1. Field
The present application relates to systems and methods employing machines (expert systems) to aid in troubleshooting customer support issues.
2. Related Art
Knowledge systems have been created to solve specific problems, and provided through customer care tools to agents and end customers via web sites, mobile applications, printed material, etc.
It is a challenge to identify the knowledge that will provide a solution when only the symptoms are known. Often a search mechanism is available to find knowledge, but this doesn't assist with the diagnosis from the symptom, and often requires the user to have an understanding of the root cause of the problem being solved before the solution can be presented.
Other knowledge system and methods employ a question and answer model, allowing the customer to search for the question, rather than the answer. Some have a usage-based learning component that will optimize the search index for future searches.
Other systems and methods have included ‘case-based reasoning’ flows—which consist of a series of steps and decisions to identify the problem and offer the solution. These require the user to identify from a pre-defined list of symptoms, rather an accurately describing their exact symptoms. These are also constrained by scope and size, so often will not cover all the domains where a fault could occur—either within the product, the network or a third party application or service. They also have to be authored up-front, and are not generated and learnt automatically.
A knowledge interface is provided that interacts with a user-operated data processing system via networked communication to identify a solution to problem or issue experienced by a customer with respect to a particular product or service. The knowledge interface includes data processing functionality that supplies information to the user-operated data processing system. The information represents a number of components that are presented in at least one display window displayed by the user-operated data processing system. The components include:
i) first data identifying a set of predetermined symptoms linked to the problem or issue experienced by the customer;
ii) first interface elements that are configured to allow the user to classify the set of predetermined symptoms of i) into two classes including a first class of symptoms representing symptoms most likely experienced by the customer and a second class of symptoms representing symptoms most likely not experienced by the customer;
iii) second data identifying a set of predetermined root causes linked to the set of predetermined symptoms of i);
iv) second interface elements that are configured to allow the user to classify the set of predetermined root causes of iii) into two classes including a first class of root causes representing root causes most likely experienced by the customer and a second class of root causes representing root causes most likely not experienced by the customer; and
v) third data identifying a set of solutions linked to the set of predetermined root causes of iii), wherein the third data identifies a best solution based upon the predetermined root causes of iii) and their associated class designations as dictated by user input with the second interface elements.
The knowledge interface can further include a user input mechanism that is configured to enable the user to specify a natural text description of at least one symptom of the problem or issue experienced by the customer with respect to a particular product or service as well as an interface to an analysis engine that is configured to identify the set of predetermined symptoms. The analysis engine can employ statistical analysis to link the natural text description specified by user operation of the user input mechanism to the set of predetermined symptoms. In one embodiment, such statistical analysis implements a naive Bayes classification methodology. The statistical analysis can associates a confidence level with the link between the natural language textual description of the problem or issue and a given predetermined symptom. The confidence levels associated with the set of predetermined symptoms can be used by the knowledge interface to arrange the display order of the set of predetermined symptoms in the at least one display window.
Context that identifies the particular product or service can be supplied to the knowledge interface and/or can be otherwise known or derived. In one embodiment, such context is derived from interaction between a call center representative and the customer. In another embodiment, such context is supplied by input from the customer.
The components that are presented in at least one display window displayed by the user-operated data processing system can further include a user input mechanism to select a best solution and trigger the display of additional information regarding the best solution to the user (such as a document or other web content, an external link, an OTA flow for mobile device configuration and programming, device attributes, and a simulation that guides the call center representative through steps to fix a particular problem or issue). The components can further include a user input mechanism to indicate whether or not the best solution was successful in solving the problem or issue.
The knowledge interface can further include a database that stores collected data derived from interaction with the user and associated with the problem or issue. The collected data can be used to train the analysis engine for subsequent operations.
In one embodiment, the user of the knowledge interface is a call center representative.
In another embodiment, the user of the knowledge interface is the customer.
The knowledge interface can be configured such that first data, the second data and the third data are displayed in a plurality of distinct regions of a display window, wherein the plurality of regions are laid out adjacent one another across the horizontal extent of the display window. The plurality of regions can include a region that displays the first data and/or the second data after being classified in accordance with user input.
