than a non-skilled person but has a high growth rate even though the past situation is similar to that of a non-skilled person and the skill level is difficult to improve is selected. The For this reason, it is possible to give unskilled people a sense of familiarity, and at the same time provide motivation such as “I want to raise the level like that person” or “I want to take action to raise the level” .

  That is, according to the present invention, it is possible to provide a person matching apparatus, method and program capable of more effectively giving an unskilled person a desire to improve.

The schematic block diagram of the system which implement | achieves the person matching service concerning embodiment of this invention. The block diagram which shows the structure of the matching server used with the system shown in FIG. The flowchart which shows the procedure and processing content of the matching process in 1st Embodiment of this invention. The flowchart which shows the procedure of the weighting adjustment process among the matching processes shown in FIG. 3, and a processing content. It is a figure which shows an example of DGM (Degree of Gap Matching) calculated by the matching process shown in FIG. The flowchart which shows the procedure and processing content of the matching process in 3rd Embodiment of this invention. The flowchart which shows the procedure of the weighting adjustment process among the matching processes shown in FIG. The figure which shows an example of the relationship between aging and physical strength. The figure which shows an example of the growth difficulty derived from the relationship between aging and physical strength.

Embodiments according to the present invention will be described below with reference to the drawings.
(First embodiment)
FIG. 1 is a schematic configuration diagram of a system for realizing a person matching service according to the first embodiment of the present invention.
This system includes a matching server SV having a function as a person matching device, and communication is performed between the matching server SV and terminals ET1 to ETn used by skilled persons and terminals FT1 to FTm used by non-experts. Communication is made possible via the network NW. The communication network NW includes, for example, an IP (Internet Protocol) network represented by the Internet and a plurality of access networks for accessing the IP network. As the access network, for example, a LAN (Local Area Network), a wireless LAN, a mobile phone network, a wired phone network, and a CATV (Cable Television) network are used.

  Terminals ET1 to ETn used by skilled workers and terminals FT1 to FTm used by unskilled persons are all composed of portable terminals such as mobile phones, smart phones, PDAs (Personal Digital Assistants), and GPS (Global Positioning System). It has a location measurement function, various sensors, a scheduler for managing an action schedule, a web access function such as a browser, an e-mail transmission / reception function, and the like.

  Each of the terminals ET1 to ETn and FT1 to FTm transmits life log data and attribute data of skilled and unskilled personnel to the matching server SV using the Web access function. Further, the terminals FT1 to FTm used by the unskilled person receive data representing the matching result from the matching server SV using the e-mail transmission / reception function, and use the web access function to send the evaluation data of the matching result to the matching server SV. Send.

  For example, the life log data is generated based on information manually input by the user and stored in the storage unit in the terminal, or is detected by a sensor such as the position measuring function or the pedometer, contents of the scheduler, electronic It is automatically generated based on the recording data of the reproduction time of the book / music and stored in the storage unit. Then, it is selectively read in response to a transmission request from the matching server SV and transmitted to the matching server SV.

  The attribute data is input by the key input operation of the skilled person and the unskilled person, and is transmitted to the matching server SV in response to a transmission request from the matching server SV. The evaluation data is input by a non-experienced person by a key input operation when receiving data representing a matching result, and is transmitted to the matching server SV using a Web access function. Note that the attribute data may be a biological age generated based on detection data of a biological sensor attached to the mobile terminal.

By the way, the matching server SV is composed of a Web server, for example, and is configured as follows. FIG. 2 is a block diagram showing the functional configuration.
That is, the matching server SV includes a communication interface unit 10, a control unit 20, and a storage unit 30.

  Under the control of the control unit 20, the communication interface unit 10 communicates the life log data and attributes between the skilled and non-experts' terminals ET1 to ETn and FT1 to FTm via the communication network NW. Data and matching result evaluation data reception processing, and data representing matching results are transmitted.

  The storage unit 30 uses a random accessible non-volatile memory using, for example, an HDD (Hard Disc Drive) or an SSD (Solid State Drive), and a skilled person as a storage unit necessary to realize the present invention. An information database 31 and an unskilled person information database 32 are provided.

  The expert information database 31 is used for storing the life log data and attribute data of the expert collected from the terminals ET1 to ETn of the expert in association with the identification information (ID) of the expert. The non-skilled person information database 32 is used to store the life log data and attribute data of the non-skilled person respectively collected from the terminals FT1 to FTn of the non-skilled person in association with the identification information (ID) of the non-skilled person. The

  The control unit 20 includes a central processing unit (CPU) as a core. As a control function necessary for carrying out the present invention, an expert life log acquisition control unit 21 and an expert attribute acquisition control are provided. A unit 22, a non-skilled person life log acquisition control unit 23, a non-skilled person attribute acquisition control unit 24, a matching processing unit 25, and a weighting adjustment unit 26. All of these control functions are realized by causing the CPU to execute an application program stored in a program memory (not shown) in the storage unit 30.

  The expert life log acquisition control unit 21 controls the communication interface unit 10 to transmit a life log transmission request to the terminals ET1 to ETn used by the expert at different timings. In response to this request, each terminal ET1 The life log data for a predetermined period transmitted from .about.ETn is received and stored in the expert information database 31 in association with the expert ID.

  The expert attribute acquisition control unit 22 controls the communication interface unit 10 to transmit attribute transmission requests to the terminals ET1 to ETn used by the expert at different timings. In response to this request, the terminals ET1 to ETn A process for receiving the current attribute data of the skilled person transmitted from, and storing it in the expert information database 31 in association with the expert ID is executed.

