JPH09305406A - System and method for processing information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve a job efficiency of a user by determining a task to be next executed based on 1st information, finding the candidate of solution of the task by performing prescribed search processing based on 2nd information previously determined by that determined task, and performing prescribed post- processing based on that candidate of the solution. SOLUTION: The task to be first executed is determined and when the conditions of execution are turned to any prescribed state, a task system determining part 12 determines the task to be next executed based on the 1st information determined in advance. Concerning that determined task, prescribed search processing is performed based on the predetermined 2nd information, and the candidate of solution of the task is found by an individual task processing determination part 14 corresponding to a search processing means. Based on that found candidate of the solution, the prescribed post-processing is performed by an individual task processing execution part 15 corresponding to a post- processing means. Thus, the job efficiency or result of the user can be improved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing system and an information processing method for executing a predetermined task or a task designated by a user and another task series related to the task.

[0002]

2. Description of the Related Art In recent years, with the development of computer technology and software science, a computer system capable of solving advanced and complicated problems that could not be conventionally solved by a computer while interacting with a user. Has been developed and is being put to practical use. A computer system belonging to a category called an expert system is a typical example. In these systems, in general, a problem given to a preset predetermined task is given by a user, and in order to solve the problem of the task, knowledge recorded in a knowledge base such as inference rules and inference control information is used. Problem solving is done. If the information necessary for solving the problem is missing, the information is acquired through interaction with the user as necessary. By these methods, these systems represented by the expert system can solve the complicated problem given by the user.

Conventionally, in these systems, a system developer organizes information necessary for problem solving for a predetermined task, enumerates possible control structures, describes them as inference rules and control codes, and stores them in a knowledge base. By recording, each system has been developed. By adopting such a method, a system capable of solving a problem belonging to a sufficiently complicated task for a computer is developed at a relatively low cost, and a relatively efficient method is available when the system is operating. I was able to solve the problem.

However, in many expert systems, since the system is constructed for each task or for each problem even in the same task, it is general that they are independently operated even if they are related to the problems handled by these plural systems. Met. Therefore, when a user wants to obtain a solution of a task or a problem that is related to each other, the user himself / herself needs to check the consistency between the inferences and the derivations of each system without the system's support. Had to manage.

Since the system is independently operated for each task or problem in this way, unless the user is aware of the related task or problem, the information on the problem by the user as described above is provided. There was a problem that even the management of the coherence was not done, and there could be inconsistency between the solutions derived in the problems that should be related to each other. In other words, if the user is unaware of the relationship between the problems, each system will solve the problem in the range that the user assumed for the task or problem in charge, and useful information for the user. It is not possible to point out the relationship between tasks and problems that the user is not supposed to do and it is not possible to create an opportunity to improve the work efficiency and results of the user, There was also the risk that the chosen solution would be inappropriate.

Furthermore, even when solving the same task or problem, the desired solution or solution may differ depending on the user. That is, even for the same task / problem, its solution method or solution may have individuality. In the conventional expert system, the knowledge base used for problem solving is generally common among users, and it has been difficult to provide such personality. Certainly, even in the conventional system, it is not impossible to give such individuality by replacing the knowledge base used for problem solving for each user. However, if an attempt is made to give individuality to a solution or solution by such a method, conversely, if there is a need to change the knowledge base in a part that does not depend on the user, maintenance of the knowledge base of all users will occur. Another problem arises, such as the need to perform.

[0007]

As described above, conventionally, there has been a problem that it is impossible to derive a solution in a consistent manner by associating problem solving processes of a plurality of related tasks. There is also a problem that a large burden is imposed on the user to associate them. Furthermore, even in a system that can execute multiple related tasks, the execution of related tasks remains within the scope of the user's expectations, so business efficiency and results that utilize the relevance of tasks that the user did not expect. There was also a problem that we could not improve.

In addition, there is a problem that a great burden is required on the user or the knowledge base manager in order to give individuality to the solution or solution when solving the problem of each task. .

The present invention has been made in consideration of the above circumstances, and leads a solution in a consistent form by associating problem solving processes of a plurality of related tasks and without burdening the user. It is an object of the present invention to provide an information processing system and an information processing method capable of improving the work efficiency and results of a user by pointing out to the user the association between a plurality of tasks that the user does not expect. And

Further, the present invention is an information processing system capable of adding a user's individuality to a solution or a solution when solving a problem of each task without burdening the user or a knowledge base manager. And an information processing method.

[0011]

The present invention has a function of managing one or more tasks and managing the association between related tasks, and from the resolution of a predetermined or user-specified initial task. Once started, when the problem solving status for a task reaches a predetermined state, it has the function of searching for the next related task and instructing execution of that individual task, and also about how to solve the problem of the instructed individual task. It is characterized by having a function of performing a search and selecting and executing an appropriate solution to a given problem.

According to the present invention, in addition to the above configuration, based on the evaluations and requests obtained from the user regarding the solution method of each individual task and specific examples of the solution, the individual task can be satisfied. The model for resolving problem solving method is updated to improve the performance and efficiency of solution search for each individual task, and also has a function to reflect the individuality of the user in the solution of individual task. It is characterized by.

The present invention (Claim 1) is an information processing system for executing a predetermined task and at least one task related to the predetermined task, wherein the predetermined task to be executed first (for example, a preset task). When a task or task instructed by the user) is determined and the execution status becomes a predetermined state (for example, when a solution is confirmed)
Determining means for determining the task to be executed next based on the first information determined in advance, and performing a predetermined search process based on the second information determined in advance for the determined task. And a post-processing unit for performing predetermined post-processing (for example, presenting a solution candidate and requesting confirmation by the user) based on the obtained solution candidate. It is characterized by having.

