JP6511333B2 - Decision support system and decision support method - Google Patents

Description

  The present invention relates to a decision support system.

  A computer system has been proposed that uses the expert's knowledge to predict future situations and the actions of persons involved.

  For example, Patent Document 1 (Japanese Patent Application Laid-Open No. 5-204991) includes a step of registering a plurality of patterns in a system consisting of a computer, a time series database, a registration pattern database and a terminal device, time series data from the time series database From the steps of reading and collating with a plurality of patterns already registered for each pattern and for each fixed period, analyzing causality regarding appearance between registered patterns, and displaying analysis results A time series data retrieval system is described. The time-series data search system described in Patent Document 1 predicts future behavior (action on a rule basis) based on a result of comparison with registered patterns.

Unexamined-Japanese-Patent No. 5-204991

  However, in the rule-based prediction system as described in Patent Document 1, a chain model is used to create a model including many rules and simulate what will occur next. For this reason, it is difficult to create a model in a rule-based prediction system. That is, in this model, all events and the actions of persons involved must be considered, and the expert's knowledge is classified in order to organize the expert's knowledge and create a model that integrates the knowledge. , It is difficult to create a model. Therefore, there is a need for a system that models expert knowledge without difficulty and predicts the future.

The following is a representative example of the invention disclosed in the present application. That is, it is a decision support system configured by a computer having a processor and a memory, wherein the memory stores a plurality of index values which quantify a plurality of situations necessary for the decision, and the decision support system An action chain model for the processor to derive a next action from a player action, and a reaction model for the processor to derive a next index value from the player action and the index value. An intention model for the processor to derive, for each player, a selection probability representing an intention of the action from the index value; an evaluation model for the processor to derive an evaluation value for each player from the index value; , an action selector for selecting an action of the player, the processor, the election Arbitrates actions of each player is, the and a mediation model to determine the actions of each player, the decision support system, using the action chain model, the player from the action of the player Derive the next action that can be taken, derive the next index value from the action of the player and the index value using the reaction model, and derive the next index derived using the intention model The selection probability of the action of each player is calculated from the value, and using the evaluation model, the action selection unit calculates an evaluation value indicating the degree to which the derived next index value is desirable for each player, The action derived using the action chain model, and the selection probability calculated using the intention model , By using the calculated evaluation value by using the evaluation model, the select multiple actions that each player can take, and outputs the probability of multiple actions and the actions the selected, the decision support The system uses the arbitration model to arbitrate the action of each player selected by the action selector to determine the action of each player .

  According to the representative form of the present invention, expert's knowledge can be easily organized, and situations and actions that will occur in the future can be predicted based on the expert's knowledge. Problems, configurations, and effects other than those described above will be clarified by the description of the following embodiments.

It is a block diagram which shows the physical structure of the decision-making assistance system of a 1st Example. It is a block diagram showing the logical composition of the decision support system of a 1st example. It is a figure explaining the action chain model of a 1st Example. It is a figure explaining the reaction model of the 1st example. It is a figure explaining the intention model of a 1st Example. It is a figure explaining the evaluation model of a 1st example. It is a flowchart of a process by the decision support system of a 1st Example. It is a figure which shows the example of the simulation result output screen of a 1st Example. It is a flowchart of a process by the decision support system of the modification of a 1st Example. It is a figure which shows the example of the star | collar table output screen of a 1st Example. It is a figure explaining the Monte Carlo tree search for comprising the star | bill of 1st Example. It is a block diagram showing the logical composition of the decision support system of the 2nd example. It is a figure explaining the mediation model of a 2nd example.

  FIG. 1 is a block diagram showing the physical configuration of the decision support system of the first embodiment.

  The decision support system of this embodiment has a plurality of computers (CALC_NODE) and a communication switch (COM_SW) that connects the plurality of computers.

  Each computer (CALC_NODE) is a communication device (CPU) that executes a program, a temporary storage device (RAM) and an auxiliary storage device (STOR) that stores data and programs, and a communication device (COM_SW) connected to a communication device. And (COM_DEV). The processor (CPU), the temporary storage device (RAM), the auxiliary storage device (STOR), and the communication device (COM_DEV) are connected by a bus (BUS).