  The non-skilled person life log acquisition control unit 23 controls the communication interface unit 10 to transmit a life log transmission request to the terminals FT1 to FTm of the non-skilled person who requested introduction of the skilled person. A process of receiving life log data for a predetermined period transmitted from each of the terminals FT1 to FTm and storing the life log data in association with the non-skilled person ID in the non-skilled person information database 32 is executed.

  The non-skilled person attribute acquisition control unit 24 controls the communication interface unit 10 to transmit an attribute transmission request to the terminals FT1 to FTm of the non-skilled person who requested the introduction of the skilled person. Processing for receiving the current attribute data of the unskilled person transmitted from FT1 to FTm and storing it in the unskilled person information database 32 in association with the unskilled person ID is executed.

  The matching processing unit 25 performs matching processing between the feature amount of the unskilled person and the feature amounts of the plurality of skilled persons, performs the process of selecting the skilled person having the highest quantitative evaluation index representing the similarity, and the selection result Is transmitted from the communication interface unit 10 to the terminal FTi used by the corresponding unskilled person. The feature amounts of the unskilled person and the plurality of skilled persons are extracted from life log data and attribute data stored in the unskilled person information database 32 and the skilled person information database 31, respectively. As a quantitative evaluation index representing the degree of similarity, DGM (Degree of Gap Matching) is used.

  The weight adjustment unit 26 receives the evaluation data for the matching result from the terminal FTi of the unskilled person who has notified the matching result via the communication interface unit 10. Based on the received evaluation data, a weighting coefficient learning process is performed using a neural network, and the weighting coefficient after the learning process is fed back to the matching processing unit 25 for a rematching process.

Next, the operation of the matching server SV configured as described above will be described.
(1) Expert Life Log Data and Attribute Data Collection Processing In the steady state, the control unit 20 activates the expert life log acquisition control unit 21, and under the control of the expert life log acquisition control unit 21, A life log transmission request is transmitted from the communication interface unit 10 to the terminals ET1 to ETn used by skilled workers registered in advance at different timings. In response to this request, life log data for a predetermined period transmitted from each of the terminals ET1 to ETn, for example, one week in the short case and 10 years in the long case is received via the communication interface unit 10, and this is received. The life log data is stored in the expert information database 31 in association with the expert ID.

  As the life log data, for example, a record of an ability level and a life pattern is used. The ability level includes, for example, the score (420 points or 930 points) in the case of an English qualification test, or the maximum strength 30 kg or 70 kg in the case of athletic ability. In the present embodiment, the ability level is treated as a level vector so that both listening and reading scores, or both muscular strengths of instantaneous power and endurance are reflected in consideration of multi-dimensional cases. Examples of lifestyle patterns include commuting time (0.5 hours and 2.5 hours), sleeping time (6 hours and 8 hours), television viewing time (0.5 hours and 4 hours), and the number of holidays per week (1st or 3rd).

  Subsequently, the control unit 20 activates the expert attribute acquisition control unit 22 and controls the terminals ET1 to ETn used by each of the previously registered experts under the control of the expert attribute acquisition control unit 22. Send attribute transmission requests at different timings. In response to this request, the current attribute data of the expert transmitted from the terminals ET1 to ETn is received via the communication interface unit 10, and the received attribute data is stored in the expert information database 31 with the expert ID. Relate to remember.

  As attribute data, for example, sex (for example, female: 1, male: expressed as 0), age (7 or 35 years old), number of families (1 or 4), and personality are used. Of these, the personality is expressed using, for example, a five-factor model. For example, if each of the 5 factors is a maximum of 10 points, diligence: 7.9, intelligence: 5.2, extroversion: 2.6, emotional stability: 5.1, coordination: 5.3 can do. The five-factor model is described in detail in Goldberg, L.R., “The development of markers for the big-five factor structure”, Psychological Assessment, vol. 4, pp 26-42, 1992.

(2) Acquisition process of life log data and attribute data of unskilled person Assume that a request for introduction of a skilled person is received from the terminal FTi of the unskilled person UFi registered for use in the system. Then, the control unit 20 first transmits a life log transmission request to the terminal FTi of the requesting non-skilled person UFi by the non-skilled person life log acquisition control unit 24. In response to this request, life log data for a predetermined period transmitted from the terminal FTi is received via the communication interface unit 10 and stored in the non-skilled person information database 32 in association with the non-skilled person ID. The content of the life log data acquired at this time is the same as that of the expert described above, but the acquisition target period is from the past one week to one month.

  Subsequently, the control unit 20 transmits an attribute transmission request to the terminal FTi of the requesting non-skilled person UFi by the non-skilled person attribute acquisition control unit 24, and in response to this request, the control unit 20 transmits the attribute of the non-skilled person UFi. The current attribute data is received via the communication interface unit 10 and stored in the non-skilled person information database 32 in association with the non-skilled person ID. The attribute data acquired at this time is the same as that of the expert described above.

(3) Matching process between non-skilled person and a plurality of skilled persons Next, the control unit 20 uses the matching processing unit 25 to perform the matching process of the feature amount between the non-skilled person UFi and the plurality of skilled persons. Perform as follows. FIG. 3 is a flowchart showing the processing procedure and processing contents.

(3-1) Extraction of Feature Value That is, first, the matching processing unit 25, in step S11, is based on the life log data and attribute data of the unskilled person UFi stored in the unskilled person information database 32. The feature amount of UFi is extracted. For example, the score of the English qualification test is extracted as the current ability level vector of the unskilled person. In addition, the values expressed by the five-factor model described above for the character of the unskilled person are added to the values obtained by quantifying the sex, age, and number of family members, and the calculated value is used as the feature amount of the unskilled person UFi To do.