According to the present invention (claim 2), in claim 1, the first information has a plurality of related tasks as nodes and transits to a directional branch for transiting from one node to another node. A directed graph with a condition added, which is capable of making a transition from the current node to another node by verifying the transition condition and selecting a directional branch satisfying the condition, and used for verifying the transition condition The determination means transitions the nodes of the directed graph starting from a node associated with the task whose execution status has become a predetermined state among the nodes included in the directed graph. , A task associated with a transition destination node is selected as a task to be executed next.

The transition condition of the first information includes a transition condition of unconditional. Therefore, transition conditions that are not unconditional are added to all directional branches, transition conditions that are not unconditional are added to some directional branches, and transition conditions of all directional branches are It may be unconditional. Regarding the first information in which the transition conditions of all the directional branches are unconditional, the information used for verification of the transition conditions is unnecessary, and the verification of the transition conditions is unnecessary.

The present invention (Claim 3) includes Claim 1 or Claim 2.
In the second information, each of the selectable solution candidates for the task is set as a terminal node, an intermediate state in the process of searching for a solution candidate is set as another node, and a transition from one node to another node is made. A directed graph in which a transition condition is added to a directional branch for the purpose of selecting a directional branch that satisfies the condition by verifying the transition condition, and transitions from the current node to another node, depending on the current situation. Includes task model information composed of a directed graph capable of repeating node transitions until reaching a terminal node, and information used for verifying the transition condition, and the search processing means is set in advance. It is characterized in that the transition on the directed graph is repeated starting from a node and a solution candidate associated with the reached terminal node is selected.

The transition condition of the second information includes a transition condition of unconditional. Therefore, transition conditions that are not unconditional are added to all directional branches, transition conditions that are not unconditional are added to some directional branches, and transition conditions of all directional branches are It may be unconditional. As for the task model information in which the transition conditions of all the directed edges are unconditional, the information used for verifying the transition conditions is unnecessary, and the verification of the transition conditions is unnecessary.

The present invention (claim 4) is based on claim 1 or 2
In the second information, each of the selectable solution candidates for the task is a leaf node, an intermediate state in the process of searching for a solution candidate is another node, and a transition from one node to another node is made. It is a directional tree in which a transition condition is added to a directional branch for the purpose of verifying the transition condition and selecting the directional branch that satisfies the condition to transition from the current node to another node. In accordance with the task model information consisting of a directed tree that can be repeated node transition until reaching a leaf node, and the information used for verification of the transition condition, the search processing means, It is characterized in that the transition on the directed tree is repeated from a preset node as a starting point and a solution candidate associated with the reached leaf node is selected.

The present invention (Claim 5) includes Claims 1 to 4.
In any one of the above items, further comprising a dialogue means for interacting with the user, wherein when the decision means cannot decide the task due to lack of information, the decision means makes the decision through the dialogue means. It is characterized in that necessary information (for example, an identifier of an initial task) is inquired to a user and acquired, and the task is determined based on the acquired information.

The present invention (Claim 6) includes Claims 1 to 4.
In any one of the above items, a dialog unit for interacting with the user is further provided, and the post-processing unit presents a solution candidate of the task to the user through the dialog unit as the post-processing, and It is characterized in that an instruction of adoption or non-adoption of a task solution candidate is accepted, and when an instruction of adoption is given, a candidate of the task solution is confirmed as a solution of the task.

The present invention (Claim 7) includes Claim 3 or 4.
At the time of the post-processing by the post-processing means,
A dialog means for presenting the solution candidates of the task to the user and accepting the user's evaluation of the task solution candidates, and an evaluation of the task solution candidates from the user is obtained through the dialog means. When at least one of the task model information (directed graph or directed tree) is added and deleted, at least one directional branch is added and deleted, and at least one node is added.
By performing at least one process of addition, deletion, and change of the transition condition added to one directional branch,
It further comprises learning means for changing the task model information so as to satisfy the obtained user evaluation.

According to the present invention (claim 8), in claim 7, when the learning means obtains, through the dialogue means, evaluation information resulting from the personality of the user regarding the candidate of the solution of the task, the evaluation information is obtained. The task model information (directed graph or directed tree) unique to the user is generated based on the information, and the generated task model information is managed in association with the user, so that it is different for each user. The task having a request / solving method is characterized in that the search processing means includes user adaptation means capable of performing the search processing according to each user.

The present invention (Claim 9) is an information processing method for executing a predetermined task and one or more tasks related to the task, which should be executed first when there is no task being executed. The predetermined task is determined, and the task to be executed next is determined based on the predetermined first information when the execution status of the task being executed reaches a predetermined state. Performing a predetermined search process on the basis of the second information predetermined for the task to obtain a solution candidate for the task, and performing a predetermined post-process on the basis of the obtained solution candidate. Characterize.

According to the present invention (claim 10), in claim 9, the first information has a plurality of related tasks as nodes, and transits to a directional branch for transiting from one node to another node. A directed graph with a condition added, which is capable of making a transition from the current node to another node by verifying the transition condition and selecting a directional branch satisfying the condition, and used for verifying the transition condition When determining the next task to be executed, the node associated with the task whose execution status has become a predetermined state among the nodes included in the directed graph is used as the starting point. The node of the directed graph is transited, and the task associated with the node of the transition destination is selected as the task to be executed next.

According to the present invention (claim 11), in the method of claim 9 or 10, the second information is a process of searching for a solution candidate with each of the selectable solution candidates for the task as a terminal node. A directed graph in which the transition state is added to the directional branch for transition from one node to another node with the intermediate state of the other node, and the directional branch that satisfies the condition is selected. The task model information consisting of a directed graph that can be repeatedly transitioned from the current node to another node in accordance with the current situation and reaches the terminal node according to the current situation, and the information used to verify the transition conditions. It includes, and in obtaining a candidate for the solution of the task, repeating the transition on the directed graph with a preset node as a starting point,
It is characterized by selecting a solution candidate associated with the reached terminal node.

