Patent Grant
7558773
1. Technical Field
The present invention relates generally to expert systems; and, more particularly, it relates to expert system supported interactive product selection and recommendation
2. Related Art
Within conventional systems and methods that are operable to perform product selection and recommendation, a key limiting factor is the skill level of the agent or user who performs the customer or client interaction in selling and marketing those products. Within the scope of products, services are also envisioned as well within the context of the generic label “product.” Regardless of the industry, the skill level of the agent in communicating the available products to a customer and, as is oftentimes the case, educating the customer and helping him to quantify his needs, are limiting factors that govern the degree of effectiveness that a particular salesperson can achieve. Within technology areas that are more amorphous and complex, the limitations of salespeople are even further compromised.
There have been a number of technology areas where the availability of adequately skilled salespersons has been limiting to the full tapping of the market. As one example, the Internet has rapidly become an indispensable system that many businesses require simply to conduct their day-to-day affairs. During Internet access blackouts within such industries, when the access to the Internet has been temporarily interrupted, it is evident the degree to which businesses have grown dependent on the availability and reliability of such services. Yet, as is often the case, there are relatively few individuals within the company who are able to seek adequate Internet services to meet the needs of their business. While there is this sometimes great limitation of sufficiently skilled individuals within customer-businesses that seek such services, when there are not sufficiently skilled and trained salespeople, or agents, to sell those services, the problem becomes ever more complex. The likelihood of a sufficient solution that will adequately satisfy the needs of the customer is relatively low.
The conventional solutions have been geared towards training the salespeople in such situations to a level at which they can adequately communicate and understand the technology that they are seeking to sell. This approach, while being effective for a particular individual once that individual is sufficiently trained, is catastrophic for an industry that has a high degree of turnover. In the example used above to illustrate the deficiency of this traditional approach, the turnover rate of individuals, once they have acquired a high skill level, is oftentimes astronomical. There is such a high need for specifically trained individuals in these key technology areas that the lateral opportunities are sometimes too good to pass up. Moreover, the problem is further complicated by the fact that companies in the business of selling such goods and services, namely providers, are loathe to invest a high degree of money and effort to train their work base in light of the radical degree of turnover within the industry.
Another limitation within conventional approaches is the turn around time simply to provide a prospective customer with a list of available and operable options that may meet his needs. In some industries, given the high degree of technical complexity, there may be a turn around time of several weeks before a customer even gets an estimate or recommendation of products that may serve his needs. This latency is oftentimes extremely costly in terms of getting work up and running.
The frustration of customers who seek such products and services, when forced to deal with the all too often poorly qualified salespeople within their given industry, leads to an interaction that is less than effective. Oftentimes, both parties leave an interaction with a high degree of frustration. The conventional solution of simply trying to train the sales force to a sufficiently high level so as to accomplish this desired effectiveness has simply been deficient in light of many of the intrinsic limitations of the high technology industries. Moreover, even within relatively low technology level industries, there is oftentimes a great deal of “expert knowledge” that must or should be acquired before making an effective salesperson. Within these industries as well, the limitations of the skill level, experience, and expertise of the agent (salesperson) is often the limiting factor in the overall effectiveness of the solution.
Further limitations and disadvantages of conventional and traditional systems will become apparent to one of skill in the art through comparison of such systems with the present invention as set forth in the remainder of the present application with reference to the drawings.
Various aspects of the present invention can be found in a method used to perform product selection and recommendation using an expert system.
The method includes determining problem domain information, determining need information of a user, inputting the user need information into the expert system, transforming the user need information into a trait, and rating a product within the number of available products using the expert system. The trait is characteristic of a product of relevance to the user using the expert system. The product of relevance is selected from the available products.
In certain embodiments of the invention, the method is performed when an agent determines the need information of the user and inputs the user need information into an expert system via a graphical user interface. The business may also further include summarizing the ratings of the available products and providing explanation of the ratings of the available products. The summary of the ratings of the available products may be presented in various forms including listing of a recommended solution, a compatible solution, and a not recommended solution. Again, the generic use of the terminology product includes both products and services. The method may also include communicating the rating from the agent to a customer in situations where the rating of the product is performed in real time. The expert system employs a fuzzy value in performing rating of the product. Other values, including crisp values, may also be used by the expert system.
Other aspects of the present invention can be found in an expert system that is operable to facilitate interactive product selection and recommendation. The expert system includes an available products/services database, an expert system that is communicatively coupled to the available products/services database, and an agent interface that permits an agent to access the functionality of the expert system. The expert system performs processing using an available product within the available products/services database and a customer's need to rate the available product.
In certain embodiments of the invention, the expert system further includes a computer network that is communicatively coupled to the expert system, and the agent is able to perform remote access of the functionality of the expert system via a graphical user interface that communicatively couples to the computer network. The expert system may be supported locally on a platform that also supports the agent interface. It may alternatively be supported remotely on any other platform that the agent may access via the computer network. The expert system performs dynamic calculation to perform the rating of the available product. The agent is able to interact with a customer in real time, and the agent can communicate information corresponding to the rating of the available product to the customer. The expert system may further include a computer network communicatively coupled to the available products/services database such that the available products/services database is updated in real time, via the computer network, with information of products/services that are provided by a plurality of providers.
