JP6343079B1 - Method and apparatus for presenting a task that matches a person, and method and apparatus for presenting a person that matches a task - Google Patents

Abstract

A method and apparatus for presenting a task that matches a person to help more efficiently match an employee with a task. A method for presenting a task that matches an employee to an employee, the step of obtaining a feature vector of the skill of the employee, a step of obtaining a feature vector of the task, A step of calculating the similarity between the feature vector and the feature vector of the task, a step of calculating the similarity to the feature vector of the plurality of tasks, and among the feature vectors of the task for which the similarity is calculated, A method for presenting a task that matches an employee, comprising: presenting a task with a feature vector of a task having a high similarity to the employee skill feature vector to the employee. [Selection] Figure 1

Description

  The present invention relates to a method and apparatus for presenting a task that matches a person, and a method and apparatus for presenting a person that matches a task. In particular, it relates to a method and apparatus used to assign tasks to employees.

When executing a certain job, it is necessary to assign “tasks” constituting the job to employees (persons).
Such assignment work is classically done exclusively by an artificial method, for example, an administrator interviews each employee, asks them about their skills and wishes, and based on the results, the boss There has been much done assigning tasks to each employee.

However, in recent years, in order to efficiently advance a large-scale project or the like, many mechanisms have been proposed that can grasp the entire progress of the project and efficiently allocate task human resources.
For example, Patent Document 1 below discloses a system that allows the progress of the entire project to be browsed and the work status and load to be ascertained. According to the system described in Patent Document 1, it is possible to estimate a task that requires a load and a period during which the load is applied, adjust a schedule, secure personnel, and the like, and smoothly advance the project. Has been.

JP 2004-118648 A

As described above, a system that grasps the status of a project and appropriately determines the allocation to employees based on the situation has been widely used.
On the other hand, in order to appropriately assign tasks to employees, the inventor of the present invention considers that it is important to compare and examine employee skills and tasks in more detail. No such system is known.
The present invention has been made in view of the above circumstances, and matches a person who helps to more effectively compare employee skills and task contents and more efficiently match employees and tasks. It is an object to provide a method and apparatus for presenting a task and a method and apparatus for presenting a person who matches the task.

  A first aspect of the present invention is a method for presenting a task that matches a person to a person, the step of obtaining a feature vector of the skill of the person, the step of obtaining a feature vector of the task, and the person Obtaining the similarity between the feature vector of the skill and the feature vector of the task, executing the step of obtaining the similarity on the feature vector of a plurality of tasks, and obtaining the similarity Presenting a task that matches a person including the step of presenting the task of the feature vector of the task having a large similarity to the feature vector of the skill of the person among the feature vectors of the task to the person Is the method.

  According to a second aspect of the present invention, in the invention according to claim 1, the feature vector of the task includes one or more tagging for the task, a level required for each of the tagged tags, It is a vector consisting of

  According to a third aspect of the present invention, in the invention according to claim 1 or 2, the person's feature vector includes one or more taggings for the person and the person for each tagged tag. Is a vector consisting of a level that is currently maintained.

  According to a fourth aspect of the present invention, in the invention according to any one of claims 1 to 3, the similarity is a cosine similarity between a feature vector of the person's skill and a feature vector of the task. It is characterized by being.

  5th this invention is an apparatus which shows the task which matches the said person with respect to the person, Comprising: The skill information database which registers the feature vector of the said person's skill, and the characteristic of one or two or more said tasks A task information database for registering a vector; a feature vector of the person's skill is extracted from the skill information database; a plurality of feature vectors of the task are extracted from the task information database; A skill-task similarity calculation unit for obtaining a similarity between the extracted feature vector of the task and the task feature vector for which the similarity is obtained, and the similarity with the feature vector of the person skill is large. A recommended task presenting unit that presents the task of the feature vector of the task to the person. It is a device that presents a task to match.

  6th this invention is the method of presenting the person who matches the said target task with respect to an object task, Comprising: The step which calculates | requires the feature vector of the task digest performance of the said one person or several persons, and the said object A step of obtaining a task feature vector, a step of obtaining an inter-task similarity between the feature vector of the person's task digestion result, and a feature vector of the target task, and a step of obtaining the inter-task similarity. A step of executing a plurality of feature digestion vector of the person's task digestion, and a step of presenting the person having the task digestion feature vector having a high similarity between the tasks to the target task. It is a method of presenting the person who matches the task.

  According to a seventh aspect of the present invention, in the invention according to claim 6, the feature vector of the person's task achievement record is a feature vector of a plurality of past record tasks of the person, and the similarity between tasks is the target task. The similarity between the feature vector of the person and the feature vector of the plurality of past achievement tasks of the person is obtained for each feature vector of the plurality of past achievement tasks, and the total similarity is a total value of the obtained plurality of similarities. It is characterized by being.

  According to an eighth aspect of the present invention, in the invention according to claim 6, the feature vector of the person's task achievement record is a feature vector of a plurality of past record tasks of the person, and the similarity between tasks is the target task. And a similarity between the feature vector of the plurality of past performance tasks of the person and the feature vector of the plurality of past performance tasks of the person, and a predetermined reference value or more of the plurality of similarities obtained The total similarity is a total value obtained by adding only the similarities.