According to the present invention (claim 12), in the method according to claim 9 or 10, the second information is a process of searching for a solution candidate by setting each of the selectable solution candidates for the task as a leaf node. Is a directional tree in which the transition state is added to the directional branch for transition from one node to another node, with the intermediate state of the other node as another node, and the directional branch that satisfies the condition is selected By doing so, transition from the current node to another node, task model information consisting of a directed tree that can repeatedly perform node transition until reaching a leaf node according to the current situation, and for verifying the transition condition It includes information to be used, and in obtaining a candidate for the solution of the task, the transition on the directed tree is repeated from a preset node as a starting point and is associated with the reached leaf node. And selects the candidate.

According to the present invention (Claim 13), in performing the post-processing according to any one of Claims 9 to 12, at least a predetermined input / output device is used to select a candidate solution for the task. In addition to presenting to the user, accepting an instruction of adoption or non-adoption of a candidate for the solution of the task, and when an instruction to adopt is given, confirm the candidate of the solution of the task as the solution of the task. Is characterized by.

According to the present invention (claim 14), in claim 11 or 12, at the time of the post-processing, a predetermined input / output device is used to present a candidate of a solution of the task to the user, and the solution of the task is solved. Accept the user's evaluation of the candidate of,
When the evaluation is obtained, at least one node is added and deleted, at least one directional branch is added and deleted, and at least one is added to the task model information.
By performing at least one process of addition, deletion, and change of the transition condition added to one directional branch,
It is characterized in that the task model information is changed so as to satisfy the obtained user evaluation.

According to the present invention (claim 15), in accepting the evaluation, the evaluation information resulting from the individuality of the user regarding the candidate of the solution of the task is received as an input, and the evaluation of the user is accepted. When the evaluation information resulting from the individuality is obtained, when the task model information is changed, the task model information unique to the user is generated based on the evaluation information, and the generated task model By managing the information in association with the user, the search processing means can perform the search processing according to each user for a task having a different request / solution method for each user. Is characterized by.

According to the present invention (claim 16), in determining the task according to any one of claims 9 to 15, when the task cannot be determined due to lack of information, It is characterized in that the information necessary for the determination is acquired by inquiring the user through a predetermined input / output device, and the task is determined based on the acquired information.

In each of the above inventions, a program for executing a corresponding processing procedure is stored in a computer-readable storage medium as a program for controlling a computer, and the computer is caused to read the program from the storage medium. , Can be run on a computer.

The operation and effect of the present invention will be described below with reference to a specific description taking the case where the present invention is applied to a "scheduling system for supporting hospital operations" as an example.

In the conventional scheduling system, for example, a system that performs a task of scheduling a doctor only performs scheduling of a doctor, and a system that performs a task of scheduling a nurse performs only scheduling of a nurse. In order to handle the related tasks of scheduling and nurse scheduling, either develop a new system that can integrate and handle these two tasks as one task, or let the user consider the relationship between these tasks. I had to manage the consistency between those tasks myself, for example, not pairing incompatible doctors and nurses.

On the other hand, the present invention has a function of managing one or more tasks and managing the association between related tasks, and starts from the resolution of a predetermined or user-specified initial task. , When a problem solving situation for a task reaches a predetermined state, by providing a means having a function of searching for the next related task and instructing the execution of the individual task thereof, for example, a doctor's scheduling task and a nurse's scheduling It is possible to solve problems sequentially without losing consistency between tasks for a plurality of related tasks, for example, to solve problems successively by linking tasks.

Further, in the conventional system, even in a system capable of executing a plurality of tasks, the execution of the related tasks remains within the scope of the assumption of the user. Even if it could be done by linking it, it was not possible to improve the work efficiency and results by utilizing the task association that the user did not expect, such as not being able to activate the social gathering scheduling task at appropriate timing. There was also a problem.

On the other hand, in the present invention, by providing a means for managing the association between the above tasks using the task chain model and searching for a task to be executed, for example, the doctor's scheduling task ends and the next task is completed. For example, if one month or more has passed since the last social gathering, the next task is to get the social gathering scheduling as a search result and propose it to the user to use the task unexpected by the user. It is possible to improve the work efficiency and results of users by making a proposal to the user.

Further, in the conventional system, even in a system capable of executing a plurality of related tasks, in order to give individuality to the solution or solution when solving the problem of each task, Users, or knowledge, such as task chain model or individual task model changed by user, or knowledge base administrator must change knowledge base of all users for maintenance of knowledge base. There was also a problem that a heavy burden on the base manager was required.

On the other hand, in the present invention, the candidate obtained from the mechanism for searching the task association or the mechanism for determining the processing method of the individual task is evaluated by the user, and the task chain model based on the evaluation is obtained. Alternatively, by providing a means for changing the part of the individual task model related to the user, it is possible to add the individuality of the user to the problem solving without imposing a heavy burden on the user.

As described above, according to the present invention, the problem solving processes of a plurality of related tasks are associated with each other, and the problem solving process is created in a natural manner by utilizing the association of task sequences without burdening the user. It is possible to improve work efficiency and achievements.

Further, according to the present invention, it becomes possible to solve the problems with respect to a plurality of related tasks in sequence without losing consistency between the tasks and without imposing a heavy burden on the user.

Furthermore, according to the present invention, it is possible to efficiently process individual tasks by tracing out and pointing out the relationships between tasks that the user did not expect. Furthermore, according to the present invention, it is possible to add the individuality of the user to the handling of the problem of each task without burdening the user or the knowledge base manager.

[0042]

Embodiments of the present invention will be described below with reference to the drawings. This embodiment is roughly 1
It has a function to manage one or more tasks and manage the relationship between related tasks, and starts from the resolution of a given task or the initial task specified by the user, and the problem solving status of a certain task is in a given state. Then, it has the function of searching the next related task and instructing the execution of the individual task, and also searching the problem solving method of the instructed individual task, and selecting the appropriate solution for the given problem. Has the function to execute. In addition, based on the evaluations and requests obtained from the user about the solution method of each individual task and specific examples of the solution, the model for searching the problem solving method of the individual task is updated so as to satisfy the evaluation and request, As a result, the solution search for each individual task is made more efficient and more efficient, and the individual task solution method has a function to reflect the individuality of the user.