  Next, in step S12, the matching processing unit 25 sequentially reads the life log data and attribute data of a plurality of experts stored in the expert information database 31 for each expert, and based on the read data, Extract the expert features. Also in this case, as in the case of the unskilled person, the current ability level vector and a numerical value based on the personality, gender, age, and the number of family members are extracted as feature quantities for each skilled person. Then, when it is confirmed in step S13 that the skill level vector and feature amount extraction processing has been completed for all skilled workers to be matched, the process proceeds to step S14.

(3-2) Normalization of Feature Amount In step S14, the matching processing unit 25 normalizes the extracted feature amounts of the unskilled person and the plurality of skilled persons. This normalization process is to divide the feature values of the unskilled person and all the skilled persons by the largest one among them so that all values fall within the range of 0 to 1.

(3-3) Calculation of DGM Next, in step S15, the matching processing unit 25 executes calculation of DGM and processing for selecting an expert based on the calculation result as follows.
First, in step S151, the weighting coefficient is adjusted. If the expert has not yet been selected, the weighting coefficient is set to a predetermined initial value. As the initial value, for example, a case where all weighting coefficients are set to “1”, a value which becomes negative (such as below average) for non-experts, for example, a case where the weighting coefficient of feature amount is set to “2”, etc. can be considered. . The reason for increasing the weighting coefficient of the feature amount that is negative for the unskilled person is that it is estimated that the unskilled person sympathizes with the skilled person as the past and the negative part of the skilled person are similar. As an average setting method, for example, a method of calculating from a feature amount of an unskilled person who has used the system so far, and a method of setting using statistical data of a certain group are conceivable.

ただし、

である。 Next, in step S152, the matching processing unit 25 calculates DGM between the unskilled person UFi and the plurality of skilled persons UE1, UE2,. DGM is a quantitative evaluation index that represents the degree of similarity of the feature amount between the unskilled person and the skilled species, and is expressed as in equation (1).

However,

It is.

が最大になるときの値（ｐ_max）である。
また、Ｗ_L[m]は各レベルベクトル要素の重み付け係数を、Ｗ_X[n]は各特徴量の重み付け係数（n個）をそれぞれ表す。 Here, L [m] is a level vector element (m), X [n] is a feature quantity (n), and L and X are subscripts of E, F, non-skilled, and c, respectively. Currently, p represents the past.
p is a set of past life log data accumulated in the expert information database 31,

Is the value (p_max) when becomes maximum.
W _L [m] represents a weighting coefficient of each level vector element, and W _X [n] represents a weighting coefficient (n) of each feature quantity.

の値は大きくなる。また、非熟練者の現在の特徴量と熟練者の過去の特徴量との間の類似度が高ければ高いほど、式(1) の二項目

は大きくなる。 When the DGM is calculated using the above equation (1), the higher the skill level of the skilled person compared with the current ability level of the unskilled person, one item of the expression (1)

The value of increases. In addition, the higher the similarity between the current feature quantity of the unskilled person and the past feature quantity of the skilled person, the higher the two items of Equation (1).

Will grow.

  An example of the DGM calculated by the above equation (1) is shown in FIG. In this example, the life log data and attribute data of the unskilled person UFi are current TOEIC score: 410 points, gender: male, age: 31 years old, number of family members: 4, commuting time: 2 hours. On the other hand, the life log data and attribute data of the expert UE1 are obtained by changing the TOEIC score from the past to the present: 400 points → 925 points, gender: male, age change: 30 years old → 35 years old, family number change. : 4 people-> 4 people, change in commuting time: 1.5 hours-> 0.5 hours, and change of life log data and attribute data of expert UE2 from past to present in TOEIC score: 480 points-> 680 Points, gender: female, change in age: 18 to 21 years old, change in the number of families: 3 to 3 people, change in commuting time: 0.5 hours to 0.5 hours.

  And based on the above life log data and attribute data, when DGM is calculated by the equation (1) between the unskilled person UFi and the skilled person UE1 and UE2, respectively, between the unskilled person UFi and the skilled person UE1 The DGM is “0.74”, and the DGM between the unskilled person UFi and the skilled person UE2 is “0.49”. In this calculation, the weighting coefficient of one item is set to “5” and the weighting coefficient of each feature amount of “2” is set to “1” so that the weighting coefficients of one item and two items in the formula (1) are the same. went.

(3-4) Selection of Experts Subsequently, in step S153, the matching processing unit 25 calculates the value of each DGM between the unskilled person UFi calculated in step S152 and the plurality of experts TE1, UE2,. And the expert UEj having the maximum DGM value is selected. For example, if the DGM value with the expert UE1 illustrated in FIG. 5 is the maximum, the expert UE1 is selected as an expert to be introduced to the non-skilled person UFi.

(4) Transmission of matching result When an expert to be introduced is selected in step S153, the control unit 20 generates notification information including life log data and attribute data of the selected expert in step S16. Then, the generated information is transmitted as data representing the matching result from the communication interface unit 10 to the terminal FTi of the unskilled person UFi. For example, e-mail is used as the transmission means at this time. Instead of using e-mail, the unskilled person UFi can acquire data representing the matching result by accessing and browsing the web page of the matching server SV using the browser function of the terminal FTi. Good.