  The ninth aspect of the present invention is an apparatus for presenting a person who matches the target task with respect to the target task, and registers a feature vector of task digest results of one or more persons. A task information database that registers the feature vector of the target task, a feature vector of the person's task digestion result, and an inter-task similarity calculation unit that calculates an inter-task similarity between the feature vector of the target task, An apparatus for presenting a person who matches a task, including a recommended target employee extracting unit that presents the person having a feature vector of a task digest performance record having a high similarity between tasks to the target task.

  A tenth aspect of the present invention is a method for presenting a task that matches a person to a person, the step of obtaining a feature vector of the task digestion result of the person, the step of obtaining a feature vector of the task, The step of obtaining the similarity between tasks between the feature vector of the task digestion result of the person and the feature vector of the task and the step of obtaining the similarity between the tasks are executed on the feature vectors of the plurality of tasks. Co-occurring with a cluster group to which a predetermined one of the tasks belongs in the step, the task for which the similarity between tasks is obtained is decomposed into cluster groups, and the task for which the similarity between tasks is obtained Extracting another cluster group having high characteristics, and among the feature vectors of the task belonging to the extracted cluster group, A method of presenting a task that matches a person, including the step of presenting the certain predetermined task to the person when the similarity between the tasks with the feature vector of the result of the digestion of the risk is large. is there.

  An eleventh aspect of the present invention is characterized in that, in the invention according to claim 10, the step of decomposing into the cluster group decomposes the task into a plurality of cluster groups by performing cluster analysis.

  In a twelfth aspect of the present invention, in the invention according to claim 10 or 11, the co-occurrence includes a first set of persons who have executed a task belonging to the predetermined cluster group and tasks belonging to the other cluster group. The number of people who belong to both the first set and the second set in relation to the number of people who belong to the entire set of the first set and the second set with respect to the second set with the executed person It is the ratio of this.

  A thirteenth aspect of the present invention is an apparatus for presenting a task that matches a person to a person, the task result information database for registering the feature vector of the person's task digest record, and one or more of the above A task information database for registering a feature vector of a task, a feature vector of a person's task digestion result, a task-task similarity calculation unit for calculating a similarity between tasks of a plurality of feature vectors of the task, and the task The task whose inter-similarity is obtained is decomposed into cluster groups, and among the tasks whose inter-task similarity is obtained, another cluster group having high co-occurrence with a cluster group to which a predetermined one task belongs A co-occurrence calculation unit that extracts the feature, and a feature vector of the task digestion result of the person among the feature vectors of the task belonging to the cluster group Wherein when the inter-task similarity is large, the predetermined one of the tasks of the present, a device for presenting the task of matching the person including a recommended task presenting unit to be presented to the person.

  A fourteenth aspect of the present invention is a method for obtaining the degree of difficulty of a person for a task, wherein the length of the projection when the feature vector of the task is projected onto the feature vector of the person's skill is represented by the feature of the skill. And calculating a difficulty level of a person for a task including a step of calculating a value obtained by dividing the vector length as the difficulty level.

  A fifteenth aspect of the present invention is an apparatus for determining the level of difficulty of a person with respect to a task, wherein the length of the projection when the feature vector of the task is projected onto the feature vector of the person's skill is represented by the feature of the skill. It is a device for obtaining a person's difficulty level for a task including a calculation unit that calculates a value obtained by dividing the vector length as the difficulty level.

  According to the present invention, a task suitable for the skill can be presented to a person, and more efficient task assignment can be performed.

It is explanatory drawing of the similarity degree of the feature vector 6 of a person's skill, and the feature vector 8 of a task. 2 is a functional block diagram of a task and employee matching apparatus 10 according to the first embodiment. FIG. 10 is a table showing priorities of the second embodiment. 6 is a functional block diagram of a task and employee matching apparatus 110 according to Embodiment 2. FIG. It is explanatory drawing which shows the mode of the cluster analysis of Embodiment 3 typically. FIG. 10 is a functional block diagram of a task and employee matching apparatus 210 according to a third embodiment.

DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, embodiments of the invention will be described with reference to the drawings.
The present embodiment relates to a mechanism for automatically recommending a task that matches an employee. For example, each configuration of the following first to fourth embodiments is preferably realized by using a computer.
Embodiment 1. FIG. Recommendation Using Employee Skill and Task Similarity (Similarity) In the first embodiment, an example of a method for recommending a task based on the employee skill and task similarity (similarity) will be described.
(1) Features First, the features of this method are as follows.
(A) First, the degree of matching between the skill of the employee and the characteristics of the task is quantified.
(B) The recommendation principle according to the first embodiment is an application of the “content-based filtering” technique used in the recommendation engine. Content-based filtering is a technique for obtaining a similarity based on a feature vector of an object person or an object and making a recommendation based on the result.
(C) Here, feature vector of employee skill

And the task feature vector

It expresses.

The task feature vector is defined by the tagging for the task and the level required for each tag. For example,
{Excel: Advanced level, Accounting knowledge: Beginner level}
It is expressed as In the example shown here, it means that the required level of the tag “Excel” is the advanced level and the required level of the tag “accounting knowledge” is the beginner level. Skill feature vectors are also represented in the same manner, and are defined by the skill level that is currently held for all defined task tags.
Tagging may be other things, such as English conversation ability, programming ability, etc. In the above description, the level is expressed as an advanced level, a beginner level, or the like. However, the level may be simply expressed in 6 levels (0 to 5) or 10 levels (0 to 9). Here, the ticket means a task. When each feature vector is expressed in this way, the similarity between the skill and the ticket (task)

Is expressed as follows.