  A processor (CPU) executes a program stored in a temporary storage device (RAM). The temporary storage device (RAM) includes a ROM, which is a non-volatile storage element, and a RAM, which is a volatile storage element. The ROM stores an immutable program (for example, BIOS). (RAM) is a high-speed volatile memory element such as D (RAM) (Dynamic Random Access Memory), and temporarily stores a program executed by a processor (CPU) and data used when the program is executed Do.

  The auxiliary storage device (STOR) is, for example, a large-capacity and non-volatile storage device such as a magnetic storage device (HDD) or a flash memory (SSD), and is used when executing programs and programs executed by a processor (CPU). Store data. That is, the program is read from the auxiliary storage device (STOR), loaded into the storage device (RAM), and executed by the processor (CPU).

  The communication device (COM_DEV) is a network interface device that controls communication with another device via the communication switch (COM_SW) according to a predetermined protocol.

  Each computer (CALC_NODE) may have an input interface and an output interface. The input interface is an interface for receiving an input from an operator, and specifically, a mouse, a keyboard, a touch panel, a microphone, and the like. The output interface is an interface that outputs the execution result of the program in a form that can be viewed by an operator, and is a display device, a printer, or the like.

  A terminal computer having an input interface and an output interface may be connected to the communication switch (COM_SW).

  The program executed by the processor (CPU) is provided to each computer (CALC_NODE) via removable media (CD-ROM, flash memory, etc.) or via a network, and is a non-transitory storage medium (STOR non-volatile storage) (STOR) Stored in). Therefore, each computer (CALC_NODE) may have an interface for reading data from removable media.

  Each computer (CALC_NODE) is a computer system that is configured physically on one computer or on a plurality of logically or physically configured computers, and operates in separate threads on the same computer It may operate on a virtual computer built on multiple physical computer resources. Also, each functional unit of each computer (CALC_NODE) may be realized on a different computer.

  FIG. 2 is a block diagram showing the logical configuration of the decision support system of the first embodiment.

  The decision support system of the present embodiment is composed of four models: action chain model 1, reaction model 2, intention model 4 and evaluation model 5. Specifically, the decision support system shown in FIG. 2 includes an action chain model 1, a reaction model 2, and a plurality of action determination units 3. The action determination unit 3 includes an intention model 4, an evaluation model 5, and an action selection unit 6, and is provided for each player.

  The action chaining model 1 is a rule-based simulator, as shown in FIG. 3, and derives a probable action to be executed next from the current action of the player. A player is an entity that decides and acts (performs actions) in the world, and is, for example, the people, government agencies (ministries and ministries), parliaments, cabinets, foreign governments, mass communication, and the like.

  The reaction model 2 is a rule-based simulator as shown in FIG. 4 and derives the next index value from the current action of each player and the current index value. The index value is, for example, an index that quantifies events (changes in the situation) occurring in the world, and is an economic index (GDP, stock price, exchange rate, etc.), opinion poll results (Cabinet approval rate, etc.).

  The action determination unit 3 is provided for each player and derives the next action of each player.

  The intention model 4 is a simulator for deriving an action intention from the following index values, as shown in FIG. The action intention is a numerical value representing the strength of the intention to execute the action adopted in a certain situation (represented by a combination of index values). That is, each player has a high probability (expected value) of selecting an action whose action intention value is large. The evaluation model 5 is a simulator for deriving an evaluation value from the following index values as shown in FIG. The evaluation value is a numerical value that represents the degree to which a certain situation is desirable for the player.

  The action selection unit 6 is a selector that derives the next action of the player from the probability of the next action of the player, the action intention, and the evaluation value. The action selection unit 6 weights the action output from the action chain model 1 with the action intention output from the intention model 4 and the evaluation value output from the evaluation model 5, for example, to perform the next action of the player. select.

  The next action derived by the action selection unit 6 is input to the action chaining model 1 and used for simulation of the next action. Further, the next index value derived by the reaction model 2 is input to the reaction model 2 and used for simulation of the next index value.

