Interpretation phase for adaptive agent oriented software architecture

• US 6,990,670 B1
• Filed: 11/06/2000
• Issued: 01/24/2006
• Est. Priority Date: 11/05/1999
• Status: Active Grant

• Current Assignee: iAnywhere Solutions Incorporated (SAP SE)
• Original Assignee: iAnywhere Solutions Incorporated (SAP SE)
• Inventors: Hodjat, Babak
• Primary Examiner(s): An, Meng-Al T.
• Assistant Examiner(s): ZHEN, LI B
• Application Number:
  US09/706,948
• Time in Patent Office: 1,905 Days
• Field of Search: 719/317, 709/202, 709/317, 706/10
• US Class Current: 719/317
• CPC Class Codes: G06F 40/20 Natural language analysis semantic analysis of natural language G06F40/30
• CPC Class Codes: G06N 5/043 Distributed expert systems; Blackboards

Abstract

Roughly described, the interpretation phase of a system using an Adaptive Agent-Oriented Software Architecture allows queried agents to respond to queries before they have all their own responses from their own downchain agents. Alternatively or additionally, the originating agent can include a “depth-or-search” indication with each query, thereby preventing propagation of the query through more than the indicated number of agents. In the latter alternative, the originating agent can subsequently make the same query to the same downchain agents, but with an increased depth-or-search indication, if for example the originating agent is not yet satisfied with the claims it received in response to the first query.

Claims

1. A computer-implemented method for use in deriving user intent from a subject message, for use further with a network of agents each having a view of its own natural language interpretation domain, the network including a first agent, an upchain agent udchain of said first agent, and a plurality of immediately downchain agents each immediately downchain of said first agent, comprising the steps of:

said first agent receiving from said upchain agent a first query inquiring whether at least part of said subject message is within the natural language interpretation domain of said first agent;

said first agent querying each of said plurality of immediately downchain agents whether the queried agent considers at least part of said subject message to be in the queried agent'"'"'s natural language interpretation domain; and

said first agent responding to said upchain agent tentatively whether at least part of said subject message is within the natural language interpretation domain of said first agent, after said first agent receives at least one response from said immediately downchain agents, but before said first agent receives all responses from said immediately downchain agents.

2. A method according to claim 1, further comprising the step of, after said step of responding, said first agent responding further to said upchain agent whether at least part of said subject message is within the natural language interpretation domain of said first agent, after said first agent receives at least one additional response from said immediately downchain agents.

3. A method according to claim 2, wherein said step of said first agent responding further occurs in response to a second query received by said first agent from said upchain agent inquiring whether at least part of said subject message is within the natural language interpretation domain of said first agent.

4. A method according to claim 2, wherein said step of said first agent responding further occurs in response to said first agent receiving said at least one additional response.

5. A method according to claim 1, wherein said first query includes a first depth-of-search indication, further comprising the steps of said first agent:

resolving any conflicting responses from said queried agents to identify a prevailing one of said immediately downchain agents to whom said subject message should be passed; and

instructing said prevailing agent to handle at least part of said subject message.

6. A method according to claim 5, further comprising the steps of a particular one of said queried agents, in response to said query:

determining whether a depth of said particular agent exceeds said depth of search indication, and if so, disclaiming said subject message.

7. A method according to claim 5, further comprising the steps of a particular one of said queried agents, in response to said query where a depth of said particular agent does not exceed said depth of search indication:

determining whether at least part of said subject message is within said particular agent'"'"'s local natural language interpretation domain, and if so, returning a response to said first agent claiming at least part of said message.

8. A method according to claim 5, further comprising the steps of a particular one of said queried agents, in response to said query where a depth of said particular agent does not exceed said depth of search indication:

determining whether at least part of said subject message is within said particular agent'"'"'s local natural language interpretation domain;

and where said subject message is not within said particular agent'"'"'s local natural language interpretation domain but said particular agent has further agents downchain of said particular agent, querying at least one of said further agents whether the further agent considers at least part of said subject message to be in the further agent'"'"'s natural language interpretation domain.