As in Equation 4, so-called cosine similarity is obtained. A conceptual diagram of this calculation is shown in FIG. Thus, what is calculated by Expression 4 is the cosine value of the angle θ formed by the skill feature vector 6 and the task feature vector 8. As shown in the above equation 4, the cosine value of the angle formed by the two vectors is the number obtained by dividing the inner product of the two vectors by the product of the sizes of the vectors.

(2) Configuration and Operation FIG. 2 is a functional block diagram showing the configuration of the task / employee matching apparatus 10. FIG. 2 shows the configuration of the task / employee matching apparatus 10 and the flow of operation. Hereinafter, the configuration and operation of the task / employee matching apparatus 10 will be described with reference to FIG.

First, the customer 12 inputs a task he / she wants to request via the task request unit 16. The task request unit 16 is an input means for inputting a task from the customer 12, and is realized by a computer keyboard, a touch panel, or the like. When a task is requested via the network, the network interface means constitutes the task request unit 16. In addition, the task request unit 16 may be realized with any configuration as long as a task can be input.
The task request unit 16 registers the input task in the task information database 20. Here, the registered task includes not only data describing the customer 12 who requests the task, the task number, the task content, and the like, but also the above-described feature vector of the task. The task information database 20 may use a predetermined storage unit such as a hard disk or a semiconductor storage unit.

Next, the employee 14 inputs his skill via the skill registration unit 18. The skill registration unit 18 is an input means for the employee 14 to input his / her skill, and is realized by a computer keyboard, a touch panel, or the like. In addition, a skill may be input via a network. In this case, the network interface means constitutes the skill registration unit 18. In addition, as long as the employee 14 can input skills, the skill registration unit 18 may be realized in any configuration.
The skill registration unit 18 registers the input skill of the employee 14 in the skill information database 22. Here, the registered skill is data describing the content of the skill, but particularly includes the above-described skill feature vector. The skill information database 22 may also use predetermined storage means such as a hard disk or semiconductor storage means.

Next, the skill / task similarity calculation unit 24 extracts a feature vector of the task from the task information database 20 for a predetermined task. Further, the skill / task similarity calculation unit 24 extracts a feature vector of the employee 14 from the skill information database 22 for the predetermined employee 14. Then, the skill / task similarity calculation unit 24 performs the calculation of the above-described formula 4 to obtain the similarity between the skill of the employee 14 and the task.
In this way, the skill / task similarity calculation unit 24 can select the employee 14 suitable for the task by obtaining the similarity between the predetermined task and the skill of each employee 14. For example, employees 14 having skills (feature vectors) whose similarity to the task is equal to or higher than a certain reference value are extracted, and these employees 14 are determined to be employees 14 suitable for the task. be able to. In this way, it is possible to create a list of employees 14 having skills whose similarity is equal to or higher than a certain reference value for each task.
In this way, the skill / task similarity calculation unit 24 can create a “recommended employee list” 26 that is a list of employees 14 having skills suitable for a certain task. The recommended target employee list 26 is a table including the task, the employee 14 suitable for the task, and the skill of the employee 14 and the similarity between the task and the task. The recommended employee list 26 may be, for example, a table including other items (year of experience, work style, etc.). In this way, the skill / task similarity calculation unit 24 can create the recommendation target employee list 26 for each task.

The recommended employee list 26 is stored in a predetermined storage means. The storage means may be provided in the skill / task similarity calculation unit 24, or a predetermined external storage means may be used. Further, the storage means may be shared with the above-described task information database 20 or the like.
The skill / task similarity calculation unit 24 can be composed of a program describing these operations and a CPU that executes the program. Further, the skill / task similarity calculation unit 24 may include a storage unit that stores the recommendation target employee list 26.

Next, the recommended task presentation unit 28 searches the recommended employee list 26 including the employee 14 for the predetermined employee 14. Then, the task of the recommendation target employee list including the employee 14 can be presented to the employee 14 as a recommended task.
The recommended task presentation unit 28 may execute the above operation according to a request from the employee 14. With such a configuration, the employee 14 can easily know a task (similar) suitable for his / her skill. For example, when the employee 14 inputs his / her employee number from a keyboard or the like, the recommended task presentation unit 28 searches the recommended employee list 26 including the number and displays the task on, for example, a computer display. , Can inform the employee 14.
As a presentation method, for example, a recommended task may be displayed on a display, or a list of recommended tasks may be printed. An email including a recommended task may be sent to the employee. The “presentation” in the embodiment may be any method as long as the recommended task can be shown to the employee 14.
Further, the “presentation to a person” does not need to be the employee 14 himself / herself. For example, when a manager of a company calculates the above-mentioned similarity using the employee number of the employee 14 and receives “presentation” of a task that matches the employee, it is also referred to as “presentation to a person”. . Further, “to be presented to a person” such as printing with a printer or sending by e-mail corresponds to “presentation to a person” here. Similarly, posting on a predetermined Web page and making it in a state where it can be viewed by a person corresponds to “presentation to a person”. The recommended task presentation unit 28 can be composed of a program describing these operations and a CPU that executes the program. In addition, a keyboard or the like may be provided for inputting an employee number or name of the employee 14, a request for the recommended employee list 26, and the like. Also. When inputting an employee number, a name, and a request via a network, a predetermined network interface may be provided.
The task / employee matching apparatus 10 may be used not only by the customer 12 and the employee 14 but also by a manager of the company. It is also a preferable usage method that a manager of a company can input an employee number to know a task suitable for the employee 14 and recommend the task to the employee 14.

