JP2001005678A - Device and method for network type information processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make processible information in proper order and in the shortest operation time without having an independent phase wherein sorting is performed by controlling whether or not a node requires recalculation as a result of the addition and deletion of a mark. SOLUTION: An input management part 3 transmits inputted information to a network structure 2 and when a variation of the input is detected, a mark addition and deletion part 4 gives a mark to a node whose output is not determined. A scheduling part 6 searches for and schedules nodes having only outputs from nodes having unmarked inputs among marked nodes and the input to a node is compared with the last input; when there is a change, the node is evaluated again to calculate a new output and the mark addition and deletion part 4 deletes the mark. The arithmetic result when all the nodes are unmarked is outputted by an output management part 7.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a network type information processing apparatus and method in which components having input / output are connected and arranged in a network structure.

[0002]

2. Description of the Related Art Conventionally, in constructing an information processing apparatus, when a large-scale and complicated process is targeted,
A method of dividing the entire process into a plurality of unit components having a small scale and a relatively simple process and connecting them to form a network structure to construct the information processing apparatus is often used. Such a method is a means of solving the problem,
Alternatively, this is one of the methods that have attracted attention from the viewpoint of measuring the reuse of processing unit components.

In such a network-type information processing apparatus, since a data flow always exists, simply connecting unit components does not operate normally. Therefore, it is necessary to perform processing for each unit component in an appropriate order in accordance with the flow of data.

Conventionally, a method of sorting by paying attention to the dependency between unit parts has been often used for controlling the processing order. For example, there is a method in which topological sorting is performed on unit components based on the network dependency, and calculations are performed in the sorted order.

[0005]

However, according to the above-described method using sorting, when the network structure itself changes due to addition of unit parts or change in the connection relationship, etc.
It is necessary to sort the whole again. Especially when the network structure is huge, the time required for such re-sorting is not negligible. Therefore,
Actually, it has a problem that it is difficult to dynamically change the network structure when executing the operation.

The present invention has been made to solve the above problems.
An object of the present invention is to provide a network-type information processing apparatus and method capable of performing processing in an appropriate order and with a minimum operation time without having an independent phase of performing sorting.

[0007]

In order to achieve the above object, a network type information processing apparatus according to the present invention is a network type information processing apparatus comprising a processing engine and network structure information. However, an input management unit that transmits an external input to the network structure information, and a mark indicating that an output result is uncertain for a node whose input has changed among nodes included in the network structure information or Add mark to delete
A deletion unit, a scheduling unit for scheduling recalculation of the node, and an output management unit for outputting a calculation result of the network structure information, and a mark addition / deletion unit when an input from the input management unit changes. A mark is added to a node having a dependency with an input in the above, and among the marked nodes, scheduling is performed from a node whose input is only an output from a node in a state where the mark is not added, and recalculation is performed. When the recalculation is completed, the mark is deleted from the node to which the mark is added, and the recalculation is repeated until the node to which the mark is added does not exist in the network structure information.

With this configuration, it is possible to control whether or not a node needs recalculation by adding and deleting a mark. If only nodes that need to be recalculated are recalculated, an appropriate order and minimum operations can be performed. Processing can be performed in time.

Further, the network type information processing apparatus according to the present invention saves the previous input and the calculation result for each node, and re-calculates when the input to the marked node matches the previous one. It is preferable that the previous calculation result be output without performing the calculation. If the input to the node matches the previous input, the output from the node does not change, so avoiding recalculation can prevent waste of computer resources. This is because the calculation processing time can be shortened.

[0010] In the network information processing apparatus according to the present invention, it is preferable that two or more nodes included in the network structure information be treated as one group, and the group be treated as a node. This is because, by recognizing a plurality of nodes as one node, the number of links to be considered can be reduced, and the calculation processing time can be reduced. In addition, the combination of nodes makes it easy to generate a node having a new function, and can promote the implementation of program components.

The network-type information processing apparatus according to the present invention further includes a structure management unit for adding / deleting nodes and links between the nodes included in the network structure information. Preferably, the deletion is performed dynamically at runtime. This is because, in an application that continuously outputs the calculation result, it is not preferable that the output is stopped or discontinuous.

Further, the present invention is characterized by software for executing the functions of the above-mentioned network type information processing apparatus as processing steps of a computer.
Specifically, in the processing engine, a step of transmitting an external input to the network structure information, and an output result is uncertain for a node whose input has changed among the nodes included in the network structure information. A step of adding or deleting a mark indicating, and a step of scheduling recalculation of the node, including a step of outputting the calculation result of the network structure information, when the input has changed,
A mark is added to a node having a dependency relationship with an input, and scheduling is performed from a node having an input that is only an output from a node in a state where the mark is not added, and recalculation is performed. When the calculation is completed, the mark is deleted from the node to which the mark is added, and the recalculation is repeated until the node with the mark no longer exists in the network structure information. It is a computer-readable recording medium recorded as a computer.