The present embodiment will be described in more detail below. FIG. 1 is a schematic configuration diagram of an information processing system according to an embodiment of the present invention. As shown in FIG. 1, the information processing system includes a dialogue unit 11, a task sequence determination unit 12, an individual task processing determination unit (corresponding to a search processing unit) 14, an individual task processing execution unit (post-processing unit). (Corresponding to the implemented one) 15, an information storage unit 13, and a processing rule learning unit 16.

In the information processing system of this embodiment, a preset task or a task instructed by the user is used as an initial task, and each of these initial tasks and a series of one or more related tasks connected from the initial tasks are individually set. The problem belonging to the task is solved in the dialogue unit 11 while interacting with the user.

The dialogue section 11 includes a task sequence determining section 12, an individual task processing determining section 14, an individual task processing executing section 15,
When there is a request from any of the processing rule learning units 16, the user is requested to input an inquiry by using the dialogue knowledge supplied from the information storage unit 13 and necessary data is acquired.
Then, the module (1
2, 14, 15 or 16).

The task sequence determination unit 12 is given an initial task after the system is activated, passes it as the current task to the individual task process determination unit 14, and activates the first individual task process. After that, the individual task processing execution unit 1
When the problem solving of each individual task in 4 has reached a predetermined state and the individual task processing determination unit 14 returns control with status information (for example, normal termination or error), the individual task to be executed next is determined.

The task sequence determination unit 12 has the current task information that was activated immediately before, and uses the information supplied from the task chain model, the reference information database and the user information database stored in the information storage unit 13 to determine the next Decide the individual task to be activated.

In the task chain model, for example, when each individual task is a node and one individual task is the current task, 0 or more conditions for selecting another individual task as the next individual task are It is a directed graph expressed as a transition condition given to a directional branch from a node corresponding to a task to a node corresponding to an individual task selected as a next individual task.

The reference information database is a database in which data necessary for verifying the transition conditions of the nodes in the task chain model and data necessary for determining and executing the processing method of the individual task and data unique to the problem are recorded. .

The user information database is a database in which information customized for individual users of the system is recorded. For example, regarding the task chain model, the user information database stores the identifier of the task chain model, the directional branch to be added and the transition condition, and the user identifier as a set.

The task sequence determination unit 12 uses the task / problem-specific information supplied from the reference information database and the user-specific information supplied from the user information database to correspond to the current task of the task chain model. A transition is made from a node to another node, and the individual task corresponding to the transition destination node is delivered to the individual task processing determination unit 14 as the current task from that point.

The individual task processing determining unit 14 is handed over the current task determined by the task sequence determining unit 12, searches and determines a solution candidate for the current task, and passes it to the individual task processing executing unit 15. When the problem solving of the current task in the individual task processing execution unit 15 reaches a predetermined state and the control is returned together with the status information, the task sequence determination unit 14 returns the control together with the status information.

The individual task processing determination unit 14 uses the information supplied from the individual task model, the reference information database and the user information database stored in the information storage unit 13 to solve the problem given as the current task. Search for candidates.

The individual task model is prepared, for example, for each individual task, and each selectable execution method (solution candidate) of the corresponding individual task is used as a leaf node, and the execution method determination process (solution candidate search process) is performed. This is the structure of the directional tree with the intermediate state of as a node.

Regarding the individual task model, the user information database stores, for example, individual task models customized for individual users of the system and user identifiers, which are stored. It is used in preference to the task model.

The individual task processing decision unit 14 uses the individual task model as a decision tree for searching a solution method of the individual task. That is, the transition from one node to another node corresponds to the transition from one intermediate state in the solution search to another state. In order to make a transition from one node to another node in the search on the individual task model, directional branches are provided between the nodes in a direction in which the transition is possible, and the directional branches are assigned to the directional branches. It is necessary to verify and satisfy zero or more transition conditions that exist.

The individual task processing decision unit 14 starts the search from the root portion of the individual task model corresponding to the current task, and from the reference information database and the user information database among the directional branches from the current node. Repeat the operation of selecting a directional branch that satisfies the transition condition based on the supplied data and making a transition, and when it finally reaches a leaf node, individually execute the individual task execution method (solution candidate) corresponding to that leaf node. It is output to the task processing execution unit 15.

The individual task process execution unit 15 solves the problem of the current task while interacting with the user through the dialogue unit 11 according to the description of the current task execution method delivered from the individual task process determination unit 14. When the final state of the designated execution method is reached, the individual task processing determination unit 14 together with the state of problem solving
Return control to

The processing rule learning unit 16 receives from the user the solution or the intermediate solution derived by each module in the respective process steps of the task sequence determining unit 12, the individual task process determining unit 14, and the individual task process executing unit 15. When the evaluation is obtained, it is activated by inputting the set of the solution or the intermediate solution and the user's evaluation for the solution.

The processing rule learning unit 16 performs a learning process based on the delivered input and a set of similar inputs obtained up to that point, and the individual task model, task chain model, and user information in the information storage unit 13 are processed. Update the database or reference information database. Which model database of the information storage unit 13 is to be updated is determined by the processing rule learning unit 13
Is determined by the task sequence determination unit 12, the individual task process determination unit 14, or the individual task process execution unit 15 and the content of the obtained input.

The processing rule learning unit 16 performs addition / deletion of nodes, addition / deletion of directional branches, addition / deletion / change of transition conditions added to directional branches, etc. Update models and databases to satisfy your rating.