Other aspects of the present invention can be found in a method to perform interactive product selection and recommendation. The method includes inputting a customer need into an expert system via a graphical user interface, performing expert system processing considering the customer need in light of business specific options, and generating output that identifies a recommended solution using dynamic calculation within the expert system by employing a fuzzy variable used to characterize the customer need. The recommended product solution is selected from the available products.
In certain embodiments of the invention, the available products include telecommunication technologies. Those telecommunication technologies may also include Internet access services and/or data network services. The method may also be performed where the inputting of the customer need into the expert system via the graphical user interface is performed during real time interaction between an agent and the customer. The generated output may also include a compatible solution, among other types of output.
Other aspects, advantages and novel features of the present invention will become apparent from the following detailed description of the invention when considered in conjunction with the accompanying drawings.
A better understanding of the present invention can be obtained when the following detailed description of various exemplary embodiments is considered in conjunction with the following drawings.
Within all of the various embodiments of the invention described below, the use of terminology of product or products is intended to include both services and products. The generic use of the term “products” or “product” is in no way to limit the operability of various aspects of the invention, as it is equally applicable within both industries selling and providing services, as well as those selling and providing products or combinations of services and/or products. For brevity, the term product is sometimes used instead of goods and/or services.
The invention is operable to allow and assist an agent to select certain solutions to meet the needs of a customer. In one embodiment, the invention is operable to help an agent select telecommunication technologies and products based on customer needs. In light of the comments made above regarding products and services, telecommunication technologies is also included within the scope and spirit of the terminology of “products and services” or simply, “products.” The invention employs functionality offered within fuzzy logic expert systems. From certain perspectives, various aspects of the invention are based on fuzzy logic expert systems.
To assist an agent to select an amenable product for a customer, the problem domain must first be determined. The problem domain is the general area where the customer needs a solution. For example, Internet access would be one problem domain. Another one would be local data services. There are numerous problem domains where the functionality of the invention may assist an agent to find a customer solution. Once the problem domain is determined, the agent may then be presented with a set of questions relating to the needs of the customer with respect to that problem domain. The customer's answers are then transformed into values for each element of a set of product traits. In addition, the relative importance of each of the traits, from a customer's perspective, may also be considered.
A trait is a characteristic of a product of relevance to the user. These traits include any number of parameters and may include measurements such as cost, bandwidth, reliability, availability, etc. Each trait that is measurable in a continuum is represented as a fuzzy variable.
Other traits are not measurable in a continuum, but are rather better represented as crisp values. One such example of a crisp value is the need for static IP addresses or the location of the site. These are represented with non-fuzzy values (crisp values). An expert system built in accordance with the invention then takes the collection of these values that represents the needs of the customer (one per trait, called needs) and attempts to match them against the corresponding collection of values provided by each product (called specs). Based on this, the expert system is operable to obtain a collection of ratings: one rating for each combination of product and trait. The expert system is also operable to summarize the ratings to give each product a summary rating. The summary rating of each product are used to sort them and classify them with respect to the level of recommendation assigned to each product.
In determining success of the operation of an expert system built in accordance with the invention, one consideration is that a customer belong to a set of “satisfied customers” once they deploy a solution suggested by an agent. In traditional set theory, an element either belongs to a set or does not. So, if customer satisfaction were treated in such a manner, then customers would either be “satisfied” or “dissatisfied.” However, customers are rarely 100% satisfied or 100% dissatisfied. Their satisfaction usually falls somewhere in the middle.
The invention is operable using Fuzzy Sets. In Fuzzy Set Theory, an element does not have to belong to a set 0 or 100%. Instead, an element may belong to the set to a given degree.
This type of model provides a lower semantic gap when working with sets that don't have precise mathematical definitions (as many don't). In fuzzy sets, each set has a corresponding membership function that maps all elements from the Universe of Discourse to a real number in the range [0, 1]. If the membership function maps an element to 1, then it completely belongs to the corresponding set. If it maps an element to 0, then it does not belong to the set at all. Other values provide a more fuzzy definition of whether the element belongs to the set. In the example of customer satisfaction, the membership function determines the level of satisfaction of a customer.
In many problems, the Universe of Discourse is a range of real numbers. As some examples: bandwidth, cost and reliability traits are all measurable and can be represented by real numbers. So, as an example, a fuzzy set may be created to represent the level of satisfaction of a customer who uses ISDN using the following membership functions as the spec of the ISDN product:
P(x)=1 for x<=48
P(x)=2−x/48 for 48<x and x<96
P(x)=0 for x>=96
Where x is measured in kbps. ISDN has a bandwidth of 64 kbps and in this case, the domain expert who came up with this membership function estimates that customers who need a bandwidth of 48 kbps or less will be completely satisfied with ISDN. Customers that need a bandwidth between 48 kbps and 96 kbps will have a linear reduction in their level of satisfaction. Finally customers who need 96 kbps or more will be completely dissatisfied with this solution.