With this configuration, by loading and executing the program on the computer, it is possible to control whether or not the node needs to be recalculated by adding or deleting a mark. If recalculation is performed, it is possible to realize a network-type information processing apparatus that can execute processing in an appropriate order and with a minimum operation time.

[0014]

(Embodiment 1) Hereinafter, a network type information processing apparatus according to Embodiment 1 of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of the network-type information processing apparatus according to the first embodiment of the present invention. In FIG. 1, 1 indicates a processing engine, and 2 indicates network structure information.

The processing engine 1 includes an input management unit 3 for transmitting input information to network structure information, a mark addition / deletion unit 4 for adding or deleting a mark to a node included in the network structure information 2, a node and a link. Generation of
The structure includes a structure management unit 5 for performing deletion and the like, a scheduling unit 6 for correctly scheduling the order of node recalculation, and an output management unit 7 for acquiring and outputting an output from the network structure information 2.

The network structure information 2 is composed of nodes having an arbitrary number of inputs and outputs and links having directionality indicating their dependencies. For example, the link 17 depends on the output from the node 8, and when the output of the node 8 changes, it is necessary to recalculate the node 11, and the node 8 outputs the output before the node 11. It means that you have to decide.

First, the input management unit 3 transmits the input information to the network structure information 2. Here, when the input is constant, the output of the network structure information 2 is also constant. For example, when the device according to the first embodiment is used for determining the action of the character displayed on the screen, the action selected by the character does not change unless the external environment changes. Therefore, the input management unit 3 can also control whether or not to transmit an input relating to the network structure information 2. By doing this,
Unnecessary operations can be avoided, and waste of computer resources can be prevented.

Next, when a change in the input is detected, the mark addition / deletion unit 4 is activated. Mark addition
The deletion unit 4 adds a mark to a node whose output is undecided, that is, a node that may need to be recalculated due to a changed input. In FIG.
FIG. 7 shows a mark situation diagram of network structure information when input 16 changes. Note that the mark addition / deletion unit 4 traces the dependent nodes from the changed input and assigns marks to the nodes having the dependency. In consideration of the assignment efficiency in this case, It is preferable to add marks according to a depth-first search as shown in FIG.

For example, in FIG.
, A mark is first added to the node 9. Next, since the node 9 has a dependency relationship with the node 10 via the link 18, the node 10 is also marked.

Here, the node 9 has a dependency with the node 12 via the link 19, but at the same time, the node 12 also has a dependency with the node 10 through the link 22. In the present invention, nodes are given a mark while performing a search with priority given to a node having a shallow link hierarchy in the network. Therefore, it is determined that the node 12 is a node having a deeper link hierarchy than the node 10, and 12
Is not a target for marking at this time.

When the mark is given to the node 10, the mark is given to the node 13 via the link 21. The output 24 of the node 13 is the network structure information 2
, So we cannot search deeper. Therefore, returning to the previous position by one link, a mark is added to the node 12 from the node 10 via the link 22.

In the following, a mark is given to the node 14 in the same manner, so that a mark is given to the hatched portion in FIG. Note that the node 12 can be marked by passing the link 19 from the node 9. However, since the mark has already been added at this point, there is no need to consider the mark state further ahead. .

Next, the structure management section 5 newly adds or deletes a node in the network structure information 2. When a new link is constructed by adding a node, the existing node and link are not affected at all, so that a new criterion is added, that is, a so-called learning condition is added. New nodes can be dynamically added. That is, there is no need to stop the program itself when adding a node, and a new node can be added even during execution.

Further, in the scheduling unit 6,
Among nodes to which the mark is added, a node whose input is only an output from a node to which the mark is not added is searched for and scheduled. For example, in FIG. 2, the node in such a state is limited to the node 9.

The recalculation will be performed from the scheduled node. Here, the following processing is performed for the case where the scheduled input to the node has not changed compared to the previous input and the case where the input has changed.

First, when the input has not changed, the total calculation amount can be reduced by reusing the previous calculation result, which leads to prevention of waste of computer resources.

On the other hand, if the input has changed, a new output is calculated by re-evaluating the node. At the same time as the recalculation, the mark of the recalculated node is released by the mark addition / deletion unit 4. In FIG.
When the recalculation of the node 9 is completed, the mark of the node 9 is released.