The present application also describes a method for identifying a solution to problem or issue experienced by a customer with respect to a particular product or service. The method includes supplying information to a user-operated data processing system via networked communication, the information representing a number of components that are presented in at least one display window displayed by the user-operated data processing system. The components include:
i) first data identifying a set of predetermined symptoms linked to the problem or issue experienced by the customer;
ii) first interface elements that are configured to allow the user to classify the set of predetermined symptoms of i) into two classes including a first class of symptoms representing symptoms most likely experienced by the customer and a second class of symptoms representing symptoms most likely not experienced by the customer;
iii) second data identifying a set of predetermined root causes linked to the set of predetermined symptoms of i);
iv) second interface elements that are configured to allow the user to classify the set of predetermined root causes of iii) into two classes including a first class of root causes representing root causes most likely experienced by the customer and a second class of root causes representing root causes most likely not experienced by the customer, and
v) third data identifying a set of solutions linked to the set of predetermined root causes of iii), wherein the third data identifies a best solution based upon the predetermined root causes of iii) and their associated class designations as dictated by user input with the second interface elements.
Additional details with respect to the method are disclosed and claimed.
In another aspect, a knowledge interface is described that interacts with a user to identify a solution to problem or issue experienced by a customer with respect to a particular product or service. The knowledge interface includes data processing functionality that presents a number of components in at least one display window displayed to the user. The components include:
i) first data identifying a set of predetermined symptoms linked to the problem or issue experienced by the customer;
ii) first interface elements that are configured to allow the user to classify the set of predetermined symptoms of i) into two classes including a first class of symptoms representing symptoms most likely experienced by the customer and a second class of symptoms representing symptoms most likely not experienced by the customer;
iii) second data identifying a set of predetermined root causes linked to the set of predetermined symptoms of i);
iv) second interface elements that are configured to allow the user to classify the set of predetermined root causes of iii) into two classes including a first class of root causes representing root causes most likely experienced by the customer and a second class of root causes representing root causes most likely not experienced by the customer; and
v) third data identifying a set of solutions linked to the set of predetermined root causes of iii), wherein the third data identifies a best solution based upon the predetermined root causes of iii) and their associated class designations as dictated by user input with the second interface elements.
Additional details with respect to the knowledge interface are disclosed and claimed.
In accordance with the present application, a system is provided for diagnosing and fixing problems or issues with a consumer product or service. The product or service can be electronic devices (such as a smart phone, personal digital assistant, tablet computer, notebook computer desktop computer, e-book reader, GPS device, mobile device, game console, set-top box, smart TV, digital video recorder, network router/gateway, automotive system), software products or services or other technical products or services. The problems or issues with such products and services are often difficult to identify and fix due to the complexity of such products or services. An additional complicating factor is the context in which it is used, which often involves connectivity to a local area network and the Internet through mechanisms that include GSM, CDMA, iDEN, WiMax, WiFi, LTE, Ethernet, Bluetooth, and a wide area network (ADSL, Cable, Fiber, etc.). The root cause of the problem or issue may lie with the product or service itself, with the network and its setup/provisioning, a technical fault elsewhere, or with a third party application or service, either running locally on the device (e.g. a smart phone application) or remotely (e.g. a cloud-based service). The requisite knowledge of all these relevant domains is often beyond the level attained by a call center representatives (sometimes referred to as customer support representative or agents) or by the end-user customer, resulting in a frustrating and time-consuming support experience, often with multiple calls for support to resolve the problem or issue.
The system of the present application can be configured to allow the symptoms of the problem or issue to be described in a natural language context, and thus avoids the need to translate the problem or issue that is being experienced to a list of pre-categorized symptoms or problems.
The system of the present application can also be configured to learn, through usage, the terminology that is used by customers and support staff to describe the symptoms of the problem or issue that is being experienced, and the subsequent mapping to a standardized list of symptoms.
The system of the present application can also be configured to learn the symptoms that are exhibited by a root cause problem or issue, so that future diagnosis of the same symptoms can be carried out in an efficient and effective manner.
