Data processing: artificial intelligence – Knowledge processing system – Knowledge representation and reasoning technique
Reexamination Certificate
2000-08-31
2004-05-25
Davis, George B. (Department: 2121)
Data processing: artificial intelligence
Knowledge processing system
Knowledge representation and reasoning technique
C706S048000, C706S045000
Reexamination Certificate
active
06741975
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to an expert system and to a method of providing automated advice.
The invention has particular, but not exclusive, application to the field of financial services.
Recently, the range of financial services products available to the consumer has begun to express a dichotomy. The introduction of new technologies, systems and practices, coupled with consumers' growing knowledge and confidence in the field of financial affairs have led to the development of “execution-only” sales. In such sales even complex financial products such as pensions and investment products—once thought of as requiring “face-to-face” consultation—are being sold on a commodity basis.
The public appetite for such products seems to be growing. Increasing affluence and a diminishing welfare state in many developed countries mean that people are considering private personal investments as desirable or even necessary. In addition, the pace of everyday life means that they have less time to spend planning and consulting advisors on these investments.
Newcomers to the field of financial services are providing execution-only financial products by a number of channels: using a traditional application form, by telephone contact and over the Internet, typically from a consumer's own personal computer (PC). All of these channels have a significantly lower cost to the provider and avoid the cost of the provision and training of a traditional network of human advisers. Consequently, such providers are able to offer a cheaper service and/or be more profitable than their more traditional competitors. Established companies in the field may be reluctant to enter the execution-only market for fear that their good name will be tarnished. Their market share is, therefore, likely to diminish.
Some consumers may be reluctant to conduct an “execution-only” transaction because of lack of confidence in their own ability in the relevant field. Equally they may be daunted by the prospect of a detailed, and possibly rather revealing, interview with a human expert. There is also the fear that the advice given must be paid for and there will be substantial hidden charges in whatever product or products are recommended. The customer thus needs to obtain the relevant assurance from a different source.
SUMMARY OF THE INVENTION
It is an object of the present invention to ameliorate the above disadvantages.
In a broad sense, a first aspect of the invention relates to an expert system comprising a knowledge base, means for deriving advice from that knowledge base and means for continually updating the knowledge base with advice given by human experts. The invention also resides in a method of providing an expert system comprising maintaining a knowledge base, deriving advice from that knowledge base and continually updating the knowledge base with advice given by human experts.
The term “advice” should be understood to comprise anything that may result from an expert consultation, for example a medical diagnosis or the results of a fault-finding procedure on a piece of engineering machinery.
Considering the financial advice scenario, by linking the expert system to the customer profiles and decisions made by a company's human financial advisers, the system is continually brought up-to-date without any need for particular and time-consuming action on the part of the operator of the system. Other benefits include the averaging of advice provided by the human experts and the transparency of the advice given. By providing continual, real data from a number of experts operating in the field, poor-quality advice given by a small number of the experts will not unduly influence the quality of advice given by the system. Consequently, an established company can provide services without human intervention and without compromising their reputation—they really can obtain the best of both Worlds.
The advice given will preferably, through its cosmetic appearance to the end-user/consumer, be associated with the company's brand rather than with a particular adviser. Consequently the company need not fear that an adviser will quit the company, taking clients with them.
The transparency of the system is extremely important should the quality of the advice given come into question. The rules applied in any particular case will be readily derivable from the system. This ensures that the company or institution which runs the system has the necessary information to counter any allegations that advice in any particular case was flawed.
It is preferred that the system further comprises means for deriving a plurality of rules in response to the knowledge base, wherein the means for deriving advice from the knowledge base comprises means for applying the plurality of rules derived from the knowledge base. This may also be expressed in method terms as deriving a plurality of rules in response to the knowledge base, and deriving advice from the knowledge base by applying the plurality of rules derived from the knowledge base. The rules suitably comprise fuzzy rules, whereby the means for deriving advice can be arranged to provide a plurality of advice together with a respective indication of suitability.
To enable explanation to lay persons not versed in machine code, it is preferred that the rules are expressed in a natural language. The system may further include means for publishing the rules used to derive the advice.
The means for deriving advice preferably comprises an agent and in any event the system is apt to be structured as a distributed system, which is preferred, although the system could be implemented as a stand-alone system.
To maintain a human face, the system preferably further comprises means for providing a user with a consultation with a human expert. Such consultation is preferably provided remotely.
Put more specifically, an aspect of the invention resides in a knowledge-based system adapted to provide a recommendation tailored to a consumer, comprising:
a knowledge base containing historical data;
rule extraction means for extracting a ruleset from the knowledge base;
a rules database for holding the ruleset;
codifying means for codifying the requirements of the consumer; and
recommendation means for applying the ruleset to the codified consumer
requirements and generating the recommendation accordingly; characterized by rule induction means for providing learning inputs to the knowledge base from a plurality of human experts as they advise and make real-life recommendations to actual or imaginary consumers, the learning inputs reflecting the recommendations made by the experts and the requirements of the consumers that they have advised; and update means for running the rule extraction means on the knowledge base to refresh the rules database by extracting an updated ruleset from the knowledge base for application by the recommendation means to the requirements of future consumers.
This aspect can also be expressed as a method of building a knowledge-based system adapted to provide a recommendation tailored to a consumer, the system operating by extracting rules from a knowledge base and applying the extracted rules to codified consumer requirements to generate the recommendation accordingly; wherein the method is characterized by providing learning inputs to the knowledge base from a plurality of human experts as they advise and make real-life recommendations to actual or imaginary consumers, the inputs reflecting the recommendations made by the experts and the requirements of the consumers that they have advised; and, after learning inputs have been provided to the knowledge base, extracting updated rules from the knowledge base for use in generating recommendations tailored to the requirements of future consumers.
The update means suitably operates periodically and the rule induction means provides learning inputs to the knowledge base at least as frequently as the update means operates to extract an updated ruleset from the knowledge base. It is preferred that the
Mackay Robin
Nakisa Ramin C.
Davis George B.
NCR Corporation
LandOfFree
Rule based expert system for consumer preference does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Rule based expert system for consumer preference, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Rule based expert system for consumer preference will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3263155