(5) Adjustment of weighting and recalculation of DGM When the data representing the matching result is transmitted from the matching server SV to the terminal FTi of the unskilled person UFi, the unskilled person UFi can use the notified matching result data for himself / herself. It is determined whether or not. When an expert with a higher or lower feature amount is desired, data for requesting a change of the feature amount is input as evaluation data while the Web page of the matching server SV is opened by the browser function.

  The control unit 20 of the matching server SV monitors the input operation of the evaluation data at the terminal FTi of the unskilled person UFi in step S21 in the weight adjustment unit 26 as shown in FIG. When the evaluation data is input to the terminal FTi of the unskilled person UFi, this evaluation data is received via the communication interface unit 10, and a weighting coefficient learning process is performed using a neural network in step S22. Then, the new weighting coefficient obtained by the weighting coefficient learning process is given to the matching processing unit 25 in step S23.

The matching processing unit 25 recalculates the DGM between the unskilled person UFi and the plurality of skilled persons UE1, UE2,... Using the adjusted weighting coefficient given above. Then, the skilled person UEk having the largest DGM is selected from the calculated DGMs, and the life log data and attribute data are transmitted to the terminal FTi of the unskilled person UFi by e-mail.
The above-described weighting adjustment processing, DGM recalculation processing, expert reselection processing, and result transmission processing are repeatedly performed each time different evaluation data is sent from the terminal FTi of the unskilled user UFi.

  As described above in detail, in the first embodiment, life log data and attribute data of a plurality of skilled workers are collected from the skilled worker's terminals ET1 to ETn and stored in the skilled worker information database 31. In this state, the life log data and attribute data of the unskilled person UFi are acquired from the terminal FTi of the unskilled person, the feature amount is extracted from these life log data and attribute data, Feature values are extracted from the life log data and attribute data of the expert. Then, the relationship between the extracted feature quantities is calculated by the DGM calculation formula between the unskilled person UFi and a plurality of skilled persons, and the expert who maximizes the DGM is selected and its life log data The attribute data is transmitted to the terminal FTi of the unskilled person UFi.

  Specifically, the DGM is calculated by the above equation (1). That is, the greater the difference in the current ability level between the unskilled person and the skilled person, the higher the DGM, and the degree of similarity between the current feature quantity of the unskilled person and the past feature quantity of the skilled person The calculation is such that the higher the value is, the higher the value of DGM is.

  Therefore, in the first embodiment, the feature amount reflecting both the current ability level difference between the unskilled person and the skilled person, the current life log data and attribute data of the unskilled person, and the past of the skilled person A DGM representing a relationship between the unskilled person and the skilled person is obtained based on the similarity between the feature amount reflecting both the life log data and the attribute data. Then, the information of the skilled person who maximizes the DGM is notified to the unskilled person UFi.

  In other words, although the skill level is much higher than that of the unskilled person UFi, the skilled person who has similar circumstances in the past is selected by the matching process and presented to the unskilled person UFi. For this reason, it becomes possible to give the unskilled person UFi a willingness to “raise the level like that person”. Moreover, since the unskilled person UFi is presented with information on a skilled person who has a similar situation in the past with the current self, the unskilled person UFi can feel that the skilled person is familiar. You can expect an effect that makes you feel like you can improve your level like a person, or "you can do it yourself".

  In the first embodiment, the evaluation data transmitted from the unskilled person UFi is received in response to the notification of the data representing the matching result, and the level and the feature amount when calculating the DGM based on the evaluation data are received. The weighting coefficient is adjusted, the DGM is recalculated, and the expert is selected again. Therefore, information about the skilled person whose feature amount is closer to that of the unskilled person UFi is presented, and thereby the motivation for improvement of the unskilled person UFi can be more effectively enhanced.

ここで、太字のＬはレベルベクトル、太字のＸは特徴ベクトルとする。レベルベクトルＬ及び特徴ベクトルＸに含まれる特徴量は式(1) と同様である。Sim は類似度を表し、この類似度Sim の計算にはユークリッド距離、ピアソン相関係数、コサイン類似度等が用いられる。
また、熟練者の現在のレベル及び特徴量は熟練者の実力レベルが最大だった過去の特徴量に置き換えることもできる。 (Second Embodiment)
In the first embodiment, the weighted sum is calculated when calculating the DGM. However, the present invention is not limited to this. For example, the form of the product may be considered as shown in the equation (3).

Here, bold L is a level vector, and bold X is a feature vector. The feature amounts included in the level vector L and the feature vector X are the same as in the equation (1). Sim represents similarity, and Euclidean distance, Pearson correlation coefficient, cosine similarity, and the like are used to calculate similarity Sim.
Further, the current level and feature amount of the skilled person can be replaced with a past feature amount in which the skill level of the skilled person is maximum.

(Third embodiment)
In the third embodiment of the present invention, the growth rate (efficiency) is calculated based on the current skill level of the expert and the time required to obtain the skill level, and for each past feature quantity of the expert. A growth difficulty level representing the difficulty of growth is calculated, and a quantitative evaluation index EP (Expert Property) that increases as the calculated growth rate and growth difficulty level increase. Then, an expert who maximizes the combination of the quantitative evaluation index DGM (Degree of Gap Matching) described in the first embodiment and the calculated quantitative evaluation index EP is selected.

  The matching server SV according to the third embodiment of the present invention has the following new functions. Also in the third embodiment, referring to FIG. 2, detailed description of the same parts as those of the first embodiment will be omitted.

  That is, the matching processing unit 25 ′ of the matching server SV has the following processing functions. All of these processing functions are realized by causing the CPU to execute application programs stored in a program memory (not shown) in the storage unit 30.

