JP2008123022A - Presentation/restriction method for input value and input operation in interactive processing and its device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To present/restrict a value to be input/changed and a sequence of input/change operations so that predefined consistency can be satisfied between the values of each data item in inputting/changing the values of a plurality of types of data items in an interactive manner. <P>SOLUTION: When an operator inputs/changes the values of data items in an arbitrary input field displayed on a screen of a display device 3 by operating a mouse 5 or the like, the types and values of the input/changed data items are transmitted through an operation terminal and a LAN 6 to a relation manager 11 on an electronic computer 1, and converted into a content in a constraint satisfaction problem and a planning problem based on a relation table 14, and the solutions of the constraint satisfaction problem and the planning problem updated by a data manger 12 and a plan manager 13 are calculated by using a general-purpose solver 15 and a general-purpose planner, and the inputable/changeable values and inputable/changeable operations are determined from them. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a system in which an operator inputs / changes values of a plurality of types of data items through interactive (formal) processing via a computer terminal, so that the values of the plurality of data items maintain predetermined consistency. As an object, the present invention relates to a technique for appropriately presenting and limiting values to be input / changed and the execution order of operations to be input / changed in the interactive processing.

  As business systems become IT-oriented, the requirements for operator's business proficiency are becoming stricter rapidly. The requirements here are the operator's ability to respond (adaptability to the complexity and diversification of the target work, the speed and accuracy of the work itself, the admissibility of ad hoc flow changes, and coordination with other departments. Intimateness). On the other hand, due to social factors, external situations (such as difficulty in securing high-quality human resources due to increase in scale, shortening of training period and employment period) are becoming more severe due to the improvement of operator proficiency. Under such circumstances, the function of reducing the burden on the operator and supporting the ability in the interactive processing of the business system is particularly important. From this point of view, there are 4 representative background technologies (2 from the middleware field and 2 from the problem solver field).

(Middleware / workflow engine)
Conventionally, in a workflow engine and a workflow management system (WFMS) developed by applying the workflow engine, a function of executing an interactive process as specified in advance as a workflow description including conditional branching and merging has been realized. In general, when performing an input operation that maintains the default consistency between data item values in multiple stages, the input order is optimally heuristic in terms of reducing unnecessary redo. Exists. The interactive function realizes a process (known example 1) in which the system presents and restricts the next operation to the operator so as to proceed according to such an optimal order (Non-Patent Document 1). reference).

(Middleware business rule engine)
Conventionally, in a business rule engine and a business rule management system (BRMS) developed by applying the same, a function for executing IF-THEN rules for a plurality of data item values has been realized. This function checks the consistency of values, propagates values to maintain consistency, prohibits execution of predetermined operations, or presents recommended operations in the relevant interactive processing scene. Are generally realized (see Non-Patent Document 2).

(Problem solver, constraint solver)
Conventionally, the consistency that the values of a plurality of data items should satisfy is defined in advance as a constraint condition, and by specifying the compensation of the values that the data item can take, consistency can be selected from these candidates. As a computer program for enumerating combinations of values satisfying the above, what is called a “constraint solver (or constraint satisfaction solver)” is generally realized.

  Although the constraint solver is not yet generally applied to the dialog processing described above, the constraint solver has already been widely used when the processing target area has a complicated structure with respect to consistency. There are a plurality of constraint solvers, such as those that can be used for free and those that can be used for commercial purposes, such as Cream, iZ-C (see Non-Patent Document 3), ILOG Solver, and Choco.

(Problem solver / planner)
Conventionally, pre-conditions and post-conditions for executing an operation are defined in advance for each type of operation, and by specifying an initial state and a target state for the target area of the operation, A computer program called “planner” is generally realized as a computer program for obtaining the order of operations to reach a target state (see Non-Patent Document 4).

Planners are not yet commonly applied to practical systems, but they are highly specialized problem area planning systems, verification of operation principles in the technical field of coordinated processing of multiple services, and prototype systems. It has already begun to be used as a means of development.
K. Tayama, T. Maruyama, H. Uno, and T. Inoue, "An Operation Support System Architecture for Network Provisioning of Optical Access Networks", IEEE NOMS2002, Florence Italy, S24.1, April 2002, pp.829-842 "ILOG JRules 6", [online], ILOG Corporation, [October 26, 2006 search], Internet <URL: http://www.ilog.co.jp./products/jrules/whatsnew/index. cfm> "Introduction to Constraint Programming", [online], NTT DATA Sekisui Systems, Inc. [Search May 29, 2006], Internet <URL: http://www.isac.co.jp/iz/ tutorial.html> A.Gerevini and D.Long, "Plan Constraints and Preference in PDDL3", Technical Report, Department of Electronics for Automation, University of Brescia, Italy, Augsut 2005 <URL: http://cs-www.cs.yale.edu /homes/dvm/papers/pddl-ipc5.pdf> Takeshi Masuda, et al. “Process Control Technology for Realizing Flexible Systems” NTT Technical Journal, Vol. 18, No. 9, September 2006, pp. 50-54 H.Oishi, T.Nakatani, K.Tayama, S.Ogasawara, T.Yamamura, "OSS Architecture for Flexible and Efficient Process Control", IEEE NOMS2006, April 2006, pp.1-13

(Middleware / workflow engine problem)
In general, there are many business systems that perform processing in units of transactions through a plurality of stages of interactive processing with an operator. If the flow of interactive processing is simple, that is, if it is easy to enumerate all branches of the flow of processing during system development and can be defined as a flow description, the workflow engine is effective.

  However, when the flow of interactive processing is complex, the number of branches in the flow of processing increases exponentially, making it difficult to enumerate all of them, making it impossible to define it as a flow description. There's a problem. In recent years, such cases have increased in accordance with the diversification of business systems, and in particular, a business in which an input operation is performed in a plurality of stages so that values of data items maintain a predetermined consistency, for example, This tendency is conspicuous in business dealing with resource allocation problems (= constraint satisfaction problem) such as facility selection, travel reservation, staffing plan, scheduling, and the like.

  In addition, not only in resource allocation problems, but in business systems that inherently have a high degree of freedom in processing order, there may be some restrictions on the processing execution order due to external factors. For example, if the variable V1 is equal to or greater than a predetermined value, there is a restriction that the process A1 must be executed. Alternatively, in order to execute the process A2, another process A3 is preceded. If it must be running, etc. In such a case, giving the system as a workflow description in advance an order that satisfies these partial restrictions without losing the degree of freedom of the original processing order. There is a problem that it is not realistic to increase.