Presumably, we will be able to obtain this type of information from domain experts. This particular fuzzy set is meant only as an example. On the other hand, the bandwidth requirements for a customer can rarely be estimated precisely. So, a fuzzy set can also be used to represent the bandwidth need of a customer:
N(x)=0 for x<=50
N(x)=x/10−5 for x>50 and x<=60
N(x)=−x/10+7 for 60<x and x<=70
N(x)=0 for x>70
This set represents a customer whose bandwidth need is estimated to be between 50 and 70 kbps. The set peaks at 60 kbps because it is the most likely bandwidth requirement. One assumption may be made: if the fuzzy set that represents a customer need is a subset of what is offered by a product, then it may be assumed that the customer will be perfectly satisfied. It may also be determined if there is no overlap, then the customer will be completely dissatisfied. If there is partial overlap there will be partial satisfaction. Fuzzy logic provides rules to determine the level of satisfaction.
There are at least three practical considerations that affect how fuzzy sets are represented. First, since fuzzy sets deal with concepts that are difficult to pin down, it is rare to come up with complex membership functions. Second, in order to simplify calculations, fuzzy sets are commonly defined with a set of linear functions. Third, although it is relatively easy to come up with equations for line segments, it quickly becomes bothersome. So, a common notation for fuzzy sets is shown as follows:
FUZZY−SET ::=[POINTS]
POINTS ::=POINT|POINT POINTS
POINT ::=Y/X
Y ::=<a number in the interval 0.0 to 1.0 inclusive>
X ::=<any number>
The first y0/x0 point delimits a horizontal line from (−infinity, y0) to (x0, y0). The last yn/xn point delimits a horizontal line from (xn, yn) to (infinity, yn). All other consecutive points (xi, yi) (xi+1, yi+1) delimit straight lines.
As examples, the 2 membership functions defined above could be written as:
ISDN−BANDWIDTH−SATISFACTION=[1/48 0/96]
CUSTOMER−BANDWIDTH−NEEDS=[0/50 1/60 0/70]
Fuzzy sets defined with 2 coordinates are called Z Fuzzy Sets when y0>y1. They are called S-Fuzzy Sets when y0<y1. This is in reference to the shape of the membership function. There are also triangular, rectangular and trapezoidal fuzzy sets, where the name is assigned based on the shape of the membership function.
In an expert system built in accordance with the invention, the use of Z and S Fuzzy Sets to define each trait of each product may be employed. For example, for traits where higher numbers are preferable, such as bandwidth and reliability (measured as MTBF or mean time between failures), a Z Fuzzy Set is used. For traits where lower numbers are preferred, such as cost or MTTR (Mean Time To Repair), an S Fuzzy Set is used. To represent customer needs we generally use triangular fuzzy sets, although other shapes are allowed and the expert system will work with any arbitrary membership function for any fuzzy set. When the needs of the customer are known precisely or where there are no means to “fuzzify” the needs of the customer, it is valid to use a crisp value to represent its needs. A crisp value is just a special case of a fuzzy value where the membership function evaluates to one on the crisp value and to zero elsewhere.
A common notation in Fuzzy Set Theory is that for a given Fuzzy Set A its membership function is also called A and defined as
A: UOD→[0, 1]
Where UOD is the Universe of Discourse. Using this notation we define the following set operations:
C=A ∩ B iff C(x)=min(A(x), B(x))∀× in UOD
C=A UB iff C(x)=max(A(x), B(x))∀× in UOD
C=Ā iff C(x)=1−A(x)∀× in UOD
The height of a set A is h(A) and is defined as the maximum value produced by its membership function.
When the spec of a product P is matched against the need N (where N is a crisp value), for a given trait, the fuzzy logic expert system produces P(N) as the matching measure.
When the spec of a product P is matched against the need N (where N is a fuzzy value), for a given trait, the fuzzy logic expert system proceeds as follows:
Good=h(P ∩ N)
Bad=h(
Rating=Good/(Good+Bad)
Rating is the matching measure. The rating formula described above is not in the literature. The customary way of obtaining the level of matching of 2 fuzzy sets is:
h(P ∩ N)
This is, however, overly optimistic.
The concept of a fuzzy variable is relevant and important as well, because not all fuzzy sets are comparable. For example, a fuzzy set defined to represent the bandwidth of a product and one defined to represent the cost cannot be usefully compared. For this reason there is another concept called fuzzy variable. A fuzzy variable defines a Universe of Discourse. Each fuzzy variable stipulates a name for the UOD as well as its valid range of values and the unit of measurement.
Fuzzy variables in an expert system built in accordance with the invention are used to model traits. For example, to model bandwidth, a fuzzy variable with the domain [0, 100,000] is created where each unit represents one kbps.
A fuzzy value is a fuzzy set in the context of a fuzzy variable. This is what may be used in the expert system to represent a user need for a given trait or the spec of a product with respect to a given trait. A fuzzy set is like a number (e.g., 50); a fuzzy value is like a number with a unit of measurement and a meaning (e.g., $50 monthly cost.).
Various embodiments of the invention that employ an expert system operable according to the parameters described above are further fleshed out as shown in the various Figures and description below.