(1) A process of extracting feature values of the unskilled person and the plurality of skilled persons from the life log data and attribute data stored in the unskilled person information database 32 and the skilled person information database 31, respectively.
(2) Based on the difference in the current skill level of the extracted unskilled person and each skilled person, and the similarity between the current feature quantity of the extracted non-skilled person and the past feature quantity of each skilled person In addition, a process of calculating a quantitative evaluation index DGM, which has a larger value as the degree of skill is greater and the degree of similarity is higher, for each combination of the unskilled person and each skilled person.

(3) Growth rate expressed by the current skill level of the extracted expert and the time taken to obtain the skill level, and difficulty of growth for each feature amount of the extracted expert A process of calculating a quantitative evaluation index EP for each combination of a non-experienced person and a plurality of skilled persons, based on the degree of difficulty of growth representing the degree of growth.
(4) The weighted addition value is calculated for all combinations of the quantitative evaluation index DGM and the quantitative evaluation index EP calculated by (2) and (3) above, and the calculated weighted addition value is maximized. Processing for selecting a skilled person having a combination of the quantitative evaluation index DGM and the quantitative evaluation index EP.
(5) Processing for transmitting the information on the selected skilled person from the communication interface unit 10 to the terminal FTi used by the corresponding unskilled person.

  Next, the operation of the matching process between the unskilled person and the plurality of skilled persons by the matching server SV configured as described above will be described. This matching process is performed as follows by the matching processing unit 25 ′ in the control unit 20. FIG. 6 is a flowchart showing the processing procedure and processing contents. In FIG. 6, the same processing parts as those in FIG.

(6-1) Extraction of Feature Value That is, first, the matching processing unit 25 ′ is based on the life log data and attribute data of the unskilled person UFi stored in the unskilled person information database 32 in step S 11. The feature amount of the person UFi is extracted. For example, the score of the English qualification test is extracted as the current ability level vector of the unskilled person. In addition, the values expressed by the five-factor model described above for the character of the unskilled person are added to the values obtained by quantifying the sex, age, and number of family members, and the calculated value is used as the feature amount of the unskilled person UFi. To do.

  Next, in step S12, the matching processing unit 25 ′ sequentially reads out the life log data and attribute data of the plurality of experts stored in the expert information database 31 for each expert, and based on the read data. Then, the feature amount of the expert is extracted. Also in this case, as in the case of the unskilled person, the current ability level vector and a numerical value based on the personality, gender, age, and the number of family members are extracted as feature quantities for each skilled person. Then, when it is confirmed in step S13 that the skill level vector and feature amount extraction processing has been completed for all skilled workers to be matched, the process proceeds to step S14.

(6-2) Normalization of Feature Amount In step S14, the matching processing unit 25 ′ normalizes the extracted feature amounts of the unskilled person and the plurality of skilled persons. This normalization process is a process for dividing each feature amount of the unskilled person and all the skilled persons by the largest one of them so that all values fall within the range of 0 to 1.

Next, the matching processing unit 25 ′ proceeds to step S 31 and executes a DGM calculation, an EP calculation, and an expert selection process based on the calculation results as follows.
(6-3) Calculation of DGM First, in step S151, the matching processing unit 25 ′ determines weighting coefficients used in the calculation of DGM as follows. That is, in a state where the expert has not yet been selected, the weighting coefficient is set to a predetermined initial value. As the initial value, for example, a case where all weighting coefficients are set to “1”, or a value that becomes negative (such as an average value or less) for non-experts, for example, a case where the weighting coefficient of a feature amount is set to “2” is considered. It is done. The reason for increasing the weighting coefficient of the feature amount that is negative for the unskilled person is that it is estimated that the unskilled person sympathizes with the skilled person as the past and the negative part of the skilled person are similar. As an average value setting method, for example, a method of calculating from a feature amount of an unskilled person who has used the system so far, and a method of setting using statistical data of a certain group are conceivable.

  Further, as shown in FIG. 4, the weighting coefficient is obtained by receiving evaluation data input by the unskilled person UFi from the terminal FTi of the unskilled person UFi and weighting using the neural network based on the received evaluation data. It is also possible to perform a coefficient learning process and use the weighting coefficient obtained by this learning process.

  Next, in step S152, the matching processing unit 25 ′ calculates DGM between the unskilled person UFi and the plurality of skilled persons UE1, UE2,. The DGM is a quantitative evaluation index that represents the relationship between the feature amounts between the unskilled person and the skilled species, and is calculated by the equation (1) described in the first embodiment. When the DGM is calculated using the above equation (1), the value of one item of the equation (1) becomes larger as the skill level of the skilled worker is higher than the current ability level of the unskilled worker. In addition, the higher the similarity between the current feature quantity of the non-expert and the past feature quantity of the expert, the greater the value of the two items in the equation (1).

(6-4) Calculation of EP Next, in step S311, the matching processing unit 25 ′ determines the weighting coefficient used in the calculation of EP as follows. That is, this EP weighting coefficient is also set to a fixed value as an initial value in the same manner as the DGM weighting coefficient described above. As this fixed value, for example, “1” is used.

ここで、ｐ_maxは第１の実施形態で述べた式(2)により表される値であり、ｔは熟練者の過去から現在までの総努力時間、Ｄ[n]は特徴量ごとの熟練者の成長困難度、Ｗ_T[m]は各レベルベクトル要素の効率の重み付け係数、Ｗ_D[n]は各特徴量の成長困難度の重み付け係数を表す。 Subsequently, the matching processing unit 25 ′ calculates an EP for each expert in step S312. As described above, the EP is the growth rate represented by the current skill level of the expert and the time required to obtain the skill level, and the growth of each of the extracted feature values of the expert in the past. It is calculated on the basis of the difficulty level of growth that represents the difficulty of the above, and is expressed by the following equation.