  On the other hand, in general, there is an input order that is optimally heuristic (empirical) from the viewpoint of reducing unnecessary redoing when allocating resources. From a practical point of view, it is possible to create a workflow so that a large percentage of transactions are executed in this order, and develop a system that does not handle exceptional flow transactions in the system. Such efficiency may not be achieved. This is because these heuristics largely depend on the type and amount of resources to be processed and the contents of the constraints, and therefore do not work effectively if they deviate from the assumed conditions. Therefore, there is a problem that an exceptional flow is unavoidable when a variety of processing targets are used and work is performed over a long period of time or multiple times. In particular, when the situation of the number of steps in the flow and lead time is long and the redo cost is high due to wide-ranging system linkage of departmental issues, the inefficiency caused by this problem will increase the efficiency of the overall business. Since it drops on average, it becomes a big obstacle in system operation.

(Middleware business rule engine problem)
It is conceivable to use a business rule engine as a means for solving the above problem. Unlike the workflow description, the business rule engine can change variable values and start processing based on complex condition judgments.

  However, the process of confirming / correcting whether or not the processing content of the IF-THEN rule in the business rule engine maintains consistency as intended is generally complicated and enormous. In particular, rule errors are likely to occur when the system is large-scale. The reason for this is that the value of the data item may be overwritten by another rule at another point during the interaction process, and the order of overwriting differs depending on the priority of the rule. Therefore, the final result is different, for example, because of the feature of the processing operation of the IF-THEN rule. That is, when the processing operation is considered as a mechanism for guaranteeing consistency, as shown in the rule description, the final data consistency is not ensured, and the operation confirmation / correction work becomes complicated. In addition, since it is necessary to actually simulate the operation for all permutation combinations of the rule execution order, and the number of cases to be considered increases exponentially, the work for checking and correcting the operation becomes enormous. . For these reasons, there is a problem that it is difficult to create and debug rules.

(Problem solver / constraint solver problem)
It is conceivable to use a constraint solver as a means for solving the above problem. Unlike the IF-THEN rule, the constraint solver guarantees that the consistency given as a constraint condition is satisfied in the finally obtained solution regardless of the progress of the process in the middle. However, the basic operation of the constraint solver is a functional operation that enumerates solutions that satisfy the constraint conditions and variables when given in a lump, and thus cannot be directly applied to the interactive processing. Therefore, for example, a method in which values input from an operator in dialog processing are sequentially added in the form of equivalence constraints, non-equivalence constraints, inequality constraints, etc., and the constraint satisfaction problem is solved by the constraint solver at each input stage (known example 2) (See Non-Patent Documents 5 and 6). The obtained solution can be used in the next input screen for presenting selection candidates for values of other data items to the operator or for limiting by checking the value range.

  However, this method has nothing to do with presentation / restriction regarding the execution order of operations. That is, the order in which the constraint conditions are added does not affect the content of the final solution, and therefore the execution order of operations for inputting and correcting values is completely free. At the same time, no recommended information is presented. At first glance, this is good for an operator who is familiar with work on the system, but from the viewpoint of reducing unnecessary redo, the system does nothing and simply delegates all processing to the operator. Only. In other words, it means that the operation is practically impossible for an operator who is not proficient in the business, and if the situation becomes complicated, it may be difficult for the skilled operator to perform the business. In the known example 2, such a problem is temporarily solved by adding a function (recommended flow presenting function) for presenting the workflow for the unskilled operator in the known example 1 described above. However, there is no effect when a situation deviates from the workflow (exception flow). That is, although the known example 2 realizes a technique for appropriately presenting and limiting the contents of values, the technique relating to the execution order of operations has the same effect as the known example 1 (in the known example 2, The two modes can be switched dynamically at runtime, but the essence of the problem remains unsolved).

  As described above, when the processing operation of the constraint solver is used in dialogue processing, control over the execution order of operations is not performed at all. There is a problem that it becomes extremely difficult. Alternatively, even when the recommended flow presentation function is applied, only the control based on the conventional fixed order is performed, so that there is a problem that the exception flow cannot be handled as in the conventional workflow management system.

  The above is a problem in the processing method when the constraint solver is applied to the interactive processing. However, in the known example 2, there is another problem in processing efficiency.

  In the above processing method, it is necessary for the constraint solver to solve the constraint satisfaction problem at each input stage, but this requires a large calculation cost. Here, it should be noted that the value of the input variable is simply confirmed (addition of a constraint on the value or change of the value domain to the input value), and the constraint propagation to the uninput variable is generated. If only the value domain is narrowed down, data item value selection candidates are not obtained. The value domain thus obtained only guarantees local consistency when focusing on a certain constraint, and does not guarantee global consistency when all values are determined. In other words, as input values in a certain scene, “all those that have at least one final solution after adopting that value” are enumerated, or “no solution that adopts that value exists, It is not possible to enumerate all the things that are logically clear that there will be no options. Therefore, even if the number of input variables is still small, that is, the solution is not sufficiently narrowed down, all the solutions (up to a huge number) when all the variables are finally input are listed. Must. As described above, when the range of the value domain is large, or when it is difficult for contradiction due to restrictions to occur, there is a problem that a very large calculation cost is required when presenting and limiting values.

(Problems with problem solvers and planners)
It is conceivable to use a planner as a means for solving the problem in the processing method of the constraint solver. The planner has a function of calculating the execution order to reach the target state based on the state at an arbitrary time point and information on the precondition / postcondition for each operation, and presenting any one of them. In other words, by using the planner output, it is possible to present an appropriate execution order in the interactive processing.

  However, on the other hand, it is not possible to directly present or restrict the next operation performed by the operator on the interactive processing even with the planner function. That is, as a next operation in a certain scene, it is logical to enumerate all those that have at least one execution order to reach the target state, or it is logical that there is no such execution procedure and it becomes clogged in the middle There is a problem that it is not possible to enumerate all obvious things. This is based on the function that, in any of the currently proposed planners, any one of the execution orders from a scene to the target state is given, or that there is no execution order to reach the target state. In order to check all the execution orders, the execution time is in the exponential order, so that the processing is substantially impossible.

  If this problem cannot be resolved, it is necessary to determine the next operation that should not be executed from the global viewpoint of whether or not the target state can be finally reached, and to execute the operation without noticing it in the interactive process. You can't limit it. In particular, in a large-scale system, when the variety of operations and the complexity of states increase, it becomes difficult to grasp the global situation. That is, even if the planner is used, the effect in terms of providing a function for reducing the load on the operator and supporting the ability becomes relatively small.