Recalculation is repeated for all nodes until it is confirmed that the mark has been released. Then, the calculation result at the time when the mark is completely released is output by the output management unit.

FIGS. 3 to 5 show the progress of the processing relating to the mark described above. That is, FIG. 3 shows a state where the node 9 is scheduled and the mark is released. Here, when searching for the next schedulable node, the nodes 10 and 12 can be considered. However, whichever is scheduled first does not affect the final result of the calculation.

FIG. 4 shows a state in which the node 10 is first scheduled. In FIG. 4, the next schedulable nodes are node 12 and node 13,
FIG. 5 shows a state in which these nodes are scheduled in parallel or in order.

In FIG. 5, since the last remaining node is only the node 14, by scheduling the node 14 at the end, the marks are removed from all the nodes.

In the first embodiment, a method of comparing with a previous calculation result at the time of scheduling a node can be used to determine whether or not the input to the node has changed. However, the method is not limited to such a method, and any method may be used as long as it has a similar effect. For example, a method may be used in which when the operation result output of a node changes, a flag is set for another node referred to. For example, in FIG. 2, when the operation output value changes as a result of re-evaluating the node 9,
A method of setting a flag indicating that the input has changed in the nodes 10 and 12 is also conceivable.

When the input does not change,
Since the node output does not change, recalculation is unnecessary. However, even if the input changes, the node output may not change. For example, considering a simple node as shown in FIG. 6, even if the input value to this node changes from 1 to 100, the output value is always "TRUE". Therefore, even if the input value changes, the output value may or may not change. Therefore, the necessity of recalculation may be determined based only on the output value.

Next, an example of the network type information processing apparatus according to the first embodiment will be described with reference to the drawings. FIG. 7 is a configuration diagram of a network type information processing apparatus according to one embodiment of the present invention. FIG. 7 shows an application of the network-type information processing apparatus according to the first embodiment to an emotion / action selection apparatus for a virtual creature. In FIG. 7, 27 is an image display device, 28 is a processing engine, 29 is network structure information, 30 is an input management device, 31 is a mark addition / deletion device, 32 is a structure management device, and 33 is a scheduler. Denotes an output management device.

First, an image display device 27 is a device for actually expressing an action selected by a virtual creature in an image.
Computer displays and LCDs are often used. The image display device 27 displays the action selected by the virtual creature and, at the same time, calculates the display result, and outputs, for example, the brightness, temperature, wind direction, and the like, which are information on the position of the virtual creature and the world where the virtual creature exists. It is possible to do.

Further, the processing engine 28 transmits the external state from the image display device 27 to the network structure information 29 via the external information node 35. Then, the internal state is calculated in the network structure information 29. For example, by changing the output of the external brightness node 36 or changing the output of the activity node 37 of the virtual creature, the emotion / behavior of the virtual creature is selected.

In this embodiment, for example, the drowsiness node 38
The behavior determination model shows that the behavior becomes larger when the outside is dark and the activity of the virtual creature is low. Similarly, the hunger node 39 also realizes an emotion determination model in which the hunger node 39 increases as the state of high activity continues. The action selection node 40 comprehensively judges these pieces of information, assigns priorities, and outputs the action to be performed next.

FIG. 8 is a diagram in which the structure management device 32 adds a happy node 41 as a new node to the network structure information 29 in the network type information processing device shown in FIG. Here, the input from the drowsiness node 38 and the hunger node 39 and the output to the action selection node 40 are newly set, and have no influence on the dependency of the existing nodes. Accordingly, it can be seen that a new criterion for selecting the action of "happiness" has been added.

Further, in FIG. 9, a destination selection node 44 is added and linked to the action selection node 40. In this case, when the input distance is shorter than a certain distance, the position is set as the destination. As an input to the destination selection node 44, for example, the structure management device 32 switches to a distance node 42 to the bait, a distance node 43 to the nest, and the like, so that the destination can be changed according to the situation. Becomes

Of course, depending on the switching of the structure management device 32, only the criterion at the destination selection node 44 changes, so that there is no influence on the dependency relation of the existing nodes. Therefore, switching of the structure management device 32 can be performed dynamically without stopping the program, even during execution of the program.

As described above, according to the first embodiment, the updating of the network structure information can be performed without going through an independent process of sorting the nodes, and the dependency between the nodes can be freely changed. It is possible to do. Further, even when a change occurs in the network structure due to addition / deletion of a node or the like, processing can be continuously performed without interrupting processing.

(Embodiment 2) Next, Embodiment 2 of the present invention will be described with reference to the drawings.