  FIG. 3 is a diagram for explaining the action chaining model 1 of the first embodiment. The action chain model 1 is represented by a Markov decision process model in which each player's action is a node. Each node is associated with a probability that the player selects an action, and an edge between the nodes is associated with a probability that a state transitions between nodes.

In the action chain model 1 shown in FIG. 3, the probability that the player 1 selects the action 1, the action 2 and the action 3 is 0.3: 0.5: 0.2, respectively. Also, if player 1 selects action 1, player 2 selects action 2 with a probability of 0.8. That is, the probability that player 2 selects action 2 can be expressed by equation 1.
1-((1-0.3) x (1-0.8)) ... (Formula 1)

  The action chaining model 1 can derive one or more next possible actions from the player's current action.

  FIG. 4 is a diagram for explaining the reaction model 2 of the first embodiment. The reaction model 2 is a model that represents the correlation between each index by a causal loop diagram.

  For example, as shown in FIG. 4, the reaction model 2 can be represented by a graphical model in which a plurality of indices are nodes and nodes are connected by edges. The solid arrows at each edge indicate positive correlation, and the dashed arrows indicate negative correlation. Furthermore, by defining a coefficient for each edge, a system dynamics model representing the behavior of each index can be obtained. The coefficient of each edge is defined as a ratio to the increase or decrease of the index. For example, if the index 3 and the index 2 have a positive correlation, and the edge coefficient is 0.5, when the index 3 is increased by 1, the index 2 is increased by 0.5. Further, the index 3 and the index 5 have a negative correlation, and if the coefficient of the edge is 1.2, the index 5 decreases by 1.2 when the index 3 increases by 1.

  In the case of performing discrete system simulation, the reaction model 2 can be represented by a recurrence equation in the computer. Further, when performing continuous system simulation, the reaction model 2 can be represented by a first order differential equation in a computer.

  There are two types of indicators: stock elements and flow elements. The stock element indicates an amount at a certain point of time, for example, a stock amount of crude oil. In addition, the flow element indicates, for example, the flow of variables in a time zone, such as the import amount (production amount) and consumption amount of crude oil. Whether an index is a stock element or a flow element may be determined according to whether the index is generally used as a stock quantity or a flow quantity in the world. In addition, an amount that can not be actually measured numerically (for example, risk impact, nationalism) may be used as an index. As described above, by composing the reaction model 2 by mixing the stock element and the flow element, it is possible to model the thinking of the expert as it is without setting any constraints.

  FIG. 5 is a diagram for explaining the intention model 4 of the first embodiment. The intention model 4 is a model that represents the correlation between each index by a causal loop diagram, and further represents the correlation between each player's action and each index.

  The intention model 4 can be represented by the same model as the reaction model 2 described above. That is, the intention model 4 can be represented, for example, by using a plurality of indices as nodes and a graphical model in which nodes are connected by edges. The solid arrows at each edge indicate positive correlation, and the dashed arrows indicate negative correlation. Furthermore, by defining a coefficient for each edge, a system dynamics model representing the behavior of each index can be obtained. The coefficient of each edge is defined as a ratio to the increase or decrease of the index.

  In the intention model 4, the indicators have a correlation with each other, but between the player's action and the indicator, only the edge from the action of each player to the indicator is defined, and the edge in the opposite direction is not defined. Also, the edge between each action is not defined. The intention model 4 can model the influence of each player's action on the indicator.

  In the case of discrete system simulation, the intention model 4 can be represented by a recurrence equation in the computer. In the case of continuous system simulation, the intention model 4 can be represented by a first-order differential equation in the computer. Similar to the response model 2, in the intention model 4, the indicators of the intention model 4 include a stock element and a flow element.

  FIG. 6 is a diagram for explaining an evaluation model 5 of the first embodiment. The evaluation model 5 represents the correlation between each index by the causal loop diagram, the correlation between each player's action and each index, the correlation between each index and each player's evaluation, and It is a model that represents the correlation between each indicator and the intention of each player's action. An evaluation is a numerical value which shows whether each player thinks the situation represented by the combination of a plurality of indicators as desirable. The evaluation may have a different evaluation value for each combination of the numerical range of the index. The action intention is represented by a set of actions that each player can take and a probability of selecting each action.