  Further, the task / employee matching apparatus 10 may be configured by a single computer or may be realized by a plurality of computers. Moreover, you may comprise with the some computer connected with the network. That is, you may comprise as an apparatus provided with the structure by which each structure was distributed on the network. Moreover, the matching apparatus 10 of a task and an employee may be comprised on a cloud, and the form which each person uses via a network may be employ | adopted, and you may comprise on-premises.

As described above, according to the first embodiment, the similarity between the skill of the employee 14 and the task can be calculated, and a task having a similarity higher than a predetermined reference value can be presented to the employee 14. Therefore, the employee 14 can know a task suitable for him and can execute the task.
The employee 14 corresponds to a preferable example of “person” in the claims, but “person” is not particularly limited to the employee 14 and may include students, students, athletes, entertainers, and the like. The task / employee matching device 10 corresponds to a preferred example of the “device for presenting a task that matches a person” in the claims. Further, the matching operation between the task and the employee according to the first embodiment corresponds to a preferable example of the “method for presenting a task that matches a person” in the claims.

Embodiment 2. FIG. Task recommendation to an employee who has performed a task with high past similarity based on an analysis of the employee's behavior history In the second embodiment, task recommendation is performed based on the similarity between the behavior history of the employee 14 and the task. An example of the method will be described.

(1) Features First, the features of this method are as follows.
(A) When a certain task (new task) is newly registered, a similar task is extracted from past task digest record information from the feature vector of the task. Here, the calculation of similarity applies the principle of content-based filtering using the feature vector of the task. That is, as in the first embodiment, the “content-based filtering” technique used in the recommendation engine is applied.
Note that a new task is a task to be presented by a matching person, and is sometimes referred to as a “target task” (see claims).
(B) Next, “employee prioritization” (described later) is performed on the employee 14 who is executing a task having a high similarity to the new task, and the “similarity between tasks” is a constant threshold value. The newly registered task is recommended for the employee 14 who exceeds. The similarity between tasks is a similarity between a feature vector of a certain task and a feature vector of another task. In the first embodiment. The similarity between the human feature vector and the task feature vector has been explained, but the similarity between the human feature and the task similarity can be distinguished from those obtained by performing the same calculation between the task feature vectors. Among these degrees, it is called “similarity between tasks”.
(C) “Recommendation prioritization” is performed as follows based on a total value obtained by adding all the similarities between tasks of past performance tasks (this value is referred to as “total similarity”). .
First, a collection of the results of tasks performed by an employee 14 in the past

It expresses. Here, n is the number of tasks executed by a certain employee 14 in the past. When a newly registered task is represented as newTask, the “total similarity” of the employee 14 is obtained as follows.

The total similarity obtained by Equation 6 is referred to as “priority”. The priority is also referred to as “recommendation degree”.

Modified example of calculation of priority (total similarity) (a) In the second embodiment. The priority is calculated as in Equation 6 above. Here, an example is shown in which similarities are calculated and added to all past performance tasks, but similarities with low similarities may be excluded from addition targets. This is because even if the similarity between the target task and an unrelated task is added, there is not much meaning. For example, in Equation 6, it is also preferable to add only similarities whose similarity is equal to or greater than a predetermined reference value.
(B) In Equation 6 above. Although all the similarities are added, an average value per task may be obtained. This is because there are a number of low similarities, and it may be preferable to give priority to one having a high degree of similarity rather than a case where the total value becomes a large value.

Table Next, FIG. 3 shows a table for which priority (total similarity) is obtained. In the table shown in the figure, a row is a task similar to the new task. In the figure, it is shown that the similarity between the task X and the new task is 0.9. Similarly, it is indicated that the similarity between the task Y and the new task is 0.8. Further, it is shown that the similarity between task Z and the new task is 0.7. In the second embodiment. These tasks X, Y, and Z are determined to be similar to the new task. In this determination, for example, a predetermined reference value may be used, and if the degree of similarity exceeds the reference value, the task may be determined to be similar to the new task.
In addition, employees are shown in the columns of FIG. First, employee A has executed task X and task Y in the past, and the priority (total similarity) is 1.7, which is the total value of the similarity 0.9 and 0.8 of each task. ) And the recommendation level. Next, employee B has executed task Y and task Z in the past, and 1.5, which is the total value of the similarity 0.8 and 0.7 of each task, is the priority (all similarities). Degree), which is the recommendation degree. Employee C has executed task Z and task X in the past, and 1.6, which is the total value of similarity 0.7 and 0.9 for each task, is the priority (total similarity). Yes, the recommendation level.
As a result, the “rank” is ranked first for the employee A who has the highest recommendation level (1.7). Subsequently, Mr. C, who has the second highest recommendation level (1.6), takes the second place. Finally, employee C, who has the smallest recommendation level (1.5), is ranked third. In FIG. 3, such a ranking is indicated by a number surrounded by a circle.
Such a process of assigning “order” is referred to as “recommendation prioritization”.