  On the other hand, planners and constraint solvers (two are collectively called problem solvers) have the perspective of how to internally represent the problem they are dealing with and how to proceed with what event This is an inherently different operation method. Although a system that calls and uses these two problem solvers from a hard-coded program for a specific problem area can be easily realized, there is a problem that they cannot be linked in a general-purpose manner. In particular, it is important to realize a cooperative operation for both data internal representation and event processing.

  For example, in terms of internal expression, the internal expression in the planner represents the action in the external world to be planned by data called “action”, and the “state” of the external world and “pre-condition / post-condition of action” are predicated on the predicate logic. This is expressed as a set of expressions. The value actually input from the operator is internally expressed as a predicate related to the data item (type) and its value. On the other hand, the internal representation in the constraint solver is expressed by constraint variables, value domains, and constraint conditions, and can be entered as a value candidate that can be input by an operator on a certain operation screen (in reality, the global consistency cannot be satisfied as described above. Is also included as a value domain taken by the constraint variable. The value actually input from the operator is internally expressed as a constraint condition that binds the constraint variable to the value. Thus, since there is a difference in the data representation method inside the computer for the same value, it is necessary to perform mutual conversion between the two.

  In addition, for example, in the event of an event, in the planner, data input by an operator (more specifically, equivalent to pressing an “OK (confirmation)” button on the input screen) corresponds to execution of an action, and a change in data value occurs. This corresponds to a change in “state (predicate set)”. This necessitates recalculation of the next action that can be performed. On the other hand, in the constraint solver, data input by an operator corresponds to addition of a constraint condition, and a change in data value corresponds to a change in value domain. This makes it necessary to recalculate changes in the value domain of other variables. As described above, since there is a difference in the operation contents in the computer for the same event, it is necessary to perform mutual communication between the two.

(Problem of the present invention)
As described above, the technique for presenting and restricting either “contents of values to be input / changed” or “execution order of operations to be input / changed” is a conventional technique (that is, the former is a constraint solver, The latter can be realized by applying a general-purpose planner), but there is no technology that satisfies both requirements by using two problem solvers simultaneously.

  In order to solve the above-described problems, a means for operating the constraint solver and the planner in cooperation is provided.

(Basic configuration)
FIG. 1 shows an outline of an apparatus for executing the interactive processing of the present invention. In the figure, 1 is an electronic computer (computer, server), 2 is an operation terminal, 3 is a display (display device), 4 is a keyboard, 5 is a pointing device (mouse), and these constitute hardware. The computer 1 and the operation terminal 2 are connected by a communication line such as a LAN 6. The computer 1 and the operation terminal 2 may be the same computer. The keyboard 4 and the mouse 5 together constitute an input device.

  The software configuration is as follows. That is, reference numeral 10 denotes an interactive processing execution program according to the present invention, which is composed of a relation manager 11, a data manager 12, and a plan manager 13, and calls a method mutually. The relation manager 11 has a relation table 14 described later, the data manager 12 has a known general-purpose constraint solver (program) 15, and the plan manager 13 has a known general-purpose planner (program) 16. The interactive processing execution program 10 may be realized in a form incorporated in a predetermined application program, or may be realized in a single form separate from the predetermined application program.

  FIG. 2 shows details of the software configuration of the apparatus for executing the present invention. In the figure, 11 is a relation manager, 17 is a database manager, and 20 is an application (program). Reference numeral 121 denotes a data controller, and 122 denotes a verifier, which constitute the data manager 12 described above. Reference numeral 131 denotes a plan controller, and 132 denotes a planner. These constitute the plan manager 13 described above. The details of each software will be described later.

  Each of the software elements described above exists in a distributed manner on a plurality of hardware connected to each other via an appropriate communication path, and can be executed while communicating with each other. For example, the application 20 may be executed on the same computer 1 as other software, and the operation terminal 2 may be provided with a screen and input / output means to the operator in the form of a WWW browser program via a Web service. Only the application 20 may be executed on the operation terminal 2 and may cooperate with other software executed on the electronic computer 1 via the LAN 6.

(Outline of means to link constraint solver and planner)
[Conversion of value contents from input to data controller / plan controller / communication of execution event]
In order to appropriately present and limit “values to be input / changed” and “execution order of input operations”, the constraint satisfaction problem (and its solution) held by the data controller 121 and the planning problem held by the plan controller 131 And change according to the input operation of the operator. That is, based on the value input by the operator by executing the input operation, the constraints of the constraint satisfaction problem are added / deleted, and the predicate set representing the state of the planning problem is added / deleted.

  For example, consider an operation in which the operator inputs a value of 0 in the input field A on the data input screen of the application 20. In this case, as a constraint corresponding to this, if the variable of the constraint satisfaction problem corresponding to the input field A is α, the constraint “α = 0” is added to the constraint satisfaction problem, and the updated constraint satisfaction problem is made into the constraint solver. Let them solve and update the solutions they hold. Also, if the corresponding predicate is “(a input)”, the predicate “(a input)” is added to the state of the previous planning problem and updated to a new planning problem.

[Conversion of value contents and transmission of execution event from data controller to plan controller]
Further, when the solution held by updating the constraint satisfaction problem changes, the planning problem is updated in a chained manner. The solution here means a solution set when the constraint satisfaction problem has a plurality of solutions. The case where the solution changes corresponds to the case where the compensation of values that the constraint variable can take as a solution changes.

  For example, consider a constraint satisfaction problem in which variables are x and y, domains are both {0, 1}, and constraints are x ≠ y (a satisfaction solution at this time is (x = 0, y = 1) or ( x = 1, y = 0)). Here, if a value input operation occurs from the application 20 and a constraint of “x = 0” is added, the satisfactory solution is x = 0, y = 1. If the value of y is determined to be one and is expressed internally in the planning problem as a predicate (y input) corresponding to the fact that the value has been determined, this is expressed as a predicate set representing the state of the planning problem. to add.

(Specific implementation method of means to link constraint solver and planner)
[Value / event correspondence table for linkage]
The above problem updating function is performed in cooperation with the data controller 121, the plan controller 131, and the relation manager 11. The data controller 121 holds the constraint satisfaction problem and its solution, and further holds the relationship between the contents of the input operation and the corresponding constraint addition / deletion. Similarly, the plan controller 131 holds the planning problem, and holds the relationship between the contents of the input operation and the corresponding predicate, and the relationship between the solution of the constraint satisfaction problem held by the data controller 121 and the corresponding predicate.

  It should be noted that the relationship held by the data controller 121 and the plan controller 131 need not define the addition / deletion of constraints or the addition / deletion of predicates for every operation. Similarly, it is not necessary to define addition / deletion of predicates for the solution of any constraint satisfaction problem.