Here, p_max is a value represented by the formula (2) described in the first embodiment, t is the total effort time from the past to the present of the expert, and D [n] is the expert for each feature amount. W _T [m] is a weighting coefficient for the efficiency of each level vector element, and W _D [n] is a weighting coefficient for the growth difficulty of each feature quantity.

  The growth difficulty level D [n] is set for each feature amount of the expert. For example, when the feature amount is “age”, “relationship between aging and physical strength” is used in a field that uses the body, and “relationship between aging and memory” is used in a field that requires memory. FIG. 8 is a graph showing an example of the relationship between aging and physical strength, and shows that physical strength declines with aging.

  When the growth difficulty level is derived using the relationship between aging and physical strength shown in FIG. 8, all physical strength values fall within the range of 0 to 1 by dividing the physical strength value of each age by the maximum value. The normalized value is norm (x). Then, 1−norm (x) which is the reciprocal of the normalized value norm (x) is obtained, and this value is set as the growth difficulty level. FIG. 9 is a graph showing an example of the growth difficulty derived from the relationship between the aging and physical strength.

  In addition, when the growth difficulty is calculated for a life pattern related to time, such as commuting time, sleep time, and TV viewing time, the difficulty increases as time not related to the skilled field increases. The value of pattern (h) / 24 (h)) is defined as the growth difficulty level. In addition, for the feature quantity for which it is difficult to assume the growth difficulty level, the growth difficulty level is ignored. That is, the difficulty level is set to 0 regardless of the feature value.

  “Relationship between aging and physical fitness” was published by the Ministry of Education, Culture, Sports, Science and Technology, “FY 2008 Physical Fitness / Exercise Survey”, and “Relationship between aging and memory” by PB Baltes, UM Staudinger, Wisdom: A metaheuristic (pragmatic) to orchestrate mind and virtue toward excellence ”The American psychologist, vol.55, pp122-136, 2000.

(6-5) Selection of expert Next, in step S313, the matching processing unit 25 ′ calculates weighted sums for all combinations of DGM and EP calculated in steps S152 and S312 with the following equations, respectively. To do.
_{W 1 · DGM + W 2 ·} EP
For the weighting coefficients W ₁ and W ₂ , the initial value of the fixed value “1” previously set in step S311 is assigned.
Then, based on the calculation result of W ₁ · DGM + W ₂ · EP, an expert who has the largest calculated value is selected.

In addition to using the above weighted sum (W ₁ · DGM + W ₂ · EP), the EP is normalized between 0.8 and 1.0 to calculate DGM × EP, and the calculation result of DGM × EP is also obtained. In addition, an expert who has the largest calculated value may be selected.

(6-5) Transmission of Matching Result When the expert to be introduced in step S313 is selected, the control unit 20 generates notification information including life log data and attribute data of the selected expert, This generated information is transmitted as data representing the matching result from the communication interface unit 10 to the terminal FTi of the unskilled person UFi. For example, e-mail is used as the transmission means at this time. Instead of using e-mail, the unskilled person UFi can acquire data representing the matching result by accessing and browsing the web page of the matching server SV using the browser function of the terminal FTi. Good.

(6-6) Learning of Weighting Factor in Weighting Adjustment Unit 26 When data representing the matching result is transmitted from the matching server SV to the terminal FTi of the unskilled person UFi, the unskilled person UFi receives the data of the notified matching result. Judge whether it is useful or not. When an expert with a higher or lower feature amount is desired, data for requesting a change of the feature amount is input as evaluation data while the Web page of the matching server SV is opened by the browser function. For example, evaluation data indicating that the user does not like the presented expert (desiring an expert with a higher or lower characteristic amount) is input.

  The control unit 20 of the matching server SV monitors the input operation of the evaluation data at the terminal FTi of the unskilled person UFi in step S41, as shown in FIG. When the evaluation data is input to the terminal FTi of the unskilled person UFi, the evaluation data is received via the communication interface unit 10, and the learning process of the weighting coefficient for EP is performed using the neural network in step S42. . Multiple learning analysis is used for this learning process. Then, the new EP weighting coefficient obtained by the weighting coefficient learning process is given to the matching processing unit 25 'in step S43.

(6-7) Recalculation processing of DGM and EP The matching processing unit 25 ′ uses the adjusted weighting coefficient for DGM given from the weighting adjustment unit 26, and in step S 152, the unskilled person UFi and each skilled person Recalculate the DGM between UE1, UE2,. At the same time, in step S312, the adjusted EP weighting coefficients given from the weighting adjustment unit 26 are used to recalculate the EPs of the skilled workers UE1, UE2,. Then, for all combinations of the calculated DGM and EP, a weighted sum (W ₁ · DGM + W ₂ · EP) or a product DGM × EP is calculated, and the calculated value is the largest based on the calculation result. Reselect the skilled person.
Then, notification information including life log data and attribute data of the selected skilled person is generated, and the generated information is transmitted from the communication interface unit 10 to the terminal FTi of the unskilled person UFi as data representing a matching result. Send by email.

  The recalculation process of DGM and EP, the reselection process of the expert, and the transmission process of the result are repeatedly performed every time different evaluation data is sent from the terminal FTi of the unskilled person UFi.