Further, the processing rule learning unit 16 is provided in the dialogue unit 1.
1, when the evaluation information resulting from the individuality of the user is received as an input, the model or database stored in the information storage unit 13 is generated for each user, and the generated user-specific model or database is generated. By storing (or the difference from the reference model or database) in the information storage unit 13 and managing it by the user identifier or the like, it is possible for each user to respond to an individual task having a different request or solution for each user. It has a user adaptation function that can perform processing.

Whether or not the evaluation information is attributed to the user's individuality with respect to the model or the database is explicitly input from the dialogue unit 11, for example. The information storage unit 13
Other modules (11, 12, 14, 15 or 16) that store model data used for problem solving of this system
To supply. As described above, the information storage unit 13 has five types of database models, that is, a dialogue knowledge database, a user information database, a reference information database, a task chain model, and an individual task model, as well as a task to be activated as an initial state of the system. It has an initial task database that it remembers. If the contents of the initial task database are not empty when the system is started, the task sequence determination unit 12
Determines its content as the first current task and delivers it to the individual task processing determining unit 14.

The above is the schematic configuration of the information processing system of the present embodiment. FIG. 2 is an example of a schematic processing procedure of the information processing system of this embodiment. The overall processing flow will be described below with reference to FIG. In the following procedure, the dialogue with the user is performed using the dialogue unit 11.

When the system is started, the task sequence determination unit 12 first refers to the start task database of the information storage unit 13 to start the individual task processing from which of the individual tasks recorded in the task chain model database. Check whether the start task of whether to do is described. If the start task is recorded in the start task database (step S1), the individual task is set as the current task (step S1).
2) Go to the current task process search process (step S3).

If the start task is not recorded (step S1), the user is prompted to input the start task (step S10). When the user inputs a start task, the task is set as the current task (step S2), and the process proceeds to the current task process search process. If there is no input, it is confirmed to the user whether or not to end (step S11), and if the user requests the end, the execution is ended.
If the end is not requested, the process returns to the start task input request again.

Current task process search process (step S
3) is a process (search process) in the individual task process determining unit 14 of FIG. 1 and transits on the individual task model while referring to the contents of the user information database and the reference information database stored in the information storage unit 13. , Search the processing method of the current task.

When the current task processing search processing in step S3 is completed and processing method candidates (solution candidates) are obtained, it is confirmed whether or not the user desires to execute the processing method (step S4). If the user desires execution, the process proceeds to the current task process execution process (step S5). When the user does not agree with the processing method searched for by the system, the processing rule learning unit 16 is activated using the processing method candidate denied by the user as an example. When the model update is necessary, the individual task model is updated (steps S12 and S13), and the process proceeds to the current task search process (step S7). If the update is not necessary (NO in step S12), the individual task model is not updated and the process proceeds to the current task search process.

Current task process execution process (step S
5) corresponds to the processing (post-processing) in the individual task processing execution unit 15 of FIG. 1, and interacts with the user through the dialogue unit 11 according to the description of the current task execution method delivered from the individual task processing determination unit 14. While solving the problem of the current task.

In the current task process execution process of step S5, if the user inputs that the result is satisfied (step S6), the process proceeds to the current task search process (step S7). If the user is not satisfied with the result, the processing rule learning unit 16 is activated and the individual task model is updated when model updating is necessary (step S13), and the current task search process (step S7) is performed as described above. ).

Current task search process (step S7)
Corresponds to the processing in the task sequence determination unit 12 of FIG. 1, and while referring to the contents of the user information database and the reference information database stored in the information storage unit 13, the task chain model starting from the node corresponding to the current task Transition to the above and search for an individual task to be executed next.

If, as a result of the current task search process in step S7, the search process end node has been reached, or if all tasks on the task chain model have been completed (YES in step S8), Ask the user again to enter the start task (step S1)
0). When the next individual task candidate is newly determined as a result of the current task search process (NO in step S8), the user is asked whether or not to execute the individual task (step S9). When the user wants to execute,
The process proceeds to the current task process search process (step S3).
When the execution is not desired, the processing rule learning unit 16 uses the result of the current task search processing, which has been denied by the user until then, as an example, as in the case where the result of the current task processing search processing is denied by the user. to start. When the model update is necessary, the task chain model is updated (steps S12 and S13), and the process proceeds to the current task search process. If update is not necessary (N in step S12)
In the case of O), the task chain model is not updated and the process proceeds to the current task search process.

The above processing is repeated until the user requests the end in step S11. The above is the configuration, functions, and processing method of the information processing system according to the present embodiment. In the following, as an application example of the “scheduling system for supporting hospital operations” in the present embodiment, FIG.
A more specific description will be given with reference to FIG.

FIG. 3 illustrates a task chain model included in the example scheduling system. This task chain model shows the relations of the three types of individual tasks, namely, doctor scheduling, nurse scheduling, and social gathering scheduling.
It is described that it is necessary to wait for one month or more from the last social gathering in order to select the directed branch from the doctor scheduling task to the social gathering scheduling task. It is assumed that such a system is used as a user by an office worker of this hospital.

It is assumed that the user activates the system to schedule a doctor's examination and work for one week from April 22 to April 28, 1996, for example. At this time, when doctor scheduling is described as an initial task in the initial task database of the information storage unit 13, the task sequence determination unit 12 determines “doctor scheduling” as the first individual task according to the information, and performs individual task processing. Call the decision unit 14.

Next, the individual task processing determination unit 14 is executed by the processing shown in FIG.
The individual task model shown in FIG. 4 is executed by using the information of the reference information database shown in FIG. 4, and the candidate of the doctor who examines is determined for each date from Monday to Saturday.

As an example, the execution process for Wednesday, April 24 will be shown. First, since the scheduling is for Wednesday from the root node 100, the process proceeds to the internal medicine node 101. Although the individual task process determination unit 14 repeats the transition on the individual task model while referring to the reference information database, FIG.
Since neither doctor A nor doctor B can work on Wednesday as shown in FIG.
5 can be reached. Next, the individual task processing determining unit 14 verifies the condition under the node 105, and it is described in the reference database that the medical institution trader P is to visit this hospital on Wednesday. Therefore, the leaf node 113 is finally selected, and the candidate for the medical examination on this day is the doctor A.