  In order to manage these relationships, the relationship manager 11 holds a relationship table 14 including input operation names (action names) targeted by the data controller 121 and the plan controller 131. In the problem update process, the relationship manager 11 first receives an event notification indicating that an input operation has been performed from the application 20. This event notification includes the action name. The relation manager 11 compares the input operation name in the relation table 14 with the received action name, and determines whether or not to update each of the constraint satisfaction problem and the planning problem.

  FIG. 3 shows a basic relationship table. The relation table 14 includes a column in which an action name representing the event content is stored, a column that represents a change content to the constraint satisfaction problem (constraint solver), and a column that represents a change content to the planning problem (planner). Is done.

  From the relation table 14, a method for transmitting an operation according to an event and a method for converting an internal representation of a value between problem solvers will be described. FIG. 4 shows an operation corresponding to the content of the input operation.

  First, a search is made for a row in which the ID of the input field (representing the type of data item) entered by the operator matches the value in the “contents of input operation” column of the relation table (201). Next, referring to the “change to constraint satisfaction problem” column of the row, a constraint condition that the value of the variable corresponding to the constraint variable ID is the same as the value of the input field is added to the constraint satisfaction problem (202 ). Next, referring to the “change to planning problem” column of the row, the predicate held therein is added as a planning problem state (predicate set) (203).

  In this process, when the value of another variable is determined by constraint propagation, the state of the planning problem corresponding to the value is added. FIG. 5 shows a secondary planning problem change operation by constraint propagation. First, a variable whose value is determined by constraint propagation is extracted (204). Thereafter, referring to the “change to planning problem” column of the row, the predicate held therein is added as a state (predicate set) of the planning problem (205).

[Operation switching according to the transmission pattern (there are three types) and each flow]
With the method described above, the relation manager 11 can know the relationship between the internal representation of values in each of the data controller 121 and the plan controller 131 and the operation for each event. As a result, the processing procedure branches into three depending on the combination of problems to be updated. The flow of each process is shown in FIGS.

(Pattern 1)
When the problem to be updated is only the constraint satisfaction problem (FIG. 6), the relation manager 11 transmits the contents of the input operation to the data controller 121 (FIG. 6 (2)). The data controller 121 converts the received operation content based on the relationship between the operation content held and the constraint, adds or deletes the constraint, and updates the constraint satisfaction problem (FIG. 6 (3)). . Finally, the data controller 121 transmits the updated constraint satisfaction problem to the general constraint solver 15 and obtains the solution from the general constraint solver 15 (FIGS. 6- (4) (5) (6)).

(Pattern 2)
When the problem to be updated is only the planning problem (FIG. 7), the relation manager 11 transmits the operation content to the plan controller 131 (FIG. 7- (2)). The plan controller 131 converts the received operation content based on the relationship between the retained operation content and the postcondition, reflects the postcondition in the state, and updates the held planning problem (FIG. 7-). (3)).

(Pattern 3)
When the problem to be updated is both the constraint satisfaction problem and the planning problem (FIG. 8), the relation manager 11 first transmits the operation content to the data controller 121. The received data controller 121 performs the process 1 described above (FIG. 8- (2) (3) (4) (5) (6)). Thereafter, the relation manager 11 acquires the solution or part of the updated constraint satisfaction problem and transmits it to the plan controller 131 (FIGS. 8- (7) (8)). The plan controller 131 converts the received solution or a part of the solution into a postcondition, reflects it in the state, and updates the held planning problem (FIG. 8- (9)).

  By performing the above processing, the values to be input / changed and the execution order of operations are dynamically presented and restricted, and the operator can be supported appropriately.

  FIG. 9 shows a flowchart of the entire interactive processing according to the present invention including the branch processing described above.

(Detailed description of means for presenting and restricting value candidates when inputting and restricting)
As pointed out in “Problem Solver / Constraint Solver Problems” described above, it is impossible to simply present and restrict candidates of input values on the operation screen in interactive processing from the function of the constraint solver. Therefore, the data controller 131 achieves this by cooperating with the constraint solver 15. This function is a value that does not cause inconsistency between the value domain of the corresponding constraint variable, the range of the value domain corresponding to other variables, and the values of other variables input up to that step. This is realized by extracting. The data controller 131 executes this function in response to a request from the application 20, and presents (recommended) values that can be input on the operation screen for each input data item to the application 20.

  10 and 11 show specific flowcharts for realizing this function. This function has two methods with different processing characteristics.

FIG. 10 shows one of these methods. First, the data controller 131 performs initialization processing based on a constraint satisfaction problem (a value input by the operator up to that point is added as a value constraint) held at the time when the operation screen is to be displayed. All solutions are obtained in (301). First, receive the constraint variable for which you want to obtain a recommended value from the outside (Note that at this time, you can also receive the constraint variable you want to obtain from the outside as an argument such as a list or array at one time, and select one of them. (302). Next, set A is an empty set, and the following processing is repeated for each element of the solution set (303).
1. The value taken by the variable in the solution is added to the set A avoiding duplication (304, 305).
After that, the contents of set A are presented to the outside as a recommended value set. (In this case, even if the constraint variables that are to be returned to the outside are temporarily accumulated for the temporary variables for return such as lists and arrays, there is an essential difference. (No.) (306). The above processing is repeatedly executed for all constraint variables to be displayed on the operation screen (307).

  This method is called method A, and method A has a high calculation cost because all solutions are obtained. However, even when there are a large number of variables to be examined, there is a feature that the constraint satisfaction problem can be solved once.

FIG. 11 shows another method. First, the data controller 131 receives a variable to be displayed on the operation screen from the outside (401). Next, set A is an empty set, and the following processing is repeated for each value in the value domain of the variable (402).
1. A value constraint on the value is added (or a new value domain is used) to solve the constraint satisfaction problem. If there are one or more solutions, the value is added to the set A (403, 404).
Thereafter, the contents of the set A are presented to the outside as a recommended value set (405). The above processing is repeatedly executed for all the constraint variables to be displayed on the operation screen (406).

  This method is called method B. In method B, the constraint satisfaction problem must be solved for each variable to be examined, but if one solution is found, the calculation in the constraint solver may be interrupted, so calculation per time The cost is small.

  For the above two methods, considering the trend of the constraint satisfaction problem that is the target, Method A queries the recommended value for the change frequency of the constraint satisfaction problem to be retained when the number of solutions to the problem size is relatively small The method B is effective when the frequency is relatively high, and the method B has a relatively low frequency of querying the recommended values for the change frequency of the constraint satisfaction problem to be retained when the number of solutions for the problem size is relatively large. It is considered effective in some cases.