The system of the present application can also be configured to learn which solutions are effective in solving the root cause of a problem or issue for certain exhibited symptoms.
The system of the present application can also be configured to deliver a wide variety of types of solutions (such as knowledge articles, interactive tutorials or automated fix processes, provisioning mechanisms or processes, including ‘off-line’ or manual processes).
The system of the present application can enable a novice end-user or new call center representatives to quickly become proficient at problem solving, by offering a step-by-step assisted diagnosis of a problem or issue.
The system of the present application can enable experienced call center representatives to quickly access the appropriate solution to a problem or issue through a natural language expression of the symptom(s) exhibited by the problem or issue, with the ability to skip step-by-step assisted diagnosis of the solution if desired.
The system of the present application can also be configured to track and record all troubleshooting steps and evidence associated with a problem or issue by users of the system. This feature can be useful for follow-up interactions with the same user (to avoid repeating the same steps) and also in performance monitoring.
The system of the present application can also be configured to track and records the magnitude/volume of problems and issues occurring for a particular product or service which can be used to inform the creation of new knowledge, update the type of product or another element (e.g. firmware update) to fix the problem, or to supply an appropriate solution in subsequent operations of the system.
The system of the present application can identify gaps in knowledge where no matching symptoms, route causes or solutions are available for an expression of a symptom.
The system of the present application can be configured to learn from users the solution to the described symptoms, and can apply this learning in other channels (e.g. self-service website or mobile application) for end-user customers to solve problems autonomously, without the need to contact a call center representative or other support function.
The system of the present application can be configured to tap into text streams where symptoms are being expressed, and to continually disambiguate to offer and refine the proposed solution. Text streams could include a chat session with customer care, a real-time voice analysis of the support call audio, a forum posting, a social network post (e.g. a Twitter message), a blog post or instant message.
The system of the present application can be configured to integrate to third party speech analytics platforms and APIs to enable a user to vocally describe a set of symptoms for input to the interface, and to have a solution proposed as a response to the voice input.
The customer service representative stations 5A, 5B are typically realized by a browser-based desktop interface served by customer relationship management platform 3 that allows a respective call center representative to log into the platform 3, to view a work queue for the call center representative as managed by the platform 3, and to provide notification that the call center representative is available to service incoming customer interactions. The browser-based interface can be adapted to allow the call center representative to manage multiple simultaneous customer service interactions (such as combinations of emails, phone calls, and Web calls). The browser-based interface can be adapted to allow the respective call center representative to accept an incoming customer interaction, release a customer interaction, or transfer a customer interaction. The browser-based interface can also be adapted to provide the respective call center representative with information about a particular customer interaction, such as how long the particular customer interaction has been assigned to the respective call center representative, and/or how long the respective call center representative has been working on the particular customer interaction, and/or customer data such as name, address, and contact information. The customer relationship management platform 3 can be realized by a wide variety of commercially-available call center platform solutions, including products from Cisco, Siebel, Amdocs, Salesforce.com, Microsoft and Oracle.
The customers can employ telephony devices (such as a mobile device 11 or a landline telephone 13) to place calls to the customer relationship management platform 3 over a radio access network/telephony network 15 for routing to a customer service representative. The customers can also employ mobile devices (such as mobile device 7) or other computer systems or communication devices (such as computer 17) to initiate other forms of communication (such as email, fax, SMS, Web collaboration, Social Media Messaging, and/or chat) to the customer relationship management platform 3 via the radio access network/telephony network 15 and/or Internet 9 for routing to a customer service representative.
The system 1 also includes a knowledge interface 19 that is operably coupled to the call center representative stations (for example the two shown as 5A, 5B) over the private network 7 or the public network 9 or a combination of both as shown. The knowledge interface 19 interacts with call center representatives via operation of the call center representative stations 5A, 5B to identify solutions to problems or issues experienced by customers in operating particular products or services. In this context, a particular product or service is identified by the customer or otherwise known. In one example, the context can be derived from interaction between the customer and a call center representative managed by the customer relationship management platform 3 where the call center representative is servicing a voice call, instant message, email or other communication from the customer who is experiencing the problem or issue in operating the product or service of interest. Such context (more specifically, data that identifies the particular device or equipment of interest) is passed by messaging or another suitable communication interface to the knowledge interface 19.