  The evaluation model 5 can be represented by a model similar to the reaction model 2 described above. That is, the evaluation model 5 can be represented by, for example, a graphical model in which a plurality of indices are nodes and nodes are connected by edges. The solid arrows at each edge indicate positive correlation, and the dashed arrows indicate negative correlation. Furthermore, by defining a coefficient for each edge, a system dynamics model representing the behavior of each index can be obtained. The coefficient of each edge is defined as a ratio to the increase or decrease of the index.

  In the evaluation model 5, the indices have a correlation with each other, but between the player's action and the index, only the edge from each player's action to the index is defined, and the opposite direction edge is not defined. Also, the edge between each action is not defined. Also, between each index and each player's evaluation, only the edge from each index to the evaluation is defined, and no reverse edge is defined. Also, between each index and each player's action intention, only the edge from each index to the action intention is defined, and the reverse direction edge is not defined. Furthermore, the edge between the intentions of the action is not defined. The evaluation model 5 can model the influence of the strength of the intention of each player's action of each index, and can set the evaluation value of each player's action.

  In the case of discrete system simulation, the intention model 4 can be represented by a recurrence equation in the computer. In the case of continuous system simulation, the intention model 4 can be represented by a first-order differential equation in the computer. In the evaluation model 5, as with the reaction model 2, the indices of the evaluation model 5 include a stock element and a flow element.

  Evaluation model 5 includes intention model 4, and intention model 4 includes reaction model 2. Thus, the reaction model 2, the intention model 4 and the evaluation model 5 may be configured by logically dividing one model.

  In the reaction model 2, the intention model 4 and the evaluation model 5, since the events represented by each node are somewhat related, an edge can be defined between almost all nodes. However, since defining the edges between all nodes complicates the model, it is preferable to construct the model by highly correlated edges (eg, edges whose coefficients are larger than a predetermined threshold).

  FIG. 7 is a flowchart of processing by the decision support system of the first embodiment.

  First, when the current situation and simulation period are input to the input interface (S101), the simulation start time is set to the initial value of the control parameter t repeatedly, and the simulation end time t_end is set. The current state to be input includes the current action of each player and each current index value.

  Next, it is determined whether t is smaller than t_end (S102). If t is smaller than t_end, the process proceeds to steps S103 and S105. If t is equal to or greater than t_end, the simulation result of the designated period is obtained, so the processing is ended and the simulation result output screen (FIG. 8) is output.

  In step S103, the action chaining model 1 is driven to derive the next action likely from the player's current action, and the next action is stored in the auxiliary storage device (STOR) (S104). In step S105, the reaction model 2 is driven, the next index value is derived from the current action of each player and the current index value, and is stored in the auxiliary storage device (STOR) (S106).

  The processes of steps S103 to S106 can be performed in parallel, but the process of activating the action chain model 1 (S103) and the process of driving the reaction model 2 (S105) may be sequentially performed.

  Next, the intention model 4 and the evaluation model 5 are driven for all players (S107, S109). The processes of steps S107 to S110 can be performed in parallel, but the process of activating the intention model 4 (S107) and the process of driving the evaluation model 5 (S109) may be sequentially performed.

  In step S107, the intention model 4 is driven to derive the action intention of each player from the next index value, and the action intention is stored in the auxiliary storage device (STOR) (S108). In step S109, the evaluation model 5 is driven, an evaluation value is derived from the next index value, and is stored in the auxiliary storage device (STOR) (S110).

  After that, the action selecting unit 6 determines the next action of the player in consideration of the probability of the next action of the player, the action intention, and the evaluation value, and stores the next action in the auxiliary storage device (STOR) (S111).

  After the next action of all players is determined, the next index value output from the reaction model 2 is set as the current index value, and the time of the reaction model 2 is advanced by one (S112). Then, 1 is added to the repetition control parameter t (S113), and the process returns to step S102. Note that 1 added to t indicates a time interval for executing the simulation, and the operator may set (for example, one day) in advance.

  By the above processing, the action of each player during the simulation period can be derived.