A number of provider(s) 132 are able to provide any number of services and/or products to a customer 104. Each of the products that are furnished by the providers 132 is catalogued within a product/service database 130. The product/service database 130 is communicatively coupled to the computer network(s) 180. If desired, the providers 132 are also provided with communicative coupling to the computer network(s) 180 as well.
An agent 102 and the customer 104 interact via any number of communication means including telephone, Internet, or any other operable communication media. The agent 102 is able to access the various functionality of an expert system 131 while interacting with the customer 104. If desired, the agent 102 is able to perform accessing of expert system 131 while offline with respect to the customer 104. For example, the agent 102 is able to perform interaction with the customer 104 initially, then to use the expert system 131 to perform analysis and provide recommendations of what products should be provided to the customer 104. However, in one embodiment of the invention, the agent 102 is afforded the opportunity to perform real time interaction with the customer 104 and to use the various functionality of the expert system 131 to provide recommendation of products provided by the provider(s) 132 that would provide a suitable solution for the customer 104. It is also noted that the customer 104 himself may access all of the functionality of the various aspects of the present invention, in similar manner that the agent 102 access the functionality, in various embodiments as well. That is to say, the customer 104 may be given access to the same functionality that the agent 104 enjoys to the various aspects of the present invention.
The interactive product selection/recommendation system 100 allows for the ability of an agent 102 of relatively low skill level to provide recommendations to the customer 104 of the available products that proffered by the various provider(s) 132, borrowing on the various functionality of the expert system 131. The interactive product selection/recommendation system 100 is amenable to providing sales support for the providers 132 without requiring very highly skilled and trained salespersons within many niche industries that typically require an enormous amount of experience, expertise, and training.
Particularly within industries where there is a gross mismatch between availability of adequate and competent salespeople and the industry's customer demand, the interactive product selection/recommendation system 100 offers a system in which anyone of the providers 132 is able to achieve significantly greater customer interaction and sales than that afforded using the conventional methods and systems that necessitate a great deal of training and expertise within a given industry before being able to provide adequate sales servicing. Whereas a conventional system is limited by the individual ability, expertise, and training of an agent operating within it, certain aspects of the invention provide for a system whose high end performance is not limited by the personal characteristics of the agents that operate it.
The particular connection to the expert system 231 from the agent interface 290, that thereby allows agent interaction with the expert system 231, is achieved any number of ways without departing from the scope and spirit of the invention. For examples, it may be that the expert system 231 operates locally on the platform on which the agent is using as the agent interface 290. Alternatively, it is run remotely on a server 281 where the expert system 231 is accessed by the agent via any number of electronic connections. For example, the agent interface 290 is operable to access the expert system 231 (supported remotely by the server 281) via a wireline network 270 that communicatively couples the agent interface 231 to the expert system via the Internet. Also, the wireline network 270 may be operable to access the expert system 231 directly, without going through the Internet 280, if the wireline network 270 is operable to communicatively couple to the server 281 on which the expert system 231 is supported.
In even other embodiments, the agent interface 290 is operable to perform wireless connectivity to the wireline network 270 and/or the Internet 280. For example, each of the various platforms may be operable to perform the agent interface 280 may themselves be operable to perform wireless communication, as shown by the antennae on the laptop computer 291, the computer 292, . . . and the handheld computer 293. In such instances, the agent interface 280 is operable to connect to the wireline network 270 and/or the Internet 280. The wireless connectivity may be achieved using a satellite 272 that communicatively couples to a satellite-based wireless network 271, or alternatively using a wireless tower 276 that communicatively couples to a terrestrial-based wireless network 275. Generically, any other wireless interface 279 may be used to communicatively couple the agent interface to one of the wireline network 270 and/or the Internet 280 to allow interaction of the agent with the expert system 231 when the expert system 231 is operated remotely on the server 281. The term “server,” for the server 281 is generically used to illustrate any number of computing platforms possessing sufficient processing resources to support the expert system 231.
The expert system 231 is operable also to access a product/service database 230. Again, as mentioned above and as will also be described in various embodiments of the invention below, the product/service database 230 is updated with the available products (including services) that may be sought by a customer. The listings of such available products may be provided via direct link by a number of provider(s) 232, or the providers 232 may provide the information to the product/service database 230 via the Internet 280. In addition, the server, on which the expert system 231 is supported for remote operation, may access the product/service database 230 directly via a dedicated link or via the Internet 280. It is understood that any number of dedicated wireline or wireless based connections, and/or networks, are also operable to perform the communicative coupling of the expert system 231 to the product/service database 230. The
The interactive product selection/recommendation method 300 is operable in real time where an agent may interact with a customer while inputting the customer's various needs and the agent can then provide recommendation to the customer. This recommendation is based on the customer's provided needs against the backdrop of the available products that are available to him. It may be that given the geographical location of the customer's site or sites that only a predetermined number of services are available to him. Such limitations are inherently provided for within the products/services rating that is performed in the block 340. As will also be shown in various embodiments of the invention below, the ratings summary provided for in the block 350 is able to also provide the underlying reasoning of why a particular solution is being recommended over other competing solutions. There is even greater detail provided, in certain embodiments, where compatible solutions are presented to the agent to provide for a more educated recommendation to a customer. In addition, detailed description of the available products is also available to a user of the interactive product selection/recommendation method 300 as well.