The knowledge interface 19 receives and processes the context to dynamically generate and supply information to the call center representative station 5 where the information is configured to initially present a display window (and subsequently update the display window) on the call center representative station 5. The knowledge interface 19 can be realized by application server and middleware software functionality executing on a suitable data processing platform. The data processing platform can be a single machine or distributed over multiple machines if desired. The information communicated between the knowledge interface 19 and the call center representative station 5 can include a wide variety of common web data types, such as HTML code, style sheets, scripts (such as javascripts and PHP scripts), XML documents and web pages and forms (such as ASP.net web pages and forms). The data types can include textual information, graphical information, and/or multimedia information (such as video files and/or audio files). The display window presented on the call center representative station 5 has an initial view that identifies the product or service of interest as dictated by the context. An example of the initial view of such display window 201 is shown in
The system 1 also includes an analysis engine 21 operably coupled to the knowledge interface 19 by messaging or another suitable communication interface. The analysis engine 21 can also be realized by application server and middleware software functionality executing on a suitable data processing platform. The data processing platform can be a single machine or distributed over multiple machines if desired. The knowledge interface 19 is configured to pass the natural language textual description of the symptom as well as data identifying the product or service of interest to the analysis engine 21. The analysis engine 21 is configured to receive and process the natural language textual description of the symptom as well as the data identifying the product or service of interest in order to link the natural language textual description of the symptom for the product or service of interest to a set of zero or more predetermined symptoms. The analysis engine 21 can associate a confidence level with the link between the natural language textual description of the symptom for the product or service of interest and a given predetermined symptom. The linking and associated confidence levels can be based on statistical analysis.
In one embodiment, the statistical analysis of the analysis engine 21 implements a naive Bayes classification methodology that uses Bayesian theory, which provides an equation for deriving the probability of a prediction based on a set of underlying evidence. The naive Bayes classification methodology makes a simplifying assumption that the pieces of evidence are not interrelated in a particular way. This assumption is what is called the naive aspect of the algorithm (here, “naive” is a technical term, not a disparagement). The naive Bayes classification methodology employs a prediction model with parameters that are generated from a training set as is well understood in the data analysis arts. It should be understood that other suitable statistical analysis methodologies can also be used by the analysis engine 21.
In conjunction with the operation of the analysis engine 21, the knowledge interface 19 defines four regions—Region 1 (labeled 221A), Region 2 (labeled 221B), Region 3 (labeled 221C), and Region 4 (labeled 221D)—of the display window 201 that is presented on the call center representative station 5 for display to the call center representative. These four regions 221A, 221B, 221C, 221D are preferably laid out adjacent one another across the horizontal extent of the display window 201 as shown in
The knowledge interface 19 is configured such that Region 1 (labeled 221A) of the display window 201 depicts evidentiary classifications for one or more symptoms and root causes as identified by interaction of the call center representative with the information depicted in Regions 2 and 3. The evidentiary classifications include two classes: Class A and Class B. Class A includes symptom(s) and/or root cause(s) that are most likely relevant to the problem or issue experienced by the customer. Typically, the class A symptom(s) and/or root cause(s) have been experienced by the customer or are currently being experienced by the customer. Class B includes symptom(s) and/or root cause(s) that are most likely not relevant to the problem or issue experienced by the customer. Typically, the class B symptom(s) and/or root cause(s) have not been experienced by the customer or are currently not being experienced by the customer. The classification of a given symptom or root cause is dictated by interaction of the call center representative with the information depicted in Region 2 (labeled 221B) and Region 3 (labeled 221C) as described herein. The evidentiary classifications for classes A and B is initially set to null so that there is no symptoms and root causes initially depicted in Region 1 for classes A and B. When depicting more than one class A symptom in Region 1, the top-to-bottom order of the class A symptoms as displayed in Region 1 can be based on the confidence levels of the symptoms such that the more-likely class A symptoms are depicted above the less-likely class A symptoms. If the number of class A symptoms is large and cannot fully be depicted in Region 1, a slider bar or other suitable interface mechanism can be used to provide the user access to all of the class A symptoms of the set. Similarly, when depicting more than one class B symptom in Region 1, the top-to-bottom order of the class B symptoms as displayed in Region 1 can be based on the confidence levels of the symptoms such that the more-likely class B symptoms are depicted above the less-likely class B symptoms. If the number if class B symptoms is large and cannot fully be depicted in Region 1, a slider bar or other suitable interface mechanism can be used to provide the user access to all of the class B symptoms of the set.