(2) Configuration and Operation FIG. 4 is a functional block diagram showing an example of the configuration of the task / employee matching apparatus 110. FIG. 4 shows the configuration of the task / employee matching apparatus 110 and the flow of operations. Hereinafter, the configuration and operation of the task / employee matching apparatus 110 will be described with reference to FIG.

  First, the customer 12 inputs a task he / she wants to request via the task request unit 16. The configurations and operations of the customer 12, the task request unit 16, and the task information database 20 in FIG. 4 are the same as those described with reference to FIG.

Next, the employee 14 inputs the result of his / her task through the task result registration unit 118. The task result registration unit 118 is an input means for the employee 14 to input the past task results. However, the task registration unit 118 differs from the skill registration unit 18 only in the registration target, and its operation is skill registration. This is the same as the part 18.
The task record registration unit 118 registers the input task record of the employee 14 in the task record information database 122. Here, the registered task performance information is data describing the contents of a task executed in the past, and particularly includes a feature vector of the task. The task performance information database 122 may also use predetermined storage means such as a hard disk or semiconductor storage means.

Next, the inter-task similarity calculation unit 124 extracts, from the task information database 20, a feature vector of the task with respect to a task that is currently determined (hereinafter referred to as a target task). Further, the inter-task similarity calculation unit 124 extracts, from the task performance information database 122, one or a plurality of predetermined employees 14 and feature vectors of tasks executed by the employees 14 in the past. Then, the inter-task similarity calculation unit 124 executes the calculation of the above-described formula 6 to obtain the similarity (inter-task similarity) between the task executed by the employee 14 in the past and the target task.
In this way, the inter-task similarity calculation unit 24 obtains the inter-task similarity between the target task and the task that each employee 14 has executed in the past. And the total value of the similarity between tasks is calculated | required for every employee 14. FIG. In this way, the similarity obtained by summing the similarities between tasks is particularly called the total similarity. Such processing is as described in FIG.・
The inter-task similarity calculation unit 124 can create such a table as illustrated in FIG. 3 by executing such processing for one or a plurality of employees 14. This table is called a similarity degree calculated task list 126 (see FIG. 4).
Further, the inter-task similarity calculation unit 124 puts a ranking according to the value of the recommendation degree (total similarity) in the recommendation degree column of the similarity degree calculated task list 126, and prioritization is performed. Thereby, the table shown in FIG. 3 is completed.
In this way, the inter-task similarity calculation unit 124 is a list including a recommendation degree (total similarity, priority) of the employee 14 who has executed a similar task with respect to the target task. A “list” 126 can be created. The similarity degree calculated task list 126 may be, for example, a table including other items (experience years, work styles, etc.).

The similarity degree calculated task list 126 is stored in a predetermined storage capacity. The storage means may be provided in the inter-task similarity calculation unit 124, or a predetermined external storage means may be used. Further, a common storage means may be shared with the above-described task information database 20 or the like.
The inter-task similarity calculation unit 124 can be composed of a program describing these operations and a CPU that executes the program. The inter-task similarity calculation unit 124 may include a storage unit that stores the similarity calculated task list 126.

  Next, the recommended target employee extraction unit 128 extracts the employees 14 of the higher rank from the similarity degree calculated task list 126. These extracted employees 14 are employees 14 recommended for the target task, and these employees 14 can be presented to the task as they are. The “presentation for the target task” mentioned here includes “presentation” to the customer who requested the target task. In addition, when the manager who has received the request for the target task wants to know the employee 14 that matches the target task, the recommended employee 14 is presented to the manager. That is, if it is output as the employee 14 recommended for the target task, it corresponds to “presentation” in the present embodiment regardless of the output destination.

The recommended target employee extraction unit 128 outputs the extracted list of employees 14 as the recommended target employee list 26. This recommended target instructor list 26 is the same as the recommended target instructor list 26 described with reference to FIG.
If the recommendation target employee list 26 already exists, the recommendation target instructor list 26 may be constructed by adding data thereto.
As shown in FIG. 4, the recommended task presentation unit 28 can present a matching task to the employee 14 based on the recommended target instructor list 26. Since the operation is the same as that of FIG. 2 of the first embodiment, the description thereof is omitted.

Thus, the task and employee matching apparatus 110 according to the second embodiment can present a person who matches the target task. For this purpose, the recommended target instructor list 26 and the recommended task presentation unit 28 are not essential components, and the recommended target employee extraction unit 128 may present a person who matches the target task. The task / employee matching apparatus 110 according to the second embodiment corresponds to a preferred example of an apparatus that presents a person who matches a task in the claims. The operation of matching tasks and employees according to the second embodiment corresponds to a preferred example of a method for presenting a person who matches a task in the scope of claims.
However, in the task-employee matching device 110 according to the second embodiment, the employee list 14 (persons) is retained in the same manner as in the first embodiment by maintaining the recommended class member list 26 and the recommended task presentation unit 28. ) Can be presented as a matching task.

  The task / employee matching apparatus 110 may be used not only by the customer 12 and the employee 14 but also by a manager of the company. The customer 12 can also know the employee 14 that matches the task he has requested. In addition, the manager of the company can input the target task instead of the customer 12 and know the matching employee 14.