  FIG. 8 is a view showing an example of the simulation result output screen 1000 of the first embodiment. The simulation result output screen 1000 represents the action selected by each player as time passes, and is displayed on the output interface (display device). For example, as shown in FIG. 8, the action at each time of each player is displayed in the form of a table in which the players are listed vertically and the times of simulation results are listed horizontally. The simulation result output screen 1000 allows the user to know the actions taken by each player in time series.

  Next, a modification of the first embodiment will be described. In the variation described below, a star calendar is used to select each player's action.

  FIG. 9 is a flowchart of processing by the decision support system of the modification of the first embodiment.

  First, when the current situation and the simulation period are input to the input interface (S121), the initial value of the repetitive control parameter t is set at the start of the simulation, and the simulation end time t_end is set. The current state to be input includes the current action of each player and the current index value.

  Next, it is determined whether t is smaller than t_end. If t is smaller than t_end, the process proceeds to steps S103 and S105 (S122). If t is equal to or greater than t_end, the simulation result of the designated period is obtained, so the processing is ended and the simulation result output screen (FIG. 8) is output.

  Next, enumerate all combinations of action options that all players can take (S123), drive action chain model 1 and reaction model 2 to t + α for all the listed combinations, and follow each player's next The action and the next index value are derived and stored in the auxiliary storage device (STOR) (S124). α is a future time at which the prediction is considered at time t in calculating the score of each action described in the star catalog.

  Thereafter, for all players, the intention model 4 and the evaluation model 5 are driven, the action intention and the evaluation value of each player are derived, and stored in the auxiliary storage device (STOR) (S125).

  Thereafter, a star list is created based on the evaluation value at time t + α (S126), and the player's action at time t is determined using the star list and stored in the auxiliary storage device (STOR) (S127). Specifically, the star calendar is a table describing one's evaluation and the evaluations of other players, and is used to select an appropriate action as described later.

  After the next action of all players is determined, the next index value output from the reaction model 2 is set as the current index value, and the time of the reaction model 2 is advanced by one (S128). Then, 1 is added to the repetitive control parameter t (S129), and the process returns to step S122.

  By the above processing, it is possible to simulate each player's action during the simulation period while considering other people's actions.

  FIG. 10 is a diagram showing an example of the star calendar output screen 1100 according to the first embodiment. The star calendar output screen is displayed on the output interface (display device) by selecting the action column on the simulation result output screen 1000 (FIG. 8). Since the star calendar indicates the relationship between the actions of the two players, the operator selects the player of the other party after selecting the action column on the simulation result output screen 1000.

  In addition, it is good to comprise a star calendar with the Monte Carlo tree shown in FIG. 11 inside a computer.

  First, the contents of the screen will be described. The star calendar output screen 1100 shown in FIG. 10 shows the relationship between the two players, and includes a star calendar 1110 in which the actions of player 1 are listed vertically and the actions of player 2 are listed horizontally. At the bottom of the screen, a "return" button 1120 is provided. The operator can return to the simulation result output screen 1000 by operating the “Return” button 1120.

  Next, the contents of the star calendar will be described. In the star calendar shown in FIG. 10, in the set of the action of the player 1 and the action of the player 2, the evaluation of the player 1 and the evaluation of the player 2 are recorded as a set. The evaluation may be represented by symbols as shown in FIG. 10 or may be represented by numerical values. By using the star calendar, it is possible to determine an evaluated action that integrates the evaluation values of a plurality of players.

  Next, the case where you are the player 1 and the other party is the player 2 will be described as to the method of selecting an action using the star calendar. Using the MinMax method, determine the action so as to minimize the expected maximum damage.

  For example, when there are two players, focusing on each row of the star list (the row where their action is the same), the evaluation value (MinMax evaluation value) of Player 1 in the action with the best evaluation value of the opponent (Player 2) is Determine the best action as your player's next action. In the case illustrated, the action of player 1 is determined to be action 2.

  In addition, when there are three or more players, the actions of other players are sequentially fixed (as the action with the highest evaluation value of the player is selected) for the action of the player (player 1), and then the evaluation value of the player is one. Determine the highest action. The details will be described below.

  If there are three players, decide the action in the following steps.

  Step 1: Set your own action (player 1) to action 1.