Within the problem domain determination functional block 410, an indefinite number of trait(s) 412 and available product(s)/service(s) 414 are used to generate specs 416 that are used as input to the expert system 431. The traits 412 is a characteristic of a product and/or service that is relevant to a particular customer. The available product(s)/service(s) 414 also provide even greater detailed information of the various available product(s)/service(s) for the user of the interactive product selection/recommendation system 400. This greater detailed information may be accessed via a link, or selectable option, or in any manner known in the art.
This is correlated to the user need determination performed in the block 420. The user need determination 420 may be performed using a number of means including asking the customer direct questions 421 and indirect questions 422. In addition, the agent using the interactive product selection/recommendation system 400 is bound to possess some degree of knowledge, though it is understood that a novice or very inexperienced user (agent) may nevertheless use the interactive product selection/recommendation system 400 as well. In situations where the agent does possess a degree of knowledge, expertise, and skill, the agent's knowledge/input 423 may also be used to perform the user need determination 420. The agent may also make modifications of a number of default user needs as shown by a block 424. The agent may use the agent modification block 424 to override system input when his experience or skill level suggests doing so.
After using any of these various means to acquire knowledge, the user's needs determination block 420 provides input to the expert system 431 where the expert system 431 uses this information, along with information from the problem domain determination block 410, to generate user's needs as shown in a block 425. Within the context of expert systems, an indefinite number of fuzzy values 426 are generated. In certain instances, there are situations where a user need must be represented by a crisp value 427, or hard/rigid value. There are other instances are where no crisp values are used at all. An example of such an instance is where questions posed by an agent to a customer do not permit the user to convey a level of uncertainty. In such instances, that particular need or needs should be represented by crisp values 427. Another example of a situation that would best be represented by a crisp value 427 would be the actually availability of certain products in a given location. There may be a situation where only certain services are available in a particular city; such a situation is better represented using crisp values 427.
To state the concept of a crisp, or non-fuzzy value, in another way, some product traits are not measurable in a continuum. For those product traits that may not be represented or are not measurable using a continuum, a crisp value may be used. For example, the customer's physical location and the need for a static IP addresses have a bearing on the recommended product. These traits are not fuzzy. So, take support for static IP address, either a product supports it or not. If a product does not support it and the customer needs it, then, for this trait, the level of customer satisfaction is marked as zero. How this is summarized depends on the relative importance of this trait. The main difference with fuzzy traits is that for crisp traits traditional logic is used and in most cases it may be concluded that the customer is either 100% satisfied or 0% satisfied.
As mentioned briefly above, the specs 416 and the needs 425 are used as inputs to the expert system 431. The expert system 431 performs product(s)/service(s) rating 440 using an indefinite number of rules 441. The product(s)/service(s) rating summary 450 includes a ranking of an indefinite number of products and services, shown as a product #1451, a product #2452, . . . , and a product #n 453 and a service #1454, a service #2455, . . . , and a service #m 456. For a given problem, a sorted set of classified solutions is returned. The classification includes the level of customer satisfaction achieved by the product. There is a natural bifurcation of performing ratings of services in one instance, and performing ratings of products in another. However, there may be some instances where a product coupled with some services may also be rated. In addition, if the problem requires a product and a service, then the returned solutions will be configured with a specific set of products and a specific set of solutions.
There are many permutations in which the product(s)/service(s) rating summary 450 may be provided to an agent without departing from the scope and spirit of the invention. In one embodiment, the product(s)/service(s) rating summary 450 is provided in a document type of format that an agent may subsequently modify. One such example is the outputting of the product(s)/service(s) rating summary 450 into an MS WORD document that an agent may then edit before forwarding a report onto a customer.
Similarly, the domain expert input 526 generates specs fuzzy values 533 that are also provided to perform the expert system rating 531. The specs fuzzy values 533 are representative of the expected level of customer satisfaction for the various information provided via the domain expert input 526. The specs fuzzy values 533 include a good value 534 and an optional bad value 535. The bad value may simply be the complement of the good value 534 in certain embodiments of the invention. These specs fuzzy values 533 are then provided to the expert system to perform expert system rating 531.
As mentioned above, the GUI 527 illustrated in this embodiment is illustrative of one manner in which the domain expert input 526 may be input into the expert system to perform expert system rating 531 that is operated using various aspects of the invention.
The GUI 527 may be presented as a table that the domain expert input 527 may use. For any number of traits, shown as a trait #1528, . . . and a trait #m 529, various information may be provided in the tabular format. For example, as illustrated for the trait #1528, a user need(s) fuzzy set may be located at the top (user need(s) fuzzy set 538) or bottom (user need(s) fuzzy set 539) of the table, as desired in a given situation. The table may be arranged showing the available products, followed by their description, and subsequently by their rank. This is shown graphically as a table arranged from left to right and top to bottom as follows: (1st line) product #1541 followed by a product #1 description 542, followed by a product #1 rating 543; (2nd line) product #2551 followed by a product #2 description 552, followed by a product #2 rating 553; . . . (nth line) product #n 561 followed by a product #n description 562, followed by a product #n rating 563. Other variations of the GUI 527 may be used without departing from the scope and spirit of the invention. The GUI 527 is exemplary one just one embodiment that is operable to achieve the domain expert input 526 in a user-friendly manner.
