Data processing: artificial intelligence – Knowledge processing system
Reexamination Certificate
1999-01-08
2003-12-23
Starks, Jr., Wilbert L. (Department: 2121)
Data processing: artificial intelligence
Knowledge processing system
C345S474000, C386S349000, C714S038110
Reexamination Certificate
active
06668249
ABSTRACT:
This application claims priority to Japanese Application No. 10-3213, filed Jan. 9, 1998.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an improvement of a technology which processes information distributed on a network using an agent, more particularly to an improvement which enhances flexibility and safety of an operation of an agent.
2. Description of the Related Art
Agent System
With reference to technologies for processing information distributed on a computer network, an agent system has been heretofore familiar. An agent is a process unit concerning software, and operates autonomously in response to peripheral circumstances. The agent system is the one that performs processing such as information collections while moving on nodes configuring the network according to demand. Here, the node is a logical unit that configures the network. A plurality of nodes can exist on a single machine, or on a computer.
FIG. 15
is a functional block diagram showing an example of a configuration of an agent system, which has been proposed in Japanese Patent Application No. 7-176181 by the applicant of the present invention as an example of such agent system. The agent system shown in
FIG. 15
connects a plurality of nodes
800
with a network
800
N, and in
FIG. 15
, only two nodes are illustrated, in spite of the fact that a plurality of nodes can be practically provided in the agent system. Among the nodes
800
N, the node used by the user for an agent generation is called a local node (
800
L), and the node to which the generated agent moves is called a remote node (
800
R).
In this agent system, each of the nodes
800
has input/output means
803
(L, R) which allows the user to perform an operation to generate the agent and receives the result of information processing by the agent. Agent management means
804
(L, R) of each node serves to generate an agent and erase an agent that fulfilled its duties. The means
804
(l, R) serves also to move the agent to the other nodes by transferring the information of the agent to other nodes and to accept an agent moved from other nodes. When the user wishes to perform some information processings using such agent system, he indicates the agent management means
804
L in the local node
800
L from the input/output means
804
L to generate an agent.
The most fundamental example will be described below. The user gives a script from the input/output means
803
L to the generated agent. The script is an action program of the agent. In the script, the action program concretely describes the contents indicating which node the agent moves to and what processing the agent performs. As a more concrete example of the script, for example, the contents are conceived, which expresses that the agent moves to the node A to transmit the copy of the file b to user's node, and then moves to the node B to transmit the copy of the file b to user's node, and so on. Interpretation execution means
802
(L, R) provided in each node executes such script to allow to the agent to act, thus realizing information processing to be objected.
In this case, agent information memory means
801
(L, R) provided in each node stores information necessary for the agent. The information necessary for the agent is each variable (called a script variable) necessary for the interpretation execution of the script, information collected by the agent and files if necessary, in addition to the foregoing script. As instructions described in the script of the agent, there is a move instruction to move the agent to other nodes, as well as an instruction capable of being executed only one node. The interpretation execution means
802
L sequentially executes the instructions of the script, and designates the node to which the agent moves when the agent is required to execute the move instruction, thereby making a request to the agent management means
804
for moving the agent.
In such agent system, when the user wishes to collect several files from the network, it is satisfactory that the user gives the action program to the agent and sends out onto the network. The agent which was sent out acts autonomously based on the presented script. For this reason, since it is unnecessary to keep communications between the user's node and the agent throughout, this agent system involves a merit that it exhibits a strong resistance to circuit troubles compared to conventional network functions such as “ftp” and “telnet”.
Example Configuration Using Planning
For the agent system shown in
FIG. 15
, agent systems has been also familiar, which are capable of changing the script that is the action program of the agent depending on circumstances.
Specifically, networks have recently became large-scale and complicated, and, when particularly they become the open system network by connecting them with a wide area network like an internet, components of the network such as a file position come to change. However, in the foregoing agent system shown in
FIG. 15
, since a fixed script is given to the agent at the time when it is generated, the agent can not change its action depending on circumstances. Accordingly, to cope with such change flexibly, the applicant of the present invention filed the agent system which possesses a planning function as the technology that automatically changes the action of the agent without any human agency.
In this technology, the action program of the agent is called a plan, and generation of the plan is called a planning. This technology appropriately remakes the plan depending on circumstances, and copes with changes of components of the network. Noted that remaking of the planning in response to the changes of the components of the network is called a re-planning.
An example of a configuration of such agent system is shown in a functional block diagram of FIG.
16
. In this technology, information used for the generation of the plan includes the information called “kowledge” and the action definition. The “knowledge” means an operation of the agent, particularly various kinds of information used for the planning. For example, the “knowledge” includes information relating to the components of the network, which, for example, indicates in which node files exist. In the example of the configuration of
FIG. 16
, the knowledge relating to the structure of the network is previously stored in the local information memory means
1
L. When the structure of the network changes, the update means
2
L reflects such change on the knowledge by an automatic detection, a manual operation or the like. Furthermore, the action definition means information indicating which kinds of instructions (action) can be used as parts for configuring the plan. The action definition is previously stored in the agent information memory means
3
.
In such agent system, the user instructing the generation of the agent gives a goal to be achieved to the node, instead of the script. Here, the goal describes a situation to be achieved as an object of the information processing, according to a predetermined grammar. The plan generating means
5
combines various kinds of actions included in the action definition while referring to the presented knowledge, and generates the plan for achieving the goal. In such agent system, changes of the structure of the network is reflected on the plan of the agent through the knowledge at the time of the planning and the re-planning, so that the agent can cope with the change of circumstances without human agency and change its action flexibbly.
Noted that means for generating such plan is called also “planner”, and its entity is a kind of program which expresses procedures of the planning. Furthermore, a wide sense concept naming the action program of the agent and each portion thereof is the script. The “plan” means the total of the script which was generated by the agent performing the planning shown in FIG.
16
.
Subsequently, concrete operation procedures of the agent system using the foregoing planning are exempl
Cho Kenta
Hattori Masanori
Kase Naoki
Banner & Witcoff , Ltd.
Starks, Jr. Wilbert L.
LandOfFree
Agent system and information processing method for such system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Agent system and information processing method for such system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Agent system and information processing method for such system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3108147