  As described above in detail, in the third embodiment, the matching processing unit 25 ′ uses the current skill level of each expert UE1, UE2,... And the time t required to obtain the skill level (the growth rate ( Efficiency) and a growth difficulty level D [n] representing the difficulty of growth for each past feature quantity of the skilled person UE1, UE2,..., And the calculated growth rate and growth difficulty level D. The quantitative evaluation index EP, which increases as [n] increases, is calculated according to the equation (4). Then, an expert who has the maximum combination of the quantitative evaluation index DGM described in the first embodiment and the calculated quantitative evaluation index EP is selected.

  Therefore, an expert is selected in consideration of not only DGM but also EP. In other words, a skilled person who has a higher growth rate than a non-skilled person but has a high growth rate even though the past situation is similar to that of a non-skilled person and the skill level is difficult to improve is selected. The For this reason, the unskilled person is given a sense of familiarity, and at the same time motivates the non-skilled person to “raise the level like that person” and “want to take action to raise the level” Is possible.

(Other embodiments)
Note that the present invention is not limited to the above-described embodiments, and the configuration of the matching server, its processing functions, processing procedures and processing contents can be variously modified without departing from the spirit of the present invention. It is.
In short, the present invention is not limited to the above-described embodiments as they are, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. For example, some components may be deleted from all the components shown in each embodiment. Furthermore, you may combine suitably the component covering different embodiment.

  SV ... matching server, NW ... communication network, ET1-ETn ... mobile terminals used by skilled workers, FT1-FTm ... mobile terminals used by non-experts, 10 ... communication interface unit, 20 ... control unit, 21 ... expert Life log acquisition control unit, 22 ... expert attribute acquisition control unit, 23 ... non-skilled person life log acquisition control unit, 24 ... non-skilled person attribute acquisition control unit, 25, 25 '... matching processing unit, 26 ... weighting adjustment unit 30 ... Storage unit, expert information database, 32 ... non-skilled person information database.

Claims (19)

Means for collecting and storing life log data and attribute data of a plurality of skilled workers from the past to the present;
Means for acquiring current life log data and attribute data of the unskilled person from the terminal of the unskilled person,
Means for extracting the current skill level of the unskilled person and each skilled person from the acquired unskilled person life log data and the stored life log data of each skilled person, respectively;
Extracting a feature quantity representing the current situation of the unskilled person from the acquired attribute data of the unskilled person, and extracting a feature quantity representing the past situation of each skilled person from the stored attribute data of each skilled person Means for extracting;
Based on the difference in the current skill level of the extracted non-skilled person and each skilled person, and the degree of similarity between the current feature quantity of the extracted non-skilled person and the past feature quantity of each skilled person The first quantitative evaluation index is calculated for each combination of the unskilled person and the plurality of skilled persons, the first quantitative evaluation index having a larger value as the difference in skill level is larger and the similarity degree is higher. Exponent calculation means;
Based on the calculated first quantitative evaluation index, an expert who satisfies a predetermined condition is selected from the plurality of experts, and information on the selected expert is sent to the terminal used by the non-skilled person. A person matching apparatus comprising: an expert selection means for transmitting. The first quantitative evaluation index calculation means includes:

And p in the stored set of past life log data,

Is the maximum value (p_max), L [m] is the element of the level vector (m), X [n] is the feature quantity (n), and E and F that are subscripts of L and X are Expert and non-skilled person, c as current, p as past, W _L [m] as weighting coefficient for each level vector element, W _X [n] as weighting coefficient for each feature quantity (n) When representing the first quantitative evaluation index DGM

The person matching apparatus according to claim 1, wherein the person matching apparatus is calculated by: The first quantitative evaluation index calculation means includes:
Means for receiving evaluation data for the transmitted information regarding the skilled person from the terminal of the unskilled person;
The person matching apparatus according to claim 2, further comprising: means for changing a weighting coefficient W _X [n] of each feature amount in the equation (1) based on the received evaluation data. The first quantitative evaluation index calculation means includes:
When L is a level vector, X is a feature vector, and the feature quantity included in these level vector L and feature vector X is expressed by the above equation (1), and Sim is a similarity, the first quantitative evaluation index DGM The

The person matching apparatus according to claim 2, wherein the person matching apparatus is calculated by the following. For each of the plurality of skilled workers, the growth rate represented by the extracted current skill level and the time required to obtain the skill level, and the difficulty of growth for each extracted past feature amount And calculating a second quantitative evaluation index that increases as the calculated growth rate and growth difficulty increase, respectively, for each of the unskilled person and the plurality of skilled persons. A second quantitative evaluation index calculation means for calculating the combination;
The expert selection means selects an expert who satisfies a predetermined condition from the plurality of experts based on the calculated first and second quantitative evaluation indexes, and information on the selected expert The person matching apparatus according to claim 1, wherein the person matching device is transmitted to a terminal used by the unskilled person. The second quantitative evaluation index calculation means includes:

And p in the stored set of past life log data,

Is the maximum value (p_max), L [m] is the element of the level vector (m), X [n] is the feature quantity (n), and E and F that are subscripts of L and X are Each skilled and non-skilled person, c as the present, p as the past, W _L [m] as the weighting coefficient for each level vector element, W _X [n] as the weighting coefficient for each feature quantity (n), Furthermore, t is the total effort time from the past p to the current c of the skilled person, D [n] is the difficulty level of the skilled person for each feature, W _T [m] is the weight of the growth rate for each level vector element, When W _D [n] is the weight of the growth difficulty of each feature amount, the second quantitative evaluation index EP is