The individual task processing determination unit 14 selects a doctor candidate from Monday to Saturday through the above process using the individual task model, and delivers it to the individual task processing execution unit 15 together with the result.

In this specific example, the individual task processing execution unit 15
First confirms the passed candidate by showing it to the user, confirms it, and then stores the confirmed schedule. Then, each schedule is transmitted to each doctor by executing a process such as automatically creating a schedule transmission document and transmitting it by electronic mail. When the individual task process is executed in this way, the individual task process execution unit 15 ends and returns control to the individual task process determination unit 14.

Further, the individual task process determining unit 14 returns the control to the task sequence determining unit 12. The task sequence determination unit 12 that has been returned to control restarts the execution of the task chain model while referring to the information in the reference information database.

Here, when the doctor's scheduling is normally completed, the nurse's scheduling is selected as the next individual task. In this case, the social gathering held between the doctor and the trader was held last time. Since more than one month has passed, social gathering scheduling is selected as the next task. In this case, since the doctor's scheduling has already been completed, the date of the social gathering can be decided only by looking for the day when the doctor is at work and a large number of traders are coming.

To determine whether one month or more has passed, for example, the history information of the schedule of the social gathering that has been held is recorded, and the date one month after the date of the last social gathering is It can be performed depending on whether or not it is included in the period to be scheduled by the doctor.

According to the present embodiment, in this way, the user, in this case, the office clerk at the hospital, traces out and points out the relation between the tasks which was not expected, so that the individual tasks can be efficiently identified. It can be processed.

In this specific example, the individual task process determining unit 14
Uses the individual task model of the get-together scheduling task as in the case of doctor scheduling, and determines candidate dates for a get-together by, for example, searching for days when doctors are at work and a large number of contractors are coming. . The individual task process execution unit 15 to which the holding schedule candidate is passed, first shows the candidate to the user and confirms it to confirm the candidate. Next, for example, a document for holding a reception of the social gathering is automatically generated and transmitted to the supposed attendees such as doctors, traders, and nurses by e-mail. Further, for example, the guidance to be posted on the bulletin board is automatically generated and printed. In this way, the processing of the individual task called social gathering scheduling is decided and executed.

When the social gathering scheduling is completed, the task sequence determination unit 12 selects the nurse scheduling as the next individual task, and the individual task processing determination unit 14 is called.

The individual task process determining section 14 creates a scheduling candidate for the nurse by using the individual task model and the information in the reference information database as shown in FIG. 4 in the doctor's scheduling, and calls the individual task process executing section 15.

Here, it is assumed that, for example, there is a situation in which the doctor A and the nurse J are not input to the system of being a couple. Then, the administrator of this hospital, who is the user of the system, takes the situation into consideration, and the doctor A shifts the working mode of the nurse J on the day when he / she works in the afternoon to the doctor A and the nurse J.
So that they can go home together,
It is assumed that the user manually changed the schedule created by the system. In a similar system in the past, the user had to perform such a complicated change operation each time the system was executed. Or
It was necessary to symbolize the above situation, input the information into the system, and modify the individual task model so that the information could be used. That is, in the execution of the individual task model, in order to derive a solution that reflects some circumstances, it was necessary to explicitly symbolize and incorporate the circumstances in a form that the system could handle.

In the system according to the present invention, in the above case, the individual task execution unit processing 15 is performed on the scheduling candidate created by the individual task processing determination unit 14.
At the step, the user is confirmed, and the user corrects the scheduling candidate for the nurse J. In response to the user's negative evaluation and correction of the solution derived by the system, the individual task process execution unit 15 issues a learning request to the processing rule learning unit 16. Processing rule learning unit 1
At 6, the individual task model or the user information database is modified according to the modification made by the user. Such decision tree and database correction processing parameterizes one case as one example for a plurality of scheduling cases for which a correction request is made, and Quinlan.JR, "C4.5 Pro
grams forMachine Learning ", Morgan Kaufmann. 1993
This can be easily realized by inductively learning the decision tree of the individual task model by using the machine learning mechanism disclosed in (1). In this case, the user repeats the correction of the schedule based on the above-mentioned favors, so that, for example, a description that doctor A and nurse J are a couple, or a nurse who is a couple and the husband's doctor It is not necessary to make a clear description that reflects the above circumstances by configuring a decision tree so that the working day is in the daytime, and a schedule that automatically meets the user's request, that is, a doctor A
A decision tree that assigns nurse J to the daytime is generated on the day when he works in the afternoon. Will be released.

The customized individual task model is stored in the user information database with a user identifier added to it, for example, and the customized individual task model will be used for the user from the next time.

As described above, according to this embodiment, by incorporating the machine learning mechanism as described above, it is possible to reflect the individuality of the user in the solution of the individual task. Now, when the nurse's scheduling is completed, control is returned to the task sequence determination unit 12, and the task chain model is interpreted and executed again. In this case, the end node is reached in the transition on the task chain model.

Through the above process, the present system can solve problems in sequence with respect to a plurality of related tasks without losing consistency between the tasks and without imposing a heavy burden on the user. It has become.

As described above, according to the present embodiment, the problem solving processes of a plurality of related tasks are associated with each other, and the problem solving process is performed in a natural manner by using the task sequence association without burdening the user. It is possible to create a flow, improve work efficiency and outcomes, and also derive solutions in a consistent manner among individual tasks. Further, according to the present embodiment, it is possible to add the individuality of the user to the handling of the problem of each task without burdening the user or the knowledge base manager.