  As described above, method A can obtain a value candidate at a small calculation cost when there are many variables to be examined. In Method B, value candidates can be obtained at a low calculation cost in the case where the narrowing down of solutions due to constraints is gradual and contradiction does not easily occur and the number of final solutions becomes enormous.

  In order to improve versatility, a method of switching the above two methods according to the nature of the constraint satisfaction problem to be handled can be considered. The simplest method is a method of switching manually by an argument at the time of starting a system program or a configuration file.

  On the other hand, as a method of automatically switching, there is a method of estimating the number of solutions of the constraint satisfaction problem. Specifically, an estimate of the size of the problem (considered to be proportional to the number of solutions) is calculated from the number of variables constituting the constraint satisfaction problem, the number of constraint conditions, and the value domain range, and set in advance separately. Judgment is made based on whether or not the specified value is exceeded. Since there is no need to find a solution, the cost of this calculation does not hinder the dialogue processing.

  In general, the solution to the constraint satisfaction problem tends to increase as the value domain of each variable and the number of variables increase, and decrease as the number of constraint conditions increases. Therefore, in order to estimate the size of the problem, define a value that increases as the value domain size and number of variables for each variable increases, and decreases as the number of constraints increases. What is necessary is just to define a threshold value.

  For example, specifically, the following mathematical formula is used as the estimated value (Ed) of the problem size. The narrowing-down coefficient Cr is a real number from 0 to 1, and is set to a small value depending on the problem when the solution narrowing down is large due to strong restrictions. When the number of constraint conditions is 0, Ed represents the number of combinations of variable values. When the estimated value does not exceed the set value, it is determined that the calculation cost for obtaining all the solutions does not exceed the standard, and the method A is applied, and when it exceeds, the method B is applied.

Number of constraint variables: Nv
Number of constraints: Nc
Average number of elements in the value domain: Nd
Refinement factor: Cr
Estimated problem size: Ed = Cr ^Nc · Nd ^Nv
Based on this mathematical formula, the estimation model can be refined by considering the average number of variables constrained by one constraint condition and the strength with which each constraint condition narrows down the solution.

  Furthermore, as a method of automatically switching according to the characteristics of each operation screen, there is a method of investigating a set of input data items on the operation screen. Specifically, an estimate of the number of inquiries is calculated from the number of input data items and the size of the value domain, and a determination is made based on whether the value exceeds a preset value separately. Since this is a simple process based on numerical values known in advance for the operation screen, the cost for this calculation does not hinder the interactive process.

  Specifically, the following formula is used as the estimated value (Rf) of the number of inquiries.

Number of input data items on the operation screen: Ni
Average number of elements in the value domain: Nd
Estimated number of inquiries: Rf = Ni · Nd
The method A is applied to the problem that the estimated value exceeds the set threshold value, and the method B is applied when the estimated value does not exceed the set threshold value. As the threshold value, the number of times of calculation for enumerating all solutions (corresponding to the method A) corresponding to the number of calculations for determining the presence or absence of the solution (corresponding to one inner loop of the method B) is set. . Depending on the problem area, if the latter calculation cost is expected to be small, a large value is set.

  As described above, as a combined method of Method A and Method B, the processing content is adaptively switched according to the size and nature of the constraint satisfaction problem and the contents of the operation screen, so that there is no bias to the constraint satisfaction problem with a wide range of properties. Calculation cost can be reduced.

(Detailed explanation of the means for presenting and limiting the execution order of input operations)
As pointed out in “Problem solver / planner problem” mentioned above, it is simple to present and restrict the execution order in interactive processing of operations on the terminal (corresponding to “action” in the internal representation of the planner). However, the function of the general planner 16 cannot be realized. Therefore, the plan controller 131 solves the problem by cooperating with the planner 132. This function is realized by extracting those operations that satisfy the preconditions in the current state and can be executed, and have an execution order that reaches the target state after the execution of the operation. The plan controller 131 executes this function in response to a request from the application 20 and presents the action to be executed next to the application 20.

  FIG. 12 shows a specific flowchart for realizing this function. First, a set of required actions is set as ACT, and an empty set is set as an initial value (501). Next, the plan controller 131 performs the following processing for each action of the action total set O held (502).

  1. An unprocessed operation is taken out from O and is set as o (503).

  2. 2. Determine whether the preconditions for operation o are met in the current state, and if so, Proceed to the process. If not, 1. Returning to step 504, processing is performed for other unprocessed operations (504).

  3. Applies the post condition of operation o to the current state and updates the state. This updated state is set as s (505).

  4). The operation order from the state s to the target state is obtained by the planner 132 (506).

  5. If there is an execution order, 6. Proceed to the process. If it does not exist Returning to step 507, other unprocessed operations are processed (507).

  6). Add operation o to ACT. 1. Return to (508).

  Thus, 1. To 6. By performing the above process on each input operation, the operation to be executed next is added to the ACT.

  According to the present invention, by associating the planner and the constraint solver via the relation manager, each element of the predefined action set is determined according to the value of the data item input in each scene on the interactive processing. It is possible to dynamically calculate whether or not the action is possible, and it is possible to present and restrict the next selectable operation to the operator. In addition, even when the value of the constraint variable is narrowed down by constraint propagation in the constraint solver, the possibility of the next action can be dynamically calculated as necessary, and the next selectable operation is presented to the operator. This produces the effect that it can be restricted.

  As described above, both the “values to be input / changed” and the “execution order of operations to be input / changed” absorb the differences in internal representation and operation while maintaining the versatility as a problem solver. The planner and constraint solver can be linked.

  In addition, according to the present invention, in all scenes that can occur in interactive processing from the viewpoint of achieving business objectives (pointing both ensuring value consistency and reducing unnecessary redoing). The operator can be supported appropriately. As a result, the problem of the workflow engine that the occurrence of an exceptional flow is inevitable can be solved.

  In addition, according to the present invention, regardless of the exponential increase of scenes that may occur in the dialog processing, the content of the finally required value consistency and the pre-conditions and post-conditions that always hold in any scene for a certain operation As described above, it is possible to execute a desired support content by describing it directly. As a result, it is possible to solve the problem of the workflow engine and the business rule engine that it is difficult to create and debug the workflow business rule.

  Further, according to the present invention, two algorithms for obtaining a recommended value by solving a problem in the constraint solver are prepared on the data controller, and can be used properly according to the problem. As a result, it is possible to suppress the calculation cost evenly for problems of a wide range of properties and solve the problem of the constraint solver that the calculation cost increases for problems of a certain property.