The knowledge interface 19 is further configured such that Region 2 (labeled 221B) of the display window 201 initially depicts the set of predetermined symptoms linked to the natural language textual description of the symptom for the product or service of interest by the analysis engine 21. When depicting more than one symptom in Region 2, the top-to-bottom order of the symptoms as displayed in Region 2 can be based on the confidence levels of the symptoms such that the more-likely symptoms are depicted above the less-likely symptoms of the set. If the set of symptoms is large and cannot fully be depicted in Region 2, a slider bar or other suitable interface mechanism can be used to provide the user access to all of the symptoms of the set.
The knowledge interface 19 is further configured such that Region 3 (labeled 221C) of the display window 201 depicts a set of root causes that may exhibit the predetermined symptoms depicted in Region 2. The association between the predetermined symptoms and the root causes can be based on an acyclic directed graph constructed by expert knowledge. Initially, Region 3 depicts a set of root causes for the predetermined symptoms depicted in Region 2. When displaying more than one root cause in Region 3, the top-to-bottom order of the root causes as displayed in Region 3 can be based on the confidence levels of the associated symptoms such that the more-likely root causes are depicted above the less-likely root causes for the problem or issue experienced by the customer.
The knowledge interface 19 is further configured such that Region 4 (labeled 221D) of the display window 201 depicts a set of solutions that solves the root causes depicted in Region 3. The association between the root causes and the solutions can be based on an acyclic directed graph constructed by expert knowledge. Initially, Region 4 depicts a set of solutions associated with the root causes depicted in Region 3. When displaying more than one solution in Region 4, the top-to-bottom order of the solutions follows the ordering of the root causes displayed in Region 3 and thus more-likely solutions are depicted above the less-likely root causes for the problem or issue experienced by the customer. Note that in any point in the process, the user is able to view the set of solutions depicted in Region 4, and can select any one of the solutions to access the corresponding additional information (such as fix information specific to the solution).
Each symptom depicted in Region 2 can include one or more interface widgets (such as a tick widget 223A or cross widget 223B as shown in
Upon classification of a given symptom, the knowledge interface 19 is configured such that the display of Region 2 is updated to remove the given symptom, and the display of Region 1 is updated to display the given symptom according to the call center representative's classification as evident from
When the call center representative classifies a symptom of Region 2 as belonging to evidentiary class A (or possibly to evidentiary class B), the knowledge interface 19 is configured to update the root causes depicted in Region 3 to identify those root causes that are related to the symptoms of evidentiary class A as depicted in Region 1. This is shown in
Each root cause depicted in Region 3 can include one or more interface widgets (such as a tick widget 225A or cross widget 225B as shown in
Upon classification of a given root cause, the knowledge interface 19 is configured such that the display of Region 3 is updated to remove the given root cause, and the display of Region 1 is updated to display the given root according to the call center representative's classification as evident from
When the call center representative classifies a symptom or root cause as belonging to evidentiary class A (or possibly to evidentiary class B), the knowledge interface 19 is configured to update the solutions depicted in Region 4 of the display window to identify the best solution for the set of evidentiary class A symptom(s) and root cause(s) depicted in Region 1 as well as for the evidentiary class B symptom(s) and root cause(s) depicted in Region 1 (if any). This is shown in
The best solution can be identified as being the solution with the highest confidence level, where the confidence level for each solution is calculated using the confidence level(s) of the root cause(s) belonging to evidentiary class A where the root cause shares a relationship with the solution. One or more acyclic directed graphs can be used to define relationships (i.e., associations) between symptoms, root causes and solutions. Such acyclic directed graph(s) can be derived from expert knowledge and updated according to the operations of the system.