The agent 602 is able to interact with the customer 604 in real time to perform input of business specific options 612 and to assess the needs of the customer 604, as shown by the needs assessment functional block 613. The business specific options 612 may include additional information regarding the various available products. As described above, any number of traits 614 may be used to perform the needs assessment. The expert system is able to receive these inputs from the business specific options 612 and the needs assessment functional block 613, including the traits 614, in real time. They are then provided to an expert system processing functional block 620. Then, the expert system provided output 615 is displayed via the GUI 610 after performing dynamic calculation. The limiting latency here is the latency of the connectivity between the GUI 610 and the expert system that is used to perform the expert system processing 620. The expert system processing is performed in nearly real time, having sub-second calculation time.
The output 615 includes recommended solutions 616, compatible solutions 617, and (if desired) not recommended solutions 618. In addition, within certain embodiments, an explanation is available for the agent 602 to describe even further the reasons why the particular solutions have been categorized into the various categories, namely, the recommended solutions 616, the compatible solutions 617, and the not recommended solutions 618. This can be most helpful during interaction with the customer, particularly when the customer desires a little more clarification of why a certain solution was recommended and why another was not. The real time input and dynamic calculation offered by the expert system built in accordance with the present invention allows the agent 602 to perform the entire fielding and servicing of a potential client Attorney Docket during a single interaction. Compare this with the oftentimes multi-week-long turnaround in getting a customer a recommendation using conventional approaches. As an example, during a single telephone conversation between the agent 602 and the customer 604, the agent 602 is able to input all of the requisite information that the expert system requires. During the same telephone conversation, after the very short duration expert system processing 620, the agent 602 is able to provide virtually immediate recommendation to the customer 604.
The practically real time interaction that is afforded to the agent 602, as well as the explanation that is provided for the various options within the output 615, allow for a relatively inexperienced agent to perform recommendation for the customer 604. The agent 602 may lean on the offered functionality of the system to fill his deficiencies in terms of skill, expertise, and experience within the particular product area. The invention is also very desirable for product providers because they are able to employ relatively low skill level personnel to perform the selling of their products and/or services. This can ensure a greater profitability. Particularly within areas where there is a high degree of turnover, the invention allows a solution where a consistent level of expertise is proffered to the customer 604. In addition, in contexts where a high degree of skill is required to perform the sale of such products and/or services, a provider need not invest such a large amount of up-front effort to get a prospective employee into a position where he may be an effective salesperson or agent.
The
The manual mode 712 allows for an agent to override any expert system processing. That is to say, the manual mode 712 allows an agent to operate using the GUI 710 in a manner analogous to that employed within conventional approaches, such as allowing the agent to perform the selection of products based on his own knowledge, expertise, and experience. Alternatively, the guidance mode 714 employs the expert system that is operable to perform guidance of the agent through the recommendation/selection process. Within either mode, the guidance mode 714 or the manual mode 712, the product description is made available to the agent to allow for further description of the various available products when interacting with the customer as also described above in various embodiments.
The computer network 880 may include the Internet 881 in certain embodiments of the invention. However, the computer network 880 may also be a dedicated network, such as a local area network (LAN) or other localized network, that is operable to communicatively couple the various service and product providers.
The databases of the various service and product providers are categorized in any number of manners. Exemplary information for the various databases are shown in the embodiment of the
In addition, any number of service providers may also be integrated into the available products/services database (catalog) 830. For example, a database for a service provider #1830 includes information such as a listing of available services 832, a listing of supported services 834, a listing of compatible services 836, . . . and a listing of any other provider information 839. Similarly, any number of other service providers may also be used as well without departing from the scope and spirit of the invention. For example, a database for a service provider #m 840 includes information such as a listing of available services 842, a listing of supported services 844, a listing of compatible services 846, . . . and a listing of any other provider information 849. Other information may also be included within the various databases for the various product and service providers without departing from the scope and spirit of the invention.
For example, a computer network 980 is operable to perform real time updating of any number of customer profiles into a customer profile database 930 that is used within an expert system. The customer profile database 930 is also communicatively coupled to the computer network 980. The updating of the number of various customer profiles is performed in any number of manners, including real time, on-line, and off-line, without departing from the scope and spirit of the invention.
The computer network 980 may include the Internet 981 in certain embodiments of the invention. However, the computer network 980 may also be a dedicated network, such as a local area network (LAN) or other localized network, that is operable to communicatively couple the various customer profiles.
The databases of the various customer profiles are categorized in any number of manners. Exemplary information for the various customer profile databases are shown in the embodiment of the
In addition, any number of other customer profiles may also be used as well without departing from the scope and spirit of the invention. For example, a database for a customer profile #n 920 includes information such as needs/desires 922, financial information 924, . . . and any other customer information 929 as well. Similarly, the financial information 924 may further include information concerning cost ranges 925.