9. A method according to claim 5, further comprising the step of, after said step of querying said immediately downchain agents a first time, querying said immediately downchain agents a second time whether the queried agent considers at least part of said subject message to be in the queried agent'"'"'s natural language interpretation domain.

10. A method according to claim 9, wherein said second query includes a second depth-of-search indication which exceeds said first depth-of-search indication.

11. A computer-implemented method for processing a subject message, by a network of agents including an originating agent, at least one agent downchain of said originating agent, and a plurality of agents each immediately downchain of a first one of said agents downchain of said originating agent, each agent in said network having a view of its own natural language interpretation domain, comprising the steps of:

said originating agent querying at least said first agent a first time, whether the first agent considers at least part of said subject message to be in the first agent'"'"'s natural language interpretation domain;

in response to said first query to said first agent, said first agent querying each of said agents immediately downchain of said first agent, whether the immediately downchain agent considers at least part of said subject message to be in the immediately downchain agent'"'"'s natural language interpretation domain;

in response to said first agent receiving at least one response from said immediately downchain agents but without waiting to receive responses from all of said immediately downchain agents, said first agent responding to said originating agent whether at least a part of said subject message is within the natural language interpretation domain of said first agent;

said originating agent subsequently querying said first agent a second time whether the first agent considers at least part of said subject message to be in the first agent'"'"'s natural language interpretation domain;

said originating agent resolving any conflicting responses from said agents downchain of said originating agent to identify a prevailing one of said agents downchain of said originating agent to whom at least part of said subject message should be passed; and

said originating agent instructing said prevailing agent to handle at least part of said subject message.

12. A method according to claim 11, wherein said prevailing agent is a community of agents.

13. A method according to claim 11, further comprising the steps of a second one of said agents queried in said step of said originating agent querying a first time:

determining whether at least part of said subject message is within said second agent'"'"'s local natural language interpretation domain;

and where at least part of said subject message is within said second agent'"'"'s local natural language interpretation domain, returning a response to said originating agent claiming at least part of said subject message.

14. A method according to claim 11, further comprising the steps of a second one of said agents queried in said step of said originating agent querying a first time:

determining whether at least part of said subject message is within said second agent'"'"'s local natural language interpretation domain;

and where said subject message is not within said second agent'"'"'s local natural language interpretation domain and said second agent has no further downchain agents, returning a response to said originating agent disclaiming said subject message.

15. A method according to claim 11, further comprising the steps of a second one of said agents queried in said step of said originating agent querying a first time:

determining whether at least part of said subject message is within said second agent'"'"'s local natural language interpretation domain;

and where said subject message is not within said second agent'"'"'s local natural language interpretation domain but said second agent has further agents downchain of said second agent, querying at least one of said further agents whether the further agent considers at least part of said subject message to be in the further agent'"'"'s natural language interpretation domain.

16. A method according to claim 11, wherein said step of said originating agent querying a first time comprises the step of providing to each agent queried in said step of said originating agent querying a first time, a first depth-of-search indication for said subject message,

and wherein said step of said originating agent querying a second time comprises the step of providing to each agent queried in said originating agent querying a second time, a second depth-of-search indication for said subject message, said second depth-of-search indication indicating a deeper search than said first depth-of-search indication.

17. A method according to claim 11, further comprising the steps of a particular one of said agents immediately downchain of said first agent:

determining in response to said first query whether at least part of said subject message is within said particular agent'"'"'s local natural language interpretation domain;

where at least part of said subject message is within said particular agent'"'"'s local natural language interpretation domain, returning a response to said first agent claiming at least part of said subject message; and

where said subject message is not within said particular agent'"'"'s local natural language interpretation domain but said particular agent has further agents downchain of said particular agent, querying in response to said second query at least one of said further agents whether the further agent considers at least part of said subject message to be in the further agent'"'"'s natural language interpretation domain.

18. A method according to claim 17, further comprising the steps of said particular queried agent:

receiving a group of at least one response from said further agents downchain of said particular agent, in response to said step of querying said further agents; and

returning a response to said first agent in response to said step of receiving.