Embodiment 3. FIG. Task recommendation to employees who performed tasks with high co-occurrence by analysis of past task digestion results In Embodiment 3, the digestion results of past tasks of employees are analyzed, and the target task and co-occurrence Explain the method to recommend the target task to employees who have performed high tasks.

(1) Features First, the features of this method are as follows.
(A) Analyzing an action history (task result information) of a certain employee 14, focusing on co-occurrence between tasks, a task that is easy to match is proposed from past task results. In the third embodiment, a collaborative filtering technique used in a so-called recommendation engine is applied.
(B) If the set of employees 14 who have performed task X is SetX and the set of employees 14 who have performed task Y is SetY, the co-occurrence of task X and task Y (task X and task Y are the same employee Probability implemented by member 14) cooccurrence (index) is expressed as follows using the Jaccard coefficient.

Focusing on such co-occurrence, an employee 14 who is performing a task with high co-occurrence with a certain target task can be determined as an employee 14 matching the target task.
(C1) Next, an example of the flow of operation will be described. First, when there is a newly registered task (hereinafter also referred to as a target task), the target task and a past task group are subjected to cluster analysis and classified into a predetermined cluster. An explanatory diagram of such classification is shown in FIG. FIG. 5 is a diagram schematically showing the division into cluster groups, in which each task (feature vector) is mapped in a predetermined space (vector space). . The new task is classified as Cluster A in FIG. In FIG. 5, various tasks are classified into clusters, and ClusterB, ClusterC, and ClusterD are depicted.
(C2) Next, a cluster group having high co-occurrence with a cluster group to which a newly registered new task belongs is calculated and extracted using the above-described cooccurrence index.
For example, in the example of FIG. 5, the cluster to which the new task belongs is ClusterA. Clusters with high co-occurrence with ClusterA are calculated and extracted using the cooccurrence index. In the example of FIG. 5, as a result of the calculation, the cooccurrence index (co-occurrence index) between Cluster A and Cluster B is 0.5. The cooccurrence index (co-occurrence index) between ClusterA and ClusterC is 0.9. The cooccurrence index (co-occurrence index) between Cluster A and Cluster D was 0.7. As a result, in FIG. 5, ranking is performed in descending order of the co-occurrence index. In the ranking, ClusterC is first, ClusterD is second, and ClusterB is third (see FIG. 5). In the example of FIG. 5, the ranking is represented by a number surrounded by a rectangle. Such a process of assigning “order” is referred to as “recommendation prioritization” as in the second embodiment.
In this way, ClusterC is extracted as a cluster group with high co-occurrence (FIG. 5). Then, the target task is recommended to the employee 14 who has executed the task belonging to the extracted cluster group ClusterC.

(2) Configuration and Operation FIG. 6 is a functional block diagram showing an example of the configuration of the task / employee matching device 210. FIG. 6 also shows the configuration of the matching device 210 for tasks and employees 14 and the flow of operations. Hereinafter, based on FIG. 6, the structure and operation | movement of the matching apparatus 210 of a task and the employee 14 are demonstrated.

  First, the customer 12 inputs a task he / she wants to request via the task request unit 16. The configurations and operations of the customer 12, the task request unit 16, and the task information database 20 in FIG. 4 are the same as those described in FIG. 2 in the first embodiment and FIG. 4 in the second embodiment, and thus description thereof is omitted here.

Next, the employee 14 inputs the result of his / her task through the task result registration unit 118.
Since the contents / operations of the task record registration unit 118 and the task record information database 122 are the same as those in FIG. 4 of the second embodiment, description thereof is omitted.

Next, the inter-task similarity calculation unit 124 extracts, from the task information database 20, a feature vector of the task with respect to a task that is currently determined (hereinafter referred to as a target task). Further, the inter-task similarity calculation unit 124 extracts, from the task performance information database 122, one or a plurality of predetermined employees 14 and feature vectors of tasks executed by the employees 14 in the past. Then, the inter-task similarity calculation unit 124 performs the calculation of Equation 6 described above to obtain the similarity between the task executed in the past by the employee and the target task (inter-task similarity).
And the total value of the similarity between tasks is calculated | required for every employee 14. FIG. In this way, the similarity obtained by summing the similarities between tasks is particularly called the total similarity.
Since the configuration / operation of the inter-task similarity calculation unit 124 is as illustrated in FIG. 4 of the second embodiment, detailed description thereof is omitted. The similarity calculated task list 126 is also the same as the configuration / operation described in FIG.

The task co-occurrence calculation unit 202 executes the processes (c1) and (c2) described above.
First, based on the contents of the similarity degree calculated task list 126, task cluster analysis is performed, and the tasks are divided into cluster groups (for example, FIG. 5).
Next, a cluster group to which a new task is added and a cluster group having a high co-occurrence are searched for and listed. That is, this is a list of tasks included in the cluster group with high co-occurrence (ClusterC in the example of FIG. 5), and this list is called a task list 204 with high co-occurrence. 1
Next, the recommended target employee extraction unit 128 extracts the employees 14 who have performed the tasks described in the task list 204 with high co-occurrence. Then, a recommended employee list 26 that is a list of the extracted employees 14 is generated.
Since the configuration after the recommended employee extraction unit 128 is the same as the configuration and operation described in FIG. 4 of the second embodiment, the description thereof is omitted.