In that state, consider a game of two players, Player 2 and Player 3.
Step 1-1: With player 2 as the player, the above-mentioned player decides the action of player 2 by two methods (MinMax according to the star calendar).
Step 1-2: With player 3 as his player, the above-mentioned player decides the action of player 3 by two methods (MinMax according to the star calendar).

  Evaluation of oneself (player 1) in the combination of each player's action decided as above (player 1 = action 1, player 2 = action decided at step 1-1, player 3 = action decided at step 1-2) The value is set to the evaluation value in action 1 of player 1.

  Step 2: Set your own (player 1) action to action 2.

  Also, the evaluation value in the action 2 of the player 1 is determined in the same manner as described above.

  Step 3: The calculation of step 1 is performed by the number of actions of one's own (player 1), and the evaluation value for one's own (player 1) determines the best action.

  Furthermore, if there are four players, the action is determined in the following steps.

Step 4: Fix your (Player 1) action. As a result, the players 2 to 4 become three games.
Step 4-1: In the game of three players, with player 2 as the player, the above player decides the action of player 2 by the method of three players.
Step 4-2: In 3 games, with player 3 as the player, the above-mentioned player decides the action of player 3 in 3 ways.
Step 4-3: In the three-player game, with the player 4 as the player, the above-mentioned player decides the action of the player 4 by the three-way method.

The evaluation value of player 1 (player 1) in the combination of each player's action decided as above (player 1 = action 1, player 2 = 4 actions determined in steps 4-1 to 4-3) Set to the evaluation value in action 1.
Step 5: The calculation of Step 4 is performed by the number of actions of one's own (player 1), and the evaluation value for one's own (player 1) determines the best action.

  FIG. 11 is a diagram for explaining a Monte Carlo tree search for constructing the star table of the first embodiment.

  That is, in the method described above with reference to FIG. 10, since the combination of all actions of all players is calculated, the amount of calculation is large. Therefore, by using Monte-Carlo Tree Search, it is possible to approximately execute the same process with a small amount of calculation. Note that if you calculate the Monte Carlo tree infinitely, you get the same result as if you calculated all the action combinations of all players.

  In the Monte Carlo tree search shown in FIG. 11, in the game by four players 1 to 4, when each player can select one of the actions 1 and 2, the evaluation value of his own (player 1) is calculated Show the process to

  First, oneself (player 1) selects action 1. Next, one of the players 2 to 4 is randomly selected (equal probability). Hereinafter, the case where the player 2 is selected will be described. The player 2 selects an action having a high average evaluation value for the player 2 at the child node.

  Next, one player 3 or 4 is randomly selected (equal probability). Hereinafter, the case where the player 3 is selected will be described. The player 3 selects an action having a high average evaluation value for the player 3 at the child node. Finally, the player 4 selects an action having a high average evaluation value for the player 4 at the child node.

  By the above processing, the set of actions of each player at time X + 0 is determined.

  Furthermore, in the case of prefetching, the following processing is performed.

  First, one of the players 1 to 4 is randomly selected (equal probability), and a child node selects an action having a high evaluation value for the selected player. Then, for the remaining players, as described above, randomly select players and determine an action. After expanding the tree to some extent, all players randomly select an action and advance the time by random playout. Then, the evaluation value of each player when the number of pre-read times set in advance is reached is calculated. Finally, each node of the tree that has passed through is traced backwards, the evaluation value at the time of prefetching is added to the evaluation value attached to the node, the average value is obtained, and the evaluation value is updated.

  After performing the above processing several hundreds times, the evaluation value of the player 1 attached to the root node becomes the evaluation value of the action 1 of the player 1.

  The above processing is performed for each action that can be taken by oneself (player 1), and the action with the highest evaluation value for oneself (player 1) is selected.

  Thus, the star calendar allows the user to know why the action was derived.

  As described above, according to the first embodiment of the present invention, since the decision support system is configured using the action chain model 1, the reaction model 2, the intention model 4 and the evaluation model 5, the expert knowledge Can be easily organized and modeled. For this reason, it is possible to predict situations or actions that will occur in the future based on expert's knowledge. In particular, since the intention model 4 and the evaluation model 5 are divided and modeled, the intention factor and the suppression factor can be separated, and the expert's knowledge can be taken into the model without processing.