  Further, according to the present invention, the operation to reach the target state or the operation that becomes clogged can be enumerated by calling the calculation suitable for the planner (in any one of the orders) multiple times by the plan controller. This solves the planner's problem that the execution time is exponential when trying to enumerate all the orders.

  As described above, it is possible to solve the problems of the workflow engine, business rule engine, constraint solver, and planner, reduce the operator's load on the interactive processing of the business system, and provide the function to support the ability it can.

  An embodiment of an apparatus for executing the interactive processing according to the present invention will be described with reference to FIG.

  First, the function of each software element in FIG. 2 will be described.

  The relation manager 11 has a function of receiving an operation performed by the operator in the application 20 and transmitting the contents to the data controller 121 and the plan controller 131. Further, the relation manager 11 reads the relation table 14 that regulates the cooperative operation from the file system or the like, and performs processing based on this.

  The data controller 121 holds a constraint satisfaction problem (a set of variables, constraints, variable domains) and its solution. Further, the data controller 121 has a function of updating the constraints according to the content of the operation performed by the application 20. The data controller 121 also presents and restricts the contents of input / changed values to the application 20 by causing the general-purpose constraint solver 15 to solve the held constraint satisfaction problem and obtaining a solution. I will provide a.

  The verifier 122 is a wrapper (or adapter) for the general-purpose constraint solver 15. As the general-purpose constraint solver 15, there is general-purpose software that is open to the public for free. When this is used, the verifier 122 receives a processing request from the data controller 121, and performs request content conversion and supplementary processing in a form suitable for the interface and function of the general-purpose constraint solver 15. The definition 123 given to the general constraint solver 15 is an internal representation of the constraint satisfaction problem. In accordance with the contents of this definition 123, constraint variables and value domains are read from the database manager 17, and a constraint satisfaction network is created.

  The usual method of defining and assigning data necessary for processing directly as a constraint satisfaction problem is sufficient for design work solving a specific optimization problem. However, when dealing with a plurality of transactions as in the business system to be solved by the present invention, it is not practical to prepare a definition for each transaction, and a separate technical solution is required. By referring to the database as in the present embodiment and dynamically taking the constraint satisfaction problem, the application area of the present invention can be expanded and a great effect can be exhibited.

  The plan controller 131 holds a planning problem (current state / collection of all operations / target state). Further, the plan controller 131 has a function of updating the current state in accordance with the content of the operation performed by the application 20. In addition, the plan controller 131 provides the application 20 with a function of presenting and restricting the execution order of operations by allowing the general-purpose planner 16 to solve the held planning problem.

  The planner 132 is a wrapper (or adapter) for the general-purpose planner 16. As the general-purpose planner 16, there is general-purpose software that is open to the public for free. When this is used, the planner 132 receives a processing request from the plan controller 131 and performs conversion of request contents and supplementary processing in a form suitable for the interface and function of the general-purpose planner 16. The definition 133 given to the general-purpose planner 16 is an internal representation of the planning problem. In accordance with the contents of this definition 133, an internal representation of the planning problem is created.

  Next, FIG. 13 shows an embodiment of a relation table. The following two points are expanded from FIG.

(Extension 1)
When it is desired to change the operation according to the contents of the value input for a certain input field, it is realized by describing a conditional expression for the contents of the value. In this case, for the confirmation of the value from the input field in the “Contents of Input Operation” column to the variable in the “Change to Constraint Satisfaction” column, a constraint condition is added so that the latter is true when the former is true. That means.

(Expansion 2)
The above conditional expression is extended to include a method name that returns a Boolean value and a method name that returns an arbitrary value. This makes it possible to call an arbitrary process in the application and acquire various states of the application, so that high extensibility can be obtained. In this implementation, the relation manager 11 has a function of interpreting and executing the conditional expression and method name.

(Example of application)
An operation when an application having a function of presenting / restricting input values and input operations in interactive processing is executed will be described.

  FIG. 14 shows a configuration example of the screen, and FIG. 14A shows an example of the input operation screen. The operation order of two or more data items displayed on one input operation screen is not limited, and a plurality of data items can be input simultaneously from the screen. Press the “OK” button to reflect the value entered on the GUI to the system. Press the “Cancel (Return)” button to ignore the entered value and return to the previous screen. FIG. 14B shows a next operation selection screen. A plurality of buttons for shifting to the next executable operation are arranged on the screen. For example, when the “input operation 02” button is pressed, the input operation screen is displayed. Transition to. FIG. 14C shows a screen transition history generated by sequentially performing input and selection. The input operation screen and the next operation selection screen are alternately shifted.

  FIG. 15 shows an application example in which the two types of screens shown in FIGS. 14A and 14B are integrated into one screen and the operator can input data while confirming the next operation option. Yes (in this case, the transition history is not alternated as shown in FIG. 14C, and the same configuration as in FIG. 15 is used, and screens with different types of data items and different operation selection buttons are connected).

  On the other hand, the internal configuration of the application follows the configuration of the application, the relation manager, the plan controller, and the data controller shown in FIG.

  The application 20 inquires of the type of input screen that can be executed next (can reach the target state) from the plan controller 131. Based on this, the buttons at the bottom of FIG. 14B or FIG. 15 are displayed. It should be noted that the button is not displayed so that the screen does not transit to an input screen where it is clear that the target state cannot be reached even if it is executed (transition is restricted). In this way, regarding the operation, the operator support is performed according to the scene of the dialogue processing.

  Further, the application 20 notifies the plan controller 131 that the OK button has been pressed. The plan controller 131 stores that the operation has been executed on the previous input screen (displayed so far). This information is referred to when determining whether or not an operation has been performed in the conditional branch 502 in FIG.

  The application 20 inquires of the data controller 121 about a value that can be input to the data item (which can finally satisfy the consistency of the value). Based on this, the input combo box options for each data item in FIG. 14A or FIG. 15 are displayed. It should be noted that the corresponding option is not displayed so that a value that is clearly not satisfactory even if it is input cannot be input (value is limited). As described above, the operator is supported in accordance with the scene of the dialogue processing regarding the input value.

  Further, the application 20 notifies the relationship manager 11 that the OK button has been pressed. At the same time, a set of the value input by the operator and the name of the data item (ID representing the input field) is sent to the relation manager 11. The relation manager 11 performs processing based on the relationship table in FIG. 3 and the flowcharts in FIGS. 4 and 5.

(Specific examples of constraint satisfaction problems and planning problems)
A specific operation example of the constraint satisfaction problem / planning problem in the apparatus and application described above will be described.

  FIG. 16 shows a constraint satisfaction problem composed of six or more variables. Each variable has a value domain such as “1-4”. There are three or more constraints between variables. FIG. 17 shows a planning problem in which four or more input operations are expressed as actions. A state with no predicate is specified as an initial state, and a state in which all variables are “inputed” is specified as a target state.