In addition, the various customer profiles may also include any number of default or user defined profiles as well. For example, a “typical” customer profile, defined by a domain expert 930, may be used in certain embodiments of the invention. This domain expert defined “typical” customer profile 930 may include similar information as that provided by the customer profiles #1910, . . . and the customer profiles #n 920. For example, the domain expert defined “typical” customer profile 930 may include information such as needs/desires 932, financial information 934, . . . and any other customer information 939. If further desired, the financial information 934 may further include information concerning cost ranges 935. However, the domain expert defined “typical” customer profile 930 may include different information as well as determined by the domain expert.
The domain expert may also define any number of “other” customer profiles, besides a simple “typical” customer profile. For example, a domain expert defined “other” customer profile 940 may include similar information as that provided by the customer profiles #1910, . . . and the customer profiles #n 920. For example, the domain expert defined “other” customer profile 940 may include information such as needs/desires 942, financial information 944, . . . and any other customer information 949. If further desired, the financial information 944 may further include information concerning cost ranges 945. In addition, if desired by the domain expert, other factors may also be included as well. A customer-specific factor 946 is also included in certain embodiments of the invention, as determined by the domain expert.
In this embodiment, the domain expert 1120 is also afforded the ability to perform expert system processing engine rule(s) modification 1140 for the processing engines employed by the expert system 1110. The modified processing engine rule(s) 1130 are then employed by the expert system 1110. Similarly, the domain expert 1120 is also afforded the ability to perform summary rating rule(s) modification 1160 for the summary rating operations within the expert system 1110. The modified summary rating rule(s) 1150 are then employed by the expert system 1110. Moreover, the domain expert 1120 is also afforded the ability to override default operation(s) 1180 for the default operations to be performed by the expert system 1110. The overridden default operation(s) 1170 are then employed within the default operations of the expert system 1110.
In addition to the robustness, variability, and adaptability of an expert system that is built in accordance with the invention, such an expert system has been generated in light of a number of goals as well. The expert system is operable to change its recommendations smoothly as the inputs change. For example, adding one more employee to a site should not lead to drastic changes to the recommendations. The expert system is operable to accommodate minor changes and deviations of inputs without catastrophically failing, or by producing radically different output.
Moreover, a domain expert of relatively low skill level, or minimal computer training, is also to perform a number of tasks, some of which are enumerated in the
As described above in many of the various embodiments, the expert system provided for sub-second response time in its own code (not counting network traffic) for any input-to-recommendation transaction. This enables real time interaction between agent and customer. In addition, the expert system is able to provide explanation for the reasons that lead to particular recommendations.
It is also noted, as will be understood by those persons having skill in the art, that certain aspects of the present invention may be stored on media, in software form having an instruction set, so that it may be transported from device to device. This way, any suitable processing circuitry, including microprocessors, computers, and other processing devices, may execute that instruction set to achieve the various aspects of the present invention.
In view of the above detailed description of the present invention and associated drawings, other modifications and variations will now become apparent to those skilled in the art. It should also be apparent that such other modifications and variations may be effected without departing from the spirit and scope of the present invention.
This non-provisional application is a continuation of U.S. Nonprovisional application Ser. No. 09/909,250, filed Jul. 19, 2001, which claims priority from U.S. Provisional Application Ser. No. 60/219,783, filed Jul. 19, 2000, and U.S. application Ser. No. 09/764,662, filed Jan. 18, 2001, which claims priority from 60/241,541, filed Oct. 18, 2000. The following U.S. Patent Applications are hereby incorporated herein by reference in their entireties and made a part of the present U.S. Patent Application for all purposes: U.S. patent application Ser. No. 09/909,241, entitled “Expert System Adapted Dedicated Internet Access Guidance Engine” and U.S. patent application Ser. No. 09/909,240 entitled “Expert System Adapted Data Network Guidance Engine.”