FIG. 10 is a view showing an example of the network structure information 2 in the network type information processing apparatus according to the second embodiment of the present invention. In FIG. 10, a part of the network structure information 2 is grouped, and the group 26 is treated as one node.

More specifically, the node 10, the node 11, and the node 12 are treated as one group 26, and the group 26 is treated as one node having inputs 17 and 18 and outputs 20 and 23. In this case, for example, when the input 16 changes, the node 9, the node 10, the node 12, the node 13, and the node 14 are set in a state where no grouping is performed.
Marks were added to the five nodes.

When these are grouped and the inside is concealed as one node as shown in FIG.
If 6 changes, marks will be added to four nodes, node 9, node 26, node 13, and node 14, and the number of marks will be reduced by simplifying the node structure.

If there is no change in the output of the node 9, there is no need to perform any scheduling inside the node 26. Therefore, by grouping, it is possible to enjoy the advantage that the total amount of calculation can be reduced.

As described above, according to the second embodiment, by recognizing a plurality of nodes as one node, it is possible to reduce the number of links to be considered and to shorten the calculation processing time. be able to. In addition, the combination of nodes makes it easy to generate a node having a new function, and can be expected to promote the implementation of program components.

Next, the flow of processing of a program for realizing the network-type information processing apparatus according to the first embodiment of the present invention will be described. FIG. 12 shows Embodiment 1 of the present invention.
Is a flowchart of processing of a program for realizing a network-type information processing apparatus according to the first embodiment.

First, when a new input is detected in the input management unit (step S121), according to the input,
The mark addition / deletion unit adds a mark to a node that is dependent on the input node (step S12).
2).

Next, the scheduling unit searches the marked nodes for a node whose input is only an output from the unmarked node, and schedules the searched node (step S12).
3). This is because it is necessary to perform the processing of such a node first. Immediately after the input, the input node is scheduled.

Then, it is determined whether or not the contents of the scheduled input to the node are different from the previous input to the node (step S).
124). If there is no change from the previous input content, the output value of the current output is the same as the previous output value (step S124: No), so that the recalculation is not performed by using the previous calculation result as the output value. (Step S12
5). By doing so, it is possible to shorten the calculation time by omitting the calculation time of the shelves. On the other hand, if there is a change from the previous input contents (step S12
4: Yes), recalculation is performed based on the contents of the input, and a new output value is calculated (step S1).
26).

For the node for which input / output has been performed, the mark added by the mark adding / deleting unit is released (step S127). This is to indicate that the arithmetic processing for the node has been completed. Then, the above processing is repeated until the marks are released for all the nodes in the network structure information (step S128: N
o), if the marks have been released for all nodes (step S128: Yes), the output management unit outputs the calculation result (step S129).

A recording medium storing a program for realizing the network type information processing apparatus according to the embodiment of the present invention is a CD-ROM as shown in the example of the recording medium shown in FIG.
Not only the portable recording medium 132 such as the ROM 132-1 and the floppy disk 132-2, but also any other storage device 132 provided at the end of the communication line, or the recording medium 134 such as a hard disk or a RAM of the computer 133 may be used. When the program is executed, the program is loaded and executed on the main memory.

Further, the recording medium on which the result of the previous arithmetic processing at each node generated by the network type information processing apparatus according to the embodiment of the present invention is recorded is also the same as the recording medium shown in FIG. , CD-ROM132-
In addition to a portable recording medium 132 such as a floppy disk 1 or a floppy disk 132-2, any other storage device 131 provided at the end of a communication line or a recording medium 134 such as a hard disk or a RAM of a computer 133 may be used. It is read by the computer 133 when using the network type information processing apparatus according to the present invention.

[0055]

As described above, according to the network-type information processing apparatus of the present invention, the updating of the network structure information can be performed without going through the independent process of sorting the nodes. Dependencies can be freely changed. Further, even when a change occurs in the network structure due to addition / deletion of a node or the like, processing can be continuously performed without interrupting processing. can do.

Further, according to the network type information processing apparatus according to the present invention, it is possible to control whether or not a node needs recalculation by adding and deleting a mark, and recalculate only the node which needs recalculation. In the right order,
In addition, it is possible to execute the processing with a minimum operation time.