In this way, the task and employee matching device 210 according to the third embodiment can present a person who matches the target task. This operation is performed by the recommended target employee extraction unit 128 as in the second embodiment.
Also, in the task-employee matching apparatus 110 according to the present embodiment, by holding the recommended class member list 26 and the recommended task presentation unit 28, the employee 14 (person) is the same as in the first embodiment. Can also be presented with matching tasks.
The task / employee matching apparatus 210 according to the third embodiment corresponds to a suitable example of an apparatus that presents a task that matches a person in the scope of claims. Further, the task and employee matching operation according to the third embodiment corresponds to a preferred example of a method for presenting a task that matches a person in the scope of claims.

  The task / employee matching device 210 may be used not only by the customer 12 and the employee 14 but also by a manager of the company. The customer 12 can also know the employee 14 that matches the task he has requested. In addition, the manager of the company can input the target task instead of the customer 12 and know the matching employee 14.

Embodiment 4 FIG. Mechanism of automatic calculation of task difficulty for person in charge In Embodiments 1 to 3, a technique for matching a person (employee) with a task has been described. In the fourth embodiment, it is convenient if the degree of difficulty of a task can be known with respect to the skill of an employee (which is called a person in charge) 14 who actually executed the task. In particular, when the degree of similarity exceeds a certain threshold (for example, when cos θ> = 0.9), the merit of quantifying such difficulty is great.
For example, it is convenient if the employee 14 can present the degree of difficulty when the employee 14 becomes a person in charge of the task while making a recommendation using the mechanism of the first to third embodiments.
(1) When the task vector is projected on the skill vector axis, the ratio is calculated by the ratio of the task length to the skill vector length. If the degree of difficulty calculated in this way is presented together with the matching task presentation or the person presentation, it can be used as a material for determining how appropriate the proposal is.
(2) In other words, the task difficulty level

This difficulty level is expressed as

It can be expressed by the following formula 9.
(3) However, if the skill and task similarity is high,

Therefore, the following approximate expression is established.

It is preferable to present such a degree of difficulty together with the presentation (recommendation) of people and tasks.
In the first embodiment, this calculation and presentation may be calculated by the skill / task similarity calculation unit 24, or may be calculated and presented by the recommended task presentation unit 28. These correspond to a preferred example of the calculation unit in the claims.
In the second embodiment, the task similarity calculation unit 124 may calculate, or the recommended target employee extraction unit 128 may calculate and present it. Further, the recommended task presentation unit 28 may calculate and present it. These correspond to a preferred example of the calculation unit in the claims.
Also in the third embodiment, the task similarity calculation unit 124 may calculate, or the recommended target employee extraction unit 128 may calculate and present it. Further, the recommended task presentation unit 28 may calculate and present it. These correspond to a preferred example of the calculation unit in the claims.

  The present embodiment described above shows an example of the configuration of the invention, and the configuration can be changed as appropriate within the scope of the invention.

6 ... Skill feature vector 8 ... Task feature vector 10, 110, 210 ... Task and employee matching device 12 ... Customer 14 ... Employee 16 ... Task request part 18 ... Skill registration part 20 ... Task information database 22 ... Skill Information database 24 ... Skill / task similarity calculation unit 26 ... Recommended employee list 18 ... Skill registration unit 20 ... Task information database 22 ... Skill information database 24 ... Skill / task similarity calculation unit 118 ... Task performance registration unit 122 ... Task result information database 124 ... inter-task similarity calculation unit 126 ... similarity calculated task list 128 ... recommended employee extraction unit 202 ... task co-occurrence calculation unit 128 ... task list with high co-occurrence

Claims (14)

A device for presenting a task that matches a person to the person,
The skill information database for registering a feature vector of the person's skill;
Skill registration to register the feature vector of the person's skill in the skill information database
And
A task information database for registering one or more feature vectors of the task;
A task request unit for registering the feature vector of the task in the task information database;
A feature vector of the person's skill is extracted from the skill information database, a plurality of feature vectors of the task are extracted from the task information database, and the extracted feature vector of the person's skill, and the extracted feature vector of the task, , A skill-task similarity calculation unit for calculating the similarity of
A recommended task presentation unit that presents the task of the feature vector of the task having a high similarity to the feature vector of the person's skill among the feature vectors of the task for which the similarity is obtained;
Only including, the degree of similarity, and the feature vector of the person of skill, a device for presenting a task that matches the person, which is a cosine similarity between the feature vector of the task (cosine similarity). A method for presenting a task that matches a person to the person using the apparatus according to claim 1 ,
The skill registration unit registering the feature vector of the person's skill in the skill information database;
The task requesting unit, registering the feature vector of the task in the task information database;
The step of calculating the similarity between the feature vector of the person's skill and the feature vector of the task;
The similarity calculating unit executing the step of obtaining the similarity with respect to feature vectors of a plurality of tasks;
The recommended task presentation unit presents, to the person, the task having the feature vector of the task having a high similarity to the feature vector of the person's skill among the feature vectors of the task whose similarity is obtained. Steps,
And the similarity is a cosine similarity between the feature vector of the person's skill and the feature vector of the task . Feature vector of the task, and one or more tagged for the task, tagged according to claim 2, characterized in that the level and consists of a vector is determined for each of the tags A way to present tasks that match people. The feature vector of the person is a vector comprising one or more tagging for the person and a level that the person maintains for each of the tagged tags. A method for presenting a task that matches a person according to claim 2 or 3 . An apparatus for presenting a person who matches the target task to the target task,
A task performance information database for registering a feature vector of task digest performance of one or more persons,
A task result registration unit for registering the feature vector of the task digest result in the task result information database;
A task information database for registering the feature vector of the target task;
A task request unit that registers the feature vector of the target task in the task information database;
An inter-task similarity calculation unit for obtaining an inter-task similarity between the feature vector of the person's task digestion result and the feature vector of the target task;
A recommended target employee extraction unit for presenting the person having the feature vector of the task digestion result having a high degree of similarity between the tasks to the target task;
A device that presents people who match a task with A method for presenting a person who matches the target task to the target task using the apparatus according to claim 5 ,
The task record registration unit obtains a feature vector of task digest record of one or more persons.
Registering in the task performance information database;