The person matching apparatus according to claim 5, wherein the person matching apparatus is calculated by: The expert selection means includes
When the weighting coefficient is W _1, to the weighting coefficient W ₂ together with the second quantitative evaluation index and EP as well as the DGM the first quantitative evaluation index,
W _{1 ·} DGM + W ₂ · EP
6. The person matching apparatus according to claim 5, wherein an expert who satisfies a predetermined condition is selected from the plurality of experts based on the calculated value. The expert selection means includes
8. The apparatus according to claim 7, further comprising means for receiving an input of an evaluation result of an unskilled person for the selection result and variably setting the weighting factors W ₁ and W ₂ based on the input evaluation result. The person matching device described. The expert selection means includes
When the first quantitative evaluation index is DGM and the second quantitative evaluation index is EP, the first and second quantitative evaluation indices DGM and EP are normalized, and the normalized DGM and EP are normalized. Product DGM × EP
6. The person matching apparatus according to claim 5, wherein an expert who satisfies a predetermined condition is selected from the plurality of experts based on the calculated value. A process of collecting and storing life log data and attribute data of a plurality of skilled workers from the past to the present,
The process of acquiring the current life log data and attribute data of the unskilled person from the terminal of the unskilled person,
A process of extracting the current skill level of each unskilled person and each skilled person from the acquired unskilled person life log data and the stored life log data of each skilled person,
Extracting a feature quantity representing the current situation of the unskilled person from the acquired attribute data of the unskilled person, and extracting a feature quantity representing the past situation of each skilled person from the stored attribute data of each skilled person Extraction process,
Based on the difference in the current skill level of the extracted non-skilled person and each skilled person, and the degree of similarity between the current feature quantity of the extracted non-skilled person and the past feature quantity of each skilled person The first quantitative evaluation index is calculated for each combination of the unskilled person and the plurality of skilled persons, the first quantitative evaluation index having a larger value as the difference in skill level is larger and the similarity degree is higher. Exponential calculation process,
Based on the calculated first quantitative evaluation index, an expert who satisfies a predetermined condition is selected from the plurality of experts, and information on the selected expert is sent to the terminal used by the non-skilled person. A person matching method comprising: an expert selection process for transmission. The first quantitative evaluation index calculation process includes:

And p in the stored set of past life log data,

Is the maximum value (p_max), L [m] is the element of the level vector (m), X [n] is the feature quantity (n), and E and F that are subscripts of L and X are Expert and non-skilled person, c as current, p as past, W _L [m] as weighting coefficient for each level vector element, W _X [n] as weighting coefficient for each feature quantity (n) When representing the first quantitative evaluation index DGM

The person matching method according to claim 10, wherein the person matching method is calculated by: The first quantitative evaluation index calculation process includes:
A process of receiving evaluation data for the transmitted information about the skilled person from the unskilled person's terminal;
The person matching method according to claim 11, further comprising a step of varying a weighting coefficient W _X [n] of each feature amount in the equation (1) based on the received evaluation data. The first quantitative evaluation index calculation process includes:
When L is a level vector and X is a feature vector, the feature quantity contained in these level vector L and feature vector X is expressed by equation (1), and Sim is a similarity, the first quantitative evaluation index DGM is

The person matching method according to claim 11, wherein the person matching method is calculated by: For each of the plurality of skilled workers, the growth rate represented by the extracted current skill level and the time required to obtain the skill level, and the difficulty of growth for each extracted past feature amount And calculating a second quantitative evaluation index that increases as the calculated growth rate and growth difficulty increase, respectively, for each of the unskilled person and the plurality of skilled persons. A second quantitative evaluation index calculation process for calculating the combination;
The expert selection process selects an expert who satisfies a predetermined condition from the plurality of experts based on the calculated first and second quantitative evaluation indexes, and information on the selected expert The person matching method according to claim 10, further comprising: transmitting a message to a terminal used by the unskilled person. The second quantitative evaluation index calculation process includes:

And p in the stored set of past life log data,

Is the maximum value (p_max), L [m] is the element of the level vector (m), X [n] is the feature quantity (n), and E and F that are subscripts of L and X are Each skilled and non-skilled person, c as the present, p as the past, W _L [m] as the weighting coefficient for each level vector element, W _X [n] as the weighting coefficient for each feature quantity (n), Furthermore, t is the total effort time from the past p to the current c of the skilled person, D [n] is the difficulty level of the skilled person for each feature, W _T [m] is the weight of the growth rate for each level vector element, When W _D [n] is the weight of the growth difficulty of each feature amount, the second quantitative evaluation index EP is

The person matching method according to claim 14, wherein the person matching method is calculated by: The expert selection process includes:
When the weighting coefficient is W _1, to the weighting coefficient W ₂ together with the second quantitative evaluation index and EP as well as the DGM the first quantitative evaluation index,
W _{1 ·} DGM + W ₂ · EP
The person matching method according to claim 14, wherein an expert who satisfies a predetermined condition is selected from the plurality of experts based on the calculated value. The expert selection process includes:
17. The method according to claim 16, further comprising a step of receiving an input of an evaluation result of an unskilled person for the selection result and variably setting the weighting factors W ₁ and W ₂ based on the input evaluation result. The person matching method described. The expert selection process includes:
When the first quantitative evaluation index is DGM and the second quantitative evaluation index is EP, the first and second quantitative evaluation indices DGM and EP are normalized, and the normalized DGM and EP are normalized. Product DGM × EP
The person matching method according to claim 14, wherein an expert who satisfies a predetermined condition is selected from the plurality of experts based on the calculated value.   The program which makes a computer perform the process which implement | achieves the means with which the person matching apparatus in any one of Claims 1 thru | or 9 is provided.

Images