In the above description, it is described that the task chain model is realized in the directed graph format and the individual task model is realized in the decision tree format. However, regarding the form of these data and the execution method thereof, It is not limited to the implementations described above. Further, in FIG. 2, the processing flow of the present system is described as a series of all the processing flows, but the present invention is not limited to such an implementation, and for example, the task series determination unit 12 It is also possible to individually separate each module such as or the individual task processing determination unit 14 into a process and realize it in a multi-process form, and such an implementation form also belongs to the technical scope of the present invention.

By the way, in the above description, a plurality of related tasks starting from the initial task are sequentially solved according to the task chain model, but subsequent tasks are not resolved before the solution of the preceding task is determined. It is also possible to boot into. For example, it is possible to suspend the determination of the solution for a certain preceding task, activate the subsequent tasks, and then return to the suspended preceding task at an appropriate timing. In addition, for example, it is possible to solve the tasks while going back and forth.

Below, an explanation will be made of the device configuration in the case of implementing this information processing system using software, with reference to FIG. In this case, as the hardware, the information processing system includes a CPU 21, a RAM 22 for storing programs and necessary data, and an input / output device 2.
4 and a storage device 25 such as a hard disk device.

The dialogue unit 11 in FIG. 1 is composed of a desired input device and an output device, for example, an input / output device 24 using a display device, a keyboard, a mouse and the like, and a program describing a processing procedure of the dialogue unit 11. It

The task sequence deciding unit 12, individual task process deciding unit 14, individual task process executing unit 15, and process rule learning unit 16 are constituted by programs describing respective process procedures.

The information storage unit 13 comprises a storage device 25. The dialogue knowledge database, the user information database, the reference information database, the task chain model, the individual task model, and the initial task database are stored in this storage device 25.

On the other hand, the processing procedure of the dialogue unit 11, the task sequence determination unit 12, the individual task process determination unit 14, the individual task process execution unit 15, and the processing rule learning unit 16 as described with reference to FIG. The described program, which is a program for controlling the computer system of FIG. 6, is stored in the RAM 22 and is executed by the CPU 21. The CPU 21 realizes a desired function by performing calculations and controlling the input / output device 24 or the storage device 25 in accordance with the procedure of the program stored in the RAM 22.

Various methods can be used to install the program in the RAM 22. For example, it is a program that describes the processing procedure of the above-mentioned program (the dialogue unit 11, the task sequence determination unit 12, the individual task process determination unit 14, the individual task process execution unit 15, and the processing rule learning unit 16 and controls the computer system. Program for storing) is stored in a computer-readable storage medium (for example, a floppy disk or a removable storage medium such as a CR-ROM). Then, as shown in FIG. 6, the program is read and stored in the RAM 22 by using the disk drive device 26 corresponding to the storage medium. Alternatively, it is once installed in the hard disk device or the like and stored in the RAM 22 from the hard disk device or the like at the time of execution.

The storage medium storing the program is I
In the case of a C card, the program can be read using an IC card reader. Alternatively, the program can be received from a predetermined interface device via the network. The present invention is not limited to the above-described embodiment, and can be implemented with various modifications within the technical scope.

[0102]

According to the present invention, by associating a plurality of related tasks with each other and automatically activating them in sequence, the consistency between the tasks is not lost, and the user is sequentially loaded without imposing a great burden on the user. It is possible to solve problems and improve the results. By tracing and pointing out the relationships between tasks that the user did not expect, it is possible to process individual tasks efficiently. It is possible to bring about a great practical effect such that the individuality of the user can be reflected in the solution of the individual task without burdening the person or the knowledge base administrator.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of an information processing system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing the flow of the overall processing of the information processing system of this embodiment.

FIG. 3 is a diagram showing an example of a task chain model.

FIG. 4 is a diagram showing an example of an individual task model.

FIG. 5 is a diagram showing an example of a reference information database.

FIG. 6 is a diagram showing a hardware configuration example of an information processing system according to the present embodiment.

[Explanation of symbols]

 11 ... Dialogue unit 12 ... Task sequence determination unit 13 ... Information storage unit 14 ... Individual task processing determination unit 15 ... Individual task processing execution unit 16 ... Processing rule learning unit 21 ... CPU 22 ... RAM 23 ... Bus 24 ... Input / output device 25 ... storage device 26 ... disk drive device

Claims (16)