Second Embodiment
Next, a second embodiment of the present invention will be described. The decision support system of the second embodiment comprises five models of an action chain model 1, a reaction model 2, an intention model 4, an evaluation model 5 and a mediation model 7. In the second embodiment, description of the same configuration and processing as those of the first embodiment described above is omitted, and different configurations and processing are described.

  FIG. 12 is a block diagram showing the logical configuration of the decision support system of the second embodiment.

  The decision support system of the second embodiment includes an action chain model 1, a reaction model 2, a plurality of action determination units 3, and an arbitration model 7. The action determination unit 3 includes an intention model 4, an evaluation model 5, and an action selection unit 6, and is provided for each player.

  The action chain model 1, the reaction model 2, the intention model 4, the evaluation model 5, and the action selection unit 6 are the same as those in the first embodiment described above. Note that the action determining unit 3 of the second embodiment outputs a plurality of actions that can be taken for each player, together with their selection rates.

  The arbitration model 7 mediates the actions output from the plurality of action determination units 3 to determine the action of each player. For example, there are contradictory actions that each player can take. The arbitration model 7 uses these relationships to select a combination of actions that can be performed simultaneously, and to determine each player's action.

  Specifically, the arbitration model 7 calculates the selection rates of the plurality of actions output from the action selection unit 6, selects the action with the highest selection rate for each player, and determines the action for each player.

  FIG. 13 is a diagram for explaining the arbitration model 7 of the second embodiment. The arbitration model 7 is a model that represents the correlation between each index (stock element, flow element, state element) by means of a causal loop diagram.

  For example, in the illustrated arbitration model 7, the stock elements 1, 2, 3 and 4 are x1, x2, x3 and x4, respectively, and the state elements 1 and 2 are y1 and y2, respectively. Define k1 to k9). The state element is a number representing the current state. For example, it can be defined as 1 when the player 1 is currently performing the action 1 and 0 when the player 1 is performing other than the action 1.

  As illustrated, the stock element and the state element determine the amount of flow that flows into the stock element and the amount of flow that flows out of the stock element, and the amount of the stock element changes in an explicit manner.

When defined as described above, in discrete system simulation, the value of each stock element at time t + 1 can be calculated by the following recurrence formula.
x1 (t + 1) = k1 × y1 (t) + k2 × x4 (t) −k3 × x3 (t)
x2 (t + 1) = x1 (t) + k3 × x3 (t) −k4 × x4 (t)
x3 (t + 1) = k5 x 4 (t)-k6 x x 1 (t)
x4 (t + 1) = k7 × x1 (t) −k8 × x3 (t) −k9 × y2 (t)

Also, in continuous system simulation, the value of each stock element can be calculated by the following differential equation.
d [x1 (t)] / dt = k1 * y1 (t) + k2 * x4 (t) -k3 * x3 (t)
d [x2 (t)] / dt = x1 (t) + k3 * x3 (t) -k4 * x4 (t)
d [x3 (t)] / dt = k5 x 4 (t)-k6 x x 1 (t)
d [x 4 (t)] / dt = k7 x x 1 (t)-k8 x 3 (t)-k9 x y 2 (t)

  In mediation model 7, a plurality of stock elements are associated and controlled by controlling the flow amount according to the stock element and the state element, the probability (selection rate) that each player selects an action is determined, and each player's action is It can be arbitrated.

  As described above, according to the second embodiment of the present invention, the arbitration model 7 mediates the action of each player selected by the action selecting unit 6 to determine the action of each player, so that the plurality of players can be determined. Models 1, 2, 4, 5 can be created without consideration of the action's mediation. That is, it is possible to create a model separately from action arbitration.

  The present invention is not limited to the embodiments described above, and includes various modifications and equivalent configurations within the scope of the appended claims. For example, the embodiments described above are described in detail to explain the present invention in an easy-to-understand manner, and the present invention is not necessarily limited to those having all the configurations described. Also, part of the configuration of one embodiment may be replaced with the configuration of another embodiment. In addition, the configuration of another embodiment may be added to the configuration of one embodiment. In addition, with respect to a part of the configuration of each embodiment, another configuration may be added, deleted, or replaced.