| Number | Name | Date | Kind |
|---|---|---|---|
| 4949278 | Davies et al. | Aug 1990 | A |
| 4992940 | Dworkin | Feb 1991 | A |
| 4999833 | Lee | Mar 1991 | A |
| 5175800 | Galis et al. | Dec 1992 | A |
| 5309355 | Lockwood | May 1994 | A |
| 5412576 | Hansen | May 1995 | A |
| 5434994 | Shaheen et al. | Jul 1995 | A |
| 5493490 | Johnson | Feb 1996 | A |
| 5546539 | Poling | Aug 1996 | A |
| 5701419 | McConnell | Dec 1997 | A |
| 5758257 | Herz et al. | May 1998 | A |
| 5784539 | Lenz | Jul 1998 | A |
| 5809282 | Cooper et al. | Sep 1998 | A |
| 5822743 | Gupta et al. | Oct 1998 | A |
| 5951694 | Choquier et al. | Sep 1999 | A |
| 5963939 | McCann et al. | Oct 1999 | A |
| 5970482 | Pham et al. | Oct 1999 | A |
| 5983220 | Schmitt | Nov 1999 | A |
| 5987376 | Olson et al. | Nov 1999 | A |
| 6012051 | Sammon, Jr. et al. | Jan 2000 | A |
| 6018722 | Ray et al. | Jan 2000 | A |
| 6029195 | Herz | Feb 2000 | A |
| 6032129 | Greef et al. | Feb 2000 | A |
| 6035283 | Rofrano | Mar 2000 | A |
| 6067525 | Johnson et al. | May 2000 | A |
| 6070149 | Tavor et al. | May 2000 | A |
| 6088722 | Herz et al. | Jul 2000 | A |
| 6105003 | Morohashi et al. | Aug 2000 | A |
| 6112186 | Bergh et al. | Aug 2000 | A |
| 6117187 | Staelin | Sep 2000 | A |
| 6119101 | Peckover | Sep 2000 | A |
| 6125356 | Brockman et al. | Sep 2000 | A |
| 6138105 | Walker et al. | Oct 2000 | A |
| 6145002 | Srinivasan | Nov 2000 | A |
| 6169979 | Johnson | Jan 2001 | B1 |
| 6178546 | McIntyre | Jan 2001 | B1 |
| 6219654 | Ruffin | Apr 2001 | B1 |
| 6220743 | Campestre et al. | Apr 2001 | B1 |
| 6233609 | Mittal | May 2001 | B1 |
| 6236978 | Tuzhilin | May 2001 | B1 |
| 6243754 | Guerin et al. | Jun 2001 | B1 |
| 6247005 | Edwards et al. | Jun 2001 | B1 |
| 6266649 | Linden et al. | Jul 2001 | B1 |
| 6285986 | Andrews | Sep 2001 | B1 |
| 6317722 | Jacobi et al. | Nov 2001 | B1 |
| 6332129 | Walker et al. | Dec 2001 | B1 |
| 6341276 | Bender et al. | Jan 2002 | B1 |
| 6356884 | Thaler | Mar 2002 | B1 |
| 6397193 | Walker et al. | May 2002 | B1 |
| 6442438 | Naillon | Aug 2002 | B1 |
| 6473791 | Al-Ghosein et al. | Oct 2002 | B1 |
| 6473794 | Guheen et al. | Oct 2002 | B1 |
| 6477520 | Malaviya et al. | Nov 2002 | B1 |
| 6519571 | Guheen et al. | Feb 2003 | B1 |
| 6526387 | Ruffin et al. | Feb 2003 | B1 |
| 6529954 | Cookmeyer, II et al. | Mar 2003 | B1 |
| 6553346 | Walker et al. | Apr 2003 | B1 |
| 6578054 | Hopmann et al. | Jun 2003 | B1 |
| 6600749 | Hayball et al. | Jul 2003 | B1 |
| 6601036 | Walker et al. | Jul 2003 | B1 |
| 6658396 | Tang et al. | Dec 2003 | B1 |
| 6671716 | Diedrichsen et al. | Dec 2003 | B1 |
| 6741975 | Nakisa et al. | May 2004 | B1 |
| 6745172 | Mancisidor et al. | Jun 2004 | B1 |
| 6748369 | Khedkar et al. | Jun 2004 | B2 |
| 6753887 | Carolan et al. | Jun 2004 | B2 |
| 6799222 | Endo | Sep 2004 | B1 |
| 6826549 | Marks et al. | Nov 2004 | B1 |
| 6981040 | Konig et al. | Dec 2005 | B1 |
| 7006990 | Stolze et al. | Feb 2006 | B2 |
| 7047518 | Little et al. | May 2006 | B2 |
| 7139728 | Rigole | Nov 2006 | B2 |
| 20010014868 | Herz et al. | Aug 2001 | A1 |
| 20010034727 | Ciurcina | Oct 2001 | A1 |
| 20010044743 | McKinley et al. | Nov 2001 | A1 |
| 20010044751 | Pugliese et al. | Nov 2001 | A1 |
| 20010044759 | Kutsumi et al. | Nov 2001 | A1 |
| 20010049632 | Rigole | Dec 2001 | A1 |
| 20010051893 | Hanai et al. | Dec 2001 | A1 |
| 20020002502 | Maes et al. | Jan 2002 | A1 |
| 20020004764 | Stolze et al. | Jan 2002 | A1 |
| 20020010645 | Hagen et al. | Jan 2002 | A1 |
| 20020036658 | Carolan et al. | Mar 2002 | A1 |
| 20020052767 | Yano | May 2002 | A1 |
| 20020055890 | Foley | May 2002 | A1 |
| 20020055906 | Katz et al. | May 2002 | A1 |
| 20020091990 | Little et al. | Jul 2002 | A1 |
| 20020107824 | Ahmed | Aug 2002 | A1 |
| 20030061202 | Coleman | Mar 2003 | A1 |
| 20030167222 | Mehrotra et al. | Sep 2003 | A1 |
| 20030191841 | DeFerranti et al. | Oct 2003 | A1 |
| 20040024656 | Coleman | Feb 2004 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20070208682 A1 | Sep 2007 | US |
| Number | Date | Country | |
|---|---|---|---|
| 60241541 | Oct 2000 | US | |
| 60219783 | Jul 2000 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 09909250 | Jul 2001 | US |
| Child | 11746767 | US | |
| Parent | 09764662 | Jan 2001 | US |
| Child | 09909250 | US |