[Claims] 1. An information processing system for executing a predetermined task and at least one task related to the predetermined task, wherein the predetermined task to be executed first is determined, and the execution status becomes a predetermined state. Determining means for determining a task to be executed next based on the first information determined in advance, and a predetermined search process based on the second information determined in advance for the determined task. An information processing system comprising: a search processing unit that performs a task candidate solution by performing the above; and a post-processing unit that performs a predetermined post-process based on the obtained solution candidate. 2. The first information is a directed graph in which a transition condition is added to a directional branch for making a transition from a certain node to another node, with each of a plurality of related tasks as a node. And a directed graph that can transit from the current node to another node by selecting a directional branch that satisfies the condition, and information used for verifying the transition condition. Means transits a node of the directed graph starting from a node associated with the task in which the execution status becomes a predetermined state among the nodes included in the directed graph, and executes a task associated with the transition destination node as follows. The information processing system according to claim 1, wherein the task is selected as a task to be executed. 3. The second information is such that each selectable solution candidate for the task is a terminal node, an intermediate state in the process of searching for a solution candidate is another node, and a certain node changes to another node. It is a directed graph in which a transition condition is added to a directional branch for transitioning to a node, and a transition is performed from the current node to another node by verifying the transition condition and selecting the directional branch that satisfies the condition. Depending on the situation, it includes task model information consisting of a directed graph that can repeatedly perform node transitions until it reaches a terminal node, and information used for verification of the transition conditions, and the search processing means is in advance. The transition on the directed graph is repeated starting from a set node, and a solution candidate associated with the reached terminal node is selected. Information processing system. 4. The second information is such that each selectable solution candidate for the task is a leaf node, an intermediate state in the process of searching for a solution candidate is another node, and a certain node is changed to another node. It is a directional tree in which transition conditions are added to directional branches for transitioning to a node, and transitions from the current node to other nodes by verifying transition conditions and selecting directional branches that satisfy the conditions, It includes task model information consisting of a directed tree that can repeat node transitions until it reaches a leaf node according to the current situation, and information used for verification of the transition condition, and the search process 3. The means according to claim 1, wherein the means repeats the transition on the directed tree starting from a preset node and selects a solution candidate associated with the reached leaf node. Processing system . 5. A dialogue means for interacting with a user is further provided, wherein the deciding means requires the decision through the dialogue means when the task cannot be decided due to lack of information. The information processing system according to claim 1, wherein the task is determined based on the acquired information by inquiring the user about the information. 6. A dialog means for interacting with a user is further provided, wherein the post-processing means presents a candidate of a solution of the task to the user through the dialogue means as the post-processing, and also When an instruction to adopt or not to accept a solution candidate is received and an instruction to adopt is given,
The information processing system according to any one of claims 1 to 4, wherein a candidate for the solution of the task is confirmed as a solution for the task. 7. A dialog unit for presenting a solution candidate of the task to a user and accepting a user's evaluation of the solution candidate of the task during the post-processing by the post-processing unit, and the dialog. When the user obtains the evaluation of the solution of the task through the means, at least one node is added and deleted, at least one directional branch is added and deleted, and at least one is added to the task model information. Learning means for changing the task model information so as to satisfy the obtained user evaluation by performing at least one process of addition, deletion, and change of the transition condition added to the directional branch. The information processing system according to claim 3 or 4, further comprising: 8. The learning means, when the evaluation information resulting from the personality of the user regarding the solution candidates of the task is acquired through the dialogue means, the task unique to the user is acquired based on the evaluation information. By generating model information and managing the generated task model information in association with the user, the search processing unit responds to each user with respect to a task having a different request / solution method for each user. The information processing system according to claim 7, further comprising a user adapting means capable of performing the search processing. 9. An information processing method for executing a predetermined task and one or more tasks related to the predetermined task, wherein the predetermined task to be executed first is determined when there is no task being executed. At the same time, the task to be executed next is determined based on the predetermined first information when the execution status of the task being executed reaches a predetermined state, and the predetermined task is determined in advance. An information processing method characterized by performing a predetermined search process based on second information to obtain a solution candidate for the task, and performing predetermined post-processing based on the obtained solution candidate. 10. The directed information is a directed graph in which a plurality of related tasks are used as nodes, and a transition condition is added to a directed edge for making a transition from a certain node to another node. And a directed graph capable of making a transition from the current node to another node by selecting a directional branch satisfying the condition, and information used for the verification of the transition condition. In determining a task to be executed, among the nodes included in the digraph, the node associated with the task in which the execution status is in a predetermined state is changed to a node of the digraph, and a transition destination node The information processing method according to claim 9, wherein a task associated with is selected as a task to be executed next. 11. The second information is such that each of the selectable solution candidates for the task is a terminal node, an intermediate state in the process of searching for a solution candidate is another node, and a certain node changes to another node. It is a directed graph in which a transition condition is added to a directional branch for transitioning to a node, and a transition is performed from the current node to another node by verifying the transition condition and selecting the directional branch that satisfies the condition. Depending on the situation, it includes task model information consisting of a directed graph that can repeatedly perform node transitions until reaching a terminal node, and information used for verification of the transition conditions, and a candidate for the solution of the task is In obtaining, it is characterized in that the transition on the directed graph is repeated starting from a preset node, and a solution candidate associated with the reached terminal node is selected. The information processing method according to claim 9 or 10. 12. The second information is such that each of the selectable solution candidates for the task is a leaf node, an intermediate state in the process of searching for a solution candidate is another node, and a certain node changes to another node. It is a directional tree in which transition conditions are added to directional branches for transitioning to a node, and transitions from the current node to other nodes by verifying transition conditions and selecting directional branches that satisfy the conditions, It includes task model information consisting of a directed tree that can repeatedly perform node transition until it reaches a leaf node according to the current situation, and information used for verification of the transition condition, 10. When obtaining a solution candidate, the transition on the directed tree is repeated starting from a preset node, and a solution candidate associated with the reached leaf node is selected. 1 0. The information processing method according to 0. 13. When performing the post-processing, at least a predetermined input / output device is used to present a solution candidate of the task to a user, and whether the solution candidate of the task is adopted or not adopted. 13. The information processing according to any one of claims 9 to 12, characterized in that, when an instruction is accepted and an instruction to adopt is given, a candidate for the solution of the task is confirmed as a solution of the task. Method. 14. In the post-processing, a candidate of a solution of the task is presented to a user by using a predetermined input / output device, a user's evaluation of the candidate of the solution of the task is accepted, and the evaluation is obtained. If at least one node is added, at least one node is added and deleted, at least one directional branch is added and deleted, and at least one is added to the task model information.
By performing at least one process of addition, deletion, and change of the transition condition added to one directional branch,
The information processing method according to claim 11 or 12, wherein the task model information is changed so as to satisfy the obtained user evaluation. 15. In the case of accepting the evaluation, when the evaluation information resulting from the individuality of the user regarding the candidate solution of the task is received as an input, and the evaluation information resulting from the individuality of the user is obtained. In changing the task model information, the task model information unique to the user is generated based on the evaluation information, and the generated task model information is managed in association with the user. 15. The information processing according to claim 14, wherein the search processing unit can perform the search processing according to each user for a task having a different request / solution method for each user. Method. 16. When deciding the task, if the task cannot be decided due to lack of information, the information necessary for the decision is obtained by inquiring the user through a predetermined input / output device. The information processing method according to any one of claims 9 to 15, wherein the task is determined based on the acquired information.