Further, according to the network-type information processing apparatus of the present invention, the number of links to be considered can be reduced by recognizing a plurality of nodes as one node, and the calculation processing time can be reduced. Can be achieved. Further, the combination of nodes makes it easy to generate a node having a new function, and can be expected to promote the implementation of program components.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a network-type information processing apparatus according to a first embodiment of the present invention;

FIG. 2 is an explanatory diagram of network structure information in the network-type information processing device according to the first embodiment of the present invention;

FIG. 3 is an exemplary diagram of a mark process in the network-type information processing apparatus according to the first embodiment of the present invention;

FIG. 4 is an exemplary diagram of mark processing in the network-type information processing apparatus according to the first embodiment of the present invention;

FIG. 5 is an exemplary diagram of mark processing in the network-type information processing apparatus according to the first embodiment of the present invention;

FIG. 6 is an exemplary diagram of a node in the network-type information processing device according to the first embodiment of the present invention;

FIG. 7 is a configuration diagram of a network-type information processing apparatus according to an embodiment of the present invention.

FIG. 8 is a configuration diagram of a network-type information processing apparatus according to another embodiment of the present invention.

FIG. 9 is a configuration diagram of a network-type information processing apparatus according to another embodiment of the present invention.

FIG. 10 is a view showing an example of network structure information in a network-type information processing apparatus according to a second embodiment of the present invention;

FIG. 11 is a view showing an example of network structure information in a network type information processing apparatus according to a second embodiment of the present invention;

FIG. 12 is a flowchart of a process in the network-type information processing apparatus according to the first embodiment of the present invention;

FIG. 13 is an exemplary view of a recording medium.

[Explanation of symbols]

1, 28 processing engine 2, 29 network structure information 3 input management unit 4 mark addition / deletion unit 5 structure management unit 6 scheduling unit 7 output management unit 8, 9, 10, 11, 12, 13, 14, 35, 36,
37, 38, 39, 40, 41, 42, 43, 44 Node 15, 16 Input 17, 18, 19, 20, 21, 22, 23 Link 24, 25 Output 26 Group 27 Image display device 30 Input management device 31 Mark Addition / deletion device 32 structure management device 33 scheduler 34 output management device 131 destination storage device 132 portable recording medium such as CD-ROM or floppy disk 132-1 CD-ROM 132-2 floppy disk 133 computer 134 on computer Recording media such as RAM / hard disk

Claims (6)

[Claims] 1. A network-type information processing apparatus including a processing engine and network structure information, wherein the processing engine transmits an external input to the network structure information, and the network structure information. Among the nodes included in, a mark addition / deletion unit that adds or deletes a mark indicating that the output result is uncertain for the node whose input has changed, and a scheduling unit that performs scheduling of recalculation of the node An output management unit that outputs a calculation result of the network structure information, wherein when an input from the input management unit changes, the mark adding / deleting unit marks the node having a dependency with the input. Is added and the mark is not added to the input in the node to which the mark is added. Scheduling and recalculating from the node that is the only output from the node, and after the recalculation is completed, the mark is deleted from the node to which the mark is added, and the node with the mark is added to the network structure. A network-type information processing apparatus wherein recalculation is repeated until the information does not exist in the information. 2. A previous input and a calculation result are stored for each of the nodes, and when an input to the marked node matches the previous one, the previous calculation result is used without recalculation. 2. The network type information processing apparatus according to claim 1, wherein the information is output. 3. The network-type information processing apparatus according to claim 1, wherein two or more nodes included in the network structure information are treated as one group, and the group is treated as the node. 4. A structure management unit for adding / deleting the nodes and links between the nodes included in the network structure information, wherein the structure management unit dynamically adds / deletes each of the nodes at the time of execution. The network-type information processing apparatus according to claim 1, wherein 5. A network-type information processing method using a processing engine and network structure information, wherein the processing engine includes a step of transmitting an external input to the network structure information; Adding or deleting a mark indicating that the output result is uncertain for the node whose input has changed among the nodes to be input, scheduling the recalculation of the node, and calculating the network structure information Outputting a result, wherein when the input changes, a mark is added to the node having a dependency relationship with the input, and in the node to which the mark has been added, an input is marked. Scheduling and recalculating from the node that is the only output from the node that is not When the process is completed, the mark is deleted from the node to which the mark is added, and re-calculation is repeated until the node to which the mark is added does not exist in the network structure information. Method. 6. A computer-readable recording medium storing a program to be executed by a computer for realizing a network-type information processing method using a processing engine and network structure information, wherein the processing engine has an external input. Transmitting to the network structure information, and adding or deleting a mark indicating that an output result is uncertain for a node whose input has changed among the nodes included in the network structure information, Scheduling a recalculation of a node, and outputting a calculation result of the network structure information, and when the input changes, adding a mark to the node having a dependency with the input, If the input is marked, Scheduling and recalculation from the node that is only the output from the node in the non-added state, and when the recalculation is completed, the mark is deleted from the node to which the mark is added, and the mark is added. A computer-readable recording medium recording a program to be executed by a computer, wherein re-calculation is repeated until the node does not exist in the network structure information.