  In addition, each configuration, function, processing unit, processing means, etc. described above may be realized by hardware, for example, by designing part or all of them with an integrated circuit, etc., and the processor realizes the respective functions. It may be realized by software by interpreting and executing the program to

  Information such as a program, a table, and a file for realizing each function can be stored in a memory, a hard disk, a storage device such as a solid state drive (SSD), or a recording medium such as an IC card, an SD card, or a DVD.

  Further, control lines and information lines indicate what is considered to be necessary for explanation, and not all control lines and information lines necessary for mounting are shown. In practice, it can be considered that almost all configurations are mutually connected.

CALC_NODE Computer CPU Processor RAM Temporary Storage Device STOR Auxiliary Storage Device COM_SW Communication Switch COM_DEV Communication Device 1 Action Chaining Model 2 Reaction Model 3 Action Determination Unit 4 Intention Model 5 Evaluation Model 6 Action Selection Unit 7 Arbitration Model

Claims (6)

A decision support system comprising a computer having a processor and a memory, the system comprising:
The memory stores a plurality of index values that quantify a plurality of situations necessary for decision making,
The decision support system is
An action chain model for the processor to derive a next action from a player action;
A reaction model for the processor to derive the next index value from the player's action and the index value;
An intention model for the processor to derive, for each player, a selection probability representing the intention of the action from the index value;
An evaluation model for the processor to derive an evaluation value for each player from the index value;
An action selection unit for selecting an action of the player by the processor ;
The processor having an arbitration model for coordinating the actions of each of the selected players to determine the actions of each of the players ;
The decision support system is
The action chain model is used to derive the next action that the player can take from the action of the player,
The following index value is derived from the action of the player and the index value using the reaction model,
Using the intention model, calculate the selection probability of the action of each player from the derived next index value,
Using the evaluation model, calculate an evaluation value indicating the degree to which the derived next index value is desirable for each player,
The action selection unit uses the action derived using the action chain model, the selection probability calculated using the intention model, and the evaluation value calculated using the evaluation model. Selecting a plurality of actions that the player can take , and outputting the selected actions and the probability of the actions;
A decision support system characterized in that the decision support system mediates the action of each player selected by the action selection unit using the mediation model to determine the action of each player .   The decision support system according to claim 1, wherein
  The said action selection part outputs the screen data for displaying the action of each said selected player in time series, The decision support system characterized by the above-mentioned.   The decision support system according to claim 1, wherein
  The decision support system according to claim 1, wherein the action selection unit outputs screen data for displaying an evaluation of each of the plurality of players.   A decision support method executed by a computer having a processor and a memory, the method comprising:
  The memory stores a plurality of index values that quantify a plurality of situations necessary for decision making,
  The calculator generates an action chain model for deriving the next action from the player action, a reaction model for deriving the next index value from the player action and the index value, and an action intention from the index value. The action model of each player is determined by mediation of the action model of the player and the evaluation model for deriving the evaluation value for each player from the index value, and the action model of the player. Have an arbitration model for
  The method is
  The processor derives the next action that the player can take from the action of the player using the action chain model, and stores it in the memory.
  The processor derives the next index value from the action of the player and the index value using the reaction model, and stores the next index value in the memory,
  The processor uses the intention model to calculate the selection probability of the action of each player from the derived next index value, and stores it in the memory.
  The processor uses the evaluation model to calculate an evaluation value representing the degree to which the derived next index value is desirable for each player, and stores the evaluation value in the memory.
  The processor uses the action derived using the action chain model, the selection probability calculated using the intention model, and the evaluation value calculated using the evaluation model. Select an action and store it in the memory,
  A decision support method, wherein the processor arbitrates the actions of the selected players using the mediation model to determine the actions of the players.   It is the decision support method according to claim 4,
  A method of supporting decision making, wherein the processor outputs screen data for displaying the actions of the selected players in time series.   It is the decision support method according to claim 4,
  7. The decision support method according to claim 1, wherein the processor outputs screen data for displaying an evaluation of each player for a set of actions of the plurality of players.

Images