  The parts V1, V2, and V3 will be described. A1 is a V1 input screen. Similarly, A2 is an input screen of V2, A3 is an input screen of V3, and A4 is an input screen of V4. Since V1 to V3 are not “inputed” in the first stage of the interactive process, it is calculated by the process of FIG. 12 that execution of A1 to A3 is necessary to reach the target state. Since A4 has a precondition and “(V3 input)” has not yet been established, it is calculated that it cannot be executed as the next operation. As a result, in the first stage, any input operation of V1 to V3 can be selected, while the input operation of V4 cannot be selected yet.

  Here, it is assumed that the user selects the input operation A1. By applying the processing of FIG. 10 or FIG. 11 to the constraint satisfaction problem, the element “4” of the value domain D1 that is clearly unsolved from the constraint condition C12 is removed, and “1, 2, 3” is the input value. It becomes an option. Thereby, presentation / restriction of values is realized.

  Here, it is assumed that V1 = 1 is input at A1. The relation table is referred to in the process s1 of FIG. 9, and the constraint condition “V1 = 1” is added to the constraint satisfaction problem by the processes s2 and s3. By the processes s4 and s5 in FIG. 9, constraint propagation occurs and V2 = 1 and V3 = 1 are determined. Through the process s6 of FIG. 9, the relation table is referred again, and “(V2 input)” and “(V3 input)” are added to the state of the planning problem. Next, the process of FIG. 12 is executed by the processes s8 and s9 of FIG. Since the “(V2 input)” and “(V3 input)” portions of the target state are already satisfied, A2 and A3 are not displayed as operations that can be selected next. On the other hand, since the precondition of A4 is satisfied, A4 is newly displayed.

  Here, as another possibility, it is assumed that V1 = 2 is input in A1. V2 = 2 is determined by constraint propagation. V3 is narrowed down to “2 or 3” but not confirmed. As a result, A2 is not displayed as the next executable operation, A3 is still displayed, and A4 is newly displayed.

  In this way, dynamic adaptation of interactive processing is realized by the input value. Even when A2 or A3 is first executed, the operation screen is omitted when the value is determined in the same manner as described above, or the operation screen is displayed when the precondition is satisfied.

  The above description assumes that one variable is input in one action. However, an actual operation screen also requires an operation screen for inputting a large number of variables at once. Further, it is assumed that only several value candidates are taken as the value domain. However, an operation screen for inputting from a very large number of value candidates such as “1 to 100” in the value domain is also necessary. For such an operation screen, the processing of the method B in FIG. 11 has a high calculation cost. Therefore, this problem can be solved by appropriately using the method A.

  In the above description, an input data item is handled as an operation target in processing, and whether or not an input has been input (“input”) as a specific state in processing. Here, an internal variable “alpha” representing various states of the application, and “one”, “greater-than-one” representing the contents of the values by using the relationship table of FIG. Can handle different states.

  Moreover, although the above description has dealt with completion of input of all data items as the target state, a wide variety of target states can be handled by the above expansion. For example, practical applications such as limiting the required items, changing the required items depending on the value of a certain data item, or using a specific state of an operation target other than the data item as a condition can be considered.

  The effect of the present invention, that is, the function of appropriately presenting and restricting the content of data item values and the execution order of operations to be input / changed in the interactive process assumes the support function that the system performs for the operator. ing. However, the present invention can be easily applied to cooperation between systems or one-to-many cooperation. That is, the present invention can be used in technical fields such as EAI and SOA.

The block diagram which shows the outline | summary of the apparatus which implements this invention The block diagram which shows the detail of the software configuration of the apparatus which performs this invention Explanatory diagram showing an example of a basic problem change correspondence table Flowchart of processing for converting the contents of input operations into the contents of restriction satisfaction and planning problems Flowchart of processing to convert constraint propagation into planning problem changes Explanatory drawing which shows an example of the update procedure of the problem accompanying input operation Explanatory drawing which shows the other example of the update procedure of the problem accompanying input operation Explanatory drawing which shows the further another example of the update procedure of the problem accompanying input operation Flowchart of overall interactive processing according to the present invention Flowchart of one process for presenting and limiting input / change value candidates Flow chart of other processing that presents and restricts input / change value candidates Flowchart of processing that presents / limits the execution order of input operations Explanatory drawing which shows an example of embodiment of a problem change correspondence table Explanatory drawing showing an example of screen structure in application Explanatory drawing showing another example of the screen configuration in the application Explanatory diagram showing a specific example of the constraint satisfaction problem Explanatory diagram showing specific examples of planning problems

Explanation of symbols

  1: computer, 2: operation terminal, 3: display, 4: keyboard, 5: pointing device (mouse), 6: LAN, 10: interactive processing execution program, 11: relation manager, 12: data manager, 13: Plan manager, 14: relation table, 15: general constraint solver, 16: general planner, 17: database manager, 20: application program, 121: data controller, 122: verifier, 131: plan controller, 132: planner, 123, 133: Definition.

Claims (10)