The task requesting unit, registering the feature vector of the target task in the task information database;
The inter-task similarity calculation unit obtains the inter-task similarity between the feature vector of the person's task digest record and the feature vector of the target task;
The inter-task similarity calculation unit performs the step of obtaining the inter-task similarity on one or a plurality of the human task digestion feature vector,
The recommended target employee extraction unit presenting the person having a feature vector of task digestion results with a high degree of similarity between tasks to the target task;
How to present people matching the task, including The person's task digestion feature vector is a feature vector of the person's past achievement tasks,
The similarity between tasks is obtained by calculating the similarity between the feature vector of the target task and the feature vector of the plurality of past achievement tasks of the person for each feature vector of the plurality of past achievement tasks. The method of presenting a person who matches a task according to claim 6, wherein the similarity is a total value of degrees. The person's task digestion feature vector is a feature vector of the person's past achievement tasks,
The similarity between tasks is obtained by calculating the similarity between the feature vector of the target task and the feature vector of the plurality of past achievement tasks of the person for each feature vector of the plurality of past achievement tasks. 7. The method for presenting a person who matches a task according to claim 6, wherein the degree of similarity is a total value that is a total value of only the similarities equal to or higher than a predetermined reference value. A device for presenting a task that matches a person to the person,
A task result information database for registering a feature vector of the person's task digestion results;
A task result registration unit for registering the feature vector of the task digest result in the task result information database;
A task information database for registering one or more feature vectors of the task;
A task request unit that registers the feature vector of the target task in the task information database;
A task-task similarity calculation unit for obtaining a similarity between tasks of the feature vector of the task digestion result of the person and a plurality of feature vectors of the tasks;
The task for which the similarity between tasks is obtained is decomposed into cluster groups, and in the task for which the similarity between tasks is obtained, another cluster having high co-occurrence with a cluster group to which a predetermined one task belongs A co-occurrence calculation unit for extracting cluster groups;
Among the feature vectors of the tasks belonging to the extracted cluster group, when the similarity between the tasks with the feature vector of the task digestion result of the person is large, the predetermined one task is assigned to the person A recommended task presentation section to be presented;
A device that presents tasks that match people including A method for presenting to a person a task that matches the person using the apparatus according to claim 9 ,
Registering the feature vector of the person's task digest record in the task record information database;
The task requesting unit, registering the feature vector of the task in the task information database;
The similarity calculation unit, a step of determining a feature vector of said human task digestion results, the feature vector of the task, the task among such acetonide,
The step of calculating the similarity between tasks by the similarity calculation unit with respect to feature vectors of a plurality of tasks;
The co-occurrence calculating unit decomposing the task for which the similarity between tasks is obtained into a cluster group;
The co-occurrence calculation unit, in the task for which the similarity between tasks is obtained, extracting another cluster group having high co-occurrence with a cluster group to which a predetermined one task belongs;
When the recommended task presenting unit has a large similarity between the tasks with the feature vector of the task digestion result of the person among the feature vectors of the task belonging to the extracted cluster group, the predetermined one task Presenting to the person,
How to present matching tasks to people including.   11. The method for presenting a task that matches a person according to claim 10, wherein the step of decomposing into cluster groups decomposes the task into a plurality of cluster groups by performing cluster analysis.   The co-occurrence is determined by comparing the first set with respect to a first set of people who have executed tasks belonging to a predetermined cluster group and a second set of people who have executed tasks belonging to another cluster group. The number of persons belonging to the first set and the second set in relation to the number of persons belonging to the entire set of the second set and the second set. A way to present tasks that match people. A device for determining the level of difficulty of a person for a task,
A calculation unit that calculates, as the difficulty level, a value obtained by dividing the length of projection of the feature vector of the task with respect to the feature vector of the skill of the person by the length of the feature vector of the skill;
A device that determines the level of difficulty of a person for a task that includes A method for determining the degree of difficulty of a person for a task using the apparatus according to claim 13 ,
When the calculation unit projects the feature vector of the task on the feature vector of the person's skill, a value obtained by dividing the projection length by the length of the skill feature vector is calculated as the difficulty level. Step to do,
A method for determining the level of difficulty of a person for a task that contains a task.