For a predetermined application program operating on an arbitrary computer, at least one input / change column for data items is included in the display device of the arbitrary computer or another computer connected to the arbitrary computer via a network. An input operation screen is displayed, and the value of the data item in the input / change column is input based on an operator's operation from an input device of the arbitrary computer or another computer connected to the arbitrary computer via a network.・ Change and input the value of two or more types of data items interactively by switching this screen for two or more input operation screens including input / change fields corresponding to different types of data items. The default consistency between the values of each data item Satisfying manner, a method of presenting, limits and values entered or changed in the input-changing field, and the execution order of operations of inputting and changing through said input operation screen,
A constraint solver and a planner operating on the arbitrary computer or another computer connected to the arbitrary computer via a network;
A variable, a value domain of the variable, and a constraint condition stored in a storage device of the arbitrary computer or another computer connected to the arbitrary computer via a network, and values of the plurality of types of data items And a constraint satisfaction problem in which a predetermined consistency that should be satisfied by the values of the plurality of types of data items is a constraint condition, and a set of operations including a target state, a current state, a precondition and a postcondition, A set consisting of a set of a processing operation target and a specific processing state for the operation target is a target state, and a predicate consisting of a main part for specifying the processing operation target and a predicate indicating the state Constraints in planning problems that represent the current state, preconditions and postconditions, and constraint satisfaction problems that correspond to the types and values of data items that have been input or changed Using a relation table that the changes made by the description of the predicates in the changes and planning problems,
The arbitrary computer or another computer connected to the arbitrary computer via a network,
The type and value of the data item input / changed from the input device is converted into the change content of the constraint condition in the constraint satisfaction problem with reference to the relation table, or the change content of the predicate in the planning problem,
Add / delete constraint conditions according to the change contents of the constraint conditions, update the constraint satisfaction problem, or add / delete predicates according to the change contents of the predicate, update the planning problem,
Send the updated constraint satisfaction problem to the constraint solver to get the solution, or send the updated planning problem to the planner to get the solution,
A value that can be input / changed in the input / change column is determined from the updated solution to the constraint satisfaction problem, or an operation that can be input / changed through the input operation screen from the updated planning problem solution. To present / limit input values and input operations in interactive processing. In the method for presenting and restricting input values and input operations in the interactive processing according to claim 1,
The target state of the planning problem is that the plurality of types of data items are operation targets for processing, and a state in which essential items are input among the plurality of types of data items is a specific state for processing. To present / limit input values and input operations in interactive processing. In the method for presenting and restricting input values and input operations in the interactive processing according to claim 1,
The constraint solver includes, as values that can be input / changed for one data item, a value domain range corresponding to another data item from a variable value domain corresponding to the one data item, and so on. A method for presenting and restricting input values and input operations in interactive processing, characterized by extracting values that do not cause inconsistencies with the values of variables corresponding to other input data items. In the method for presenting and restricting input values and input operations in the interactive processing according to claim 3,
A first algorithm that calculates all solutions of the constraint satisfaction problem by a constraint solver and generates a set whose elements are values taken by the variable in each solution, and for each value in the value domain of the variable, A new constraint satisfaction problem in which the variable takes the value is generated, a value that can be determined to have a solution by the constraint solver is extracted, and a second algorithm that removes the value that is not prepared is prepared.
From the quantitative features of the input operation screen and the operation target data item, an index of the amount of calculation of presentation / restriction of the input value on the input operation screen is calculated, and the calculated value is compared with a reference value designated in advance. A method of presenting and restricting input values and input operations in interactive processing, wherein one of the first and second algorithms, which is expected to have a low calculation amount, is selected and executed. In the method for presenting and restricting input values and input operations in the interactive processing according to claim 1,
The planner executes the operation from among the input operations that can be input and changed through the input operation screen at that time and the precondition is satisfied in the current state and can be executed, and the state is updated according to the post-condition. And a method of presenting and restricting input values and input operations in interactive processing, characterized in that an operation having an execution order that reaches a target state is extracted. For a predetermined application program operating on an arbitrary computer, at least one input / change column for data items is included in the display device of the arbitrary computer or another computer connected to the arbitrary computer via a network. An input operation screen is displayed, and the value of the data item in the input / change column is input based on an operator's operation from an input device of the arbitrary computer or another computer connected to the arbitrary computer via a network.・ Change and input the value of two or more types of data items interactively by switching this screen for two or more input operation screens including input / change fields corresponding to different types of data items. The default consistency between the values of each data item Satisfying way, the values entered or changed to said input-change box, a device for presenting or restrict the execution order of the operation of inputting and changing through said input operation screen,
A constraint solver and a planner operating on the arbitrary computer or another computer connected to the arbitrary computer via a network;
A constraint satisfaction problem comprising a variable, a value domain of the variable, and a constraint condition, wherein the values of the plurality of types of data items are variables, and the default consistency to be satisfied by the values of the plurality of types of data items is a constraint condition; , Consisting of a set of operations including a target state, a current state, a precondition and a postcondition, and a set consisting of a set of a processing operation target and a specific state of processing for the operation target as a target state, A planning problem that represents the current state, preconditions, and postconditions by a predicate consisting of a main part that identifies the operation target for processing and a predicate that indicates the state, and the type and value of the data item that has been input or changed The arbitrary computer storing the change contents of the constraint condition in the corresponding constraint satisfaction problem and the relation table describing the change contents of the predicate in the planning problem or the arbitrary computer Along with storage unit of the other connected computers over a network to the computer,
The following means provided on the arbitrary computer or another computer connected to the arbitrary computer via a network:
Means for converting the type and value of the data item input / changed from the input device into the change content of the constraint condition in the constraint satisfaction problem with reference to the relation table, or the change content of the predicate in the planning problem;
A means for adding / deleting a constraint condition according to a change content of a constraint condition, updating the constraint satisfaction problem, or adding / deleting a predicate according to a change content of a predicate, and updating the planning problem;
Means for sending an updated constraint satisfaction problem to a constraint solver to obtain its solution, or sending an updated planning problem to a planner to obtain its solution;
A value that can be input / changed in the input / change column from the updated solution to the constraint satisfaction problem, or a means for determining an operation that can be input / changed through the input operation screen from the updated solution to the planning problem An apparatus for presenting and limiting input values and input operations in interactive processing characterized by the above. In the apparatus for presenting / limiting input values and input operations in the interactive processing according to claim 6,
The target state of the planning problem is that the plurality of types of data items are operation targets for processing, and a state in which essential items are input among the plurality of types of data items is a specific state for processing. Device for presenting and restricting input values and input operations in interactive processing. In the apparatus for presenting / limiting input values and input operations in the interactive processing according to claim 6,
The constraint solver includes, as values that can be input / changed for one data item, a value domain range corresponding to another data item from a variable value domain corresponding to the one data item, and so on. A device for presenting and limiting input values and input operations in interactive processing, characterized by extracting values that do not cause inconsistencies with the values of variables corresponding to other input data items. An apparatus for presenting / limiting input values and input operations in interactive processing according to claim 8,
A first algorithm that calculates all solutions of the constraint satisfaction problem by a constraint solver and generates a set whose elements are values taken by the variable in each solution, and for each value in the value domain of the variable, A new constraint satisfaction problem in which the variable takes the value is generated, a value that can be determined to have a solution by the constraint solver is extracted, and a second algorithm that removes the value that is not prepared is prepared.
From the quantitative features of the input operation screen and the operation target data item, an index of the amount of calculation of presentation / restriction of the input value on the input operation screen is calculated, and the calculated value is compared with a reference value designated in advance. An apparatus for presenting and limiting input values and input operations in interactive processing, wherein one of the first and second algorithms that is expected to have a low calculation amount is selected and executed. In the apparatus for presenting / limiting input values and input operations in the interactive processing according to claim 6,
The planner executes the operation from among the input operations that can be input and changed through the input operation screen at that time and the precondition is satisfied in the current state and can be executed, and the state is updated according to the post-condition. After that, an input value and input operation presentation / restriction device in interactive processing characterized by extracting an operation having an execution order to reach a target state.

Images