It is contemplated that the call center representative user may move a symptom or route cause between the evidentiary classes A and B as depicted in Region 1, which will update the root cause and solution regions accordingly. It is also contemplated that the call center representative user may remove (declassify) a symptom or route cause from both the evidentiary classes A and B as depicted in Region 1, which will update the root cause and solution regions accordingly.
The operations of the knowledge interface 19 allows the call center representative to traverse through the symptoms that are potentially relevant to the problem or issue experienced by the customer and identify those symptoms that are relevant to the problem or issue experienced by the customer (and also possibly identify those symptoms that are not relevant to the problem or issue experienced by the customer). For some symptoms, this can involve querying the customer to gather additional information. The process also allows the call center representative to traverse through root causes that are potentially relevant to the problem or issue experienced by the customer and identify those root causes that are relevant to the problem or issue experienced by the customer (and also possibly identify those root causes that are not relevant to the problem or issue experienced by the customer). Such user-identified information is used to identify the best solution for the problem or issue experienced by the customer. This solution is presented to the call center representative in order to attempt to fix the problem or issue.
The depiction of the best solution in Region 4 can provide a user input mechanism to access additional information regarding the best solution. The additional information can be a document or other web content, an external link, an OTA flow for mobile device configuration and programming, device attributes, and/or a simulator that guides the call center representative through steps to fix a particular problem or issue. For example, it is contemplated that the call center representative can click on the depiction of the best solution in Region 4 to display a window that displays detailed instructions 226 for carrying out the best solution as well as feedback widgets 227A, 227B indicating whether or not the solution was successful as shown in
Such interaction can be used to store the user-identified evidence (the natural language textual description of the symptom(s), the user identified symptom(s), the ruled-out symptom(s), the user-identified root cause(s), the ruled-out root cause(s), the best solution and the ruled-out solution(s)) in a database as depicted in the call log of
In alternate embodiments, it is contemplated that the natural language textual description of the symptom can be supplied by other mechanisms, such as in an email or instant message communication from the customer. In yet other embodiments, the processing of the knowledge interface 19 and the analysis engine 21 can be integrated into the hardware and software functionality of the customer relationship management platform 3.
It is also contemplated that context can be derived from interaction between the customer and a self-service kiosk, a mobile application, a web site or other suitable user interface. In such applications, the processing of the knowledge interface 19 and the analysis engine 21 can allow for customers interactions via the self-service kiosk, the mobile application, the web site or the other suitable user interface as shown in the system 100 of
The system of the present application has the potential to change the paradigm of technical support call centers by making call center representatives effective much more quickly (faster ‘onboarding’), with reduced training. It can provide a higher level of first call resolution (FCR), and reduce the impact of attrition—as there is no longer a reliance on the knowledge and troubleshooting awareness in the representative's head.
Additionally, the system of the present application can allow customer interactions of a more technical nature to be handled by a level 1 resource, which is typically a lower cost than a level 2 technical environment.
Furthermore, the system of the present application can significantly improve the capability of self-care web sites, enabling a much higher call deflection rate, lowering the total cost to support complex electronic and technical products on behalf of wireless network operators, device manufacturers, service providers, and other entities that today provide support for these products.
There have been described and illustrated herein several embodiments of a system and method to identify a solution to problem or issue experienced by a customer with respect to a particular product or service. While particular embodiments of the invention have been described, it is not intended that the invention be limited thereto, as it is intended that the invention be as broad in scope as the art will allow and that the specification be read likewise. Thus, while particular display schemes and interface elements have been disclosed, it will be appreciated that other display schemes and interface elements as well. In addition, while particular types of statistical analysis methodologies have been disclosed, it will be understood that other suitable statistical analysis methodologies can be used. Moreover, while particular system configurations, architectures and corresponding methodologies have been disclosed, it will be appreciated that other system configurations, architectures and methodologies could be used as well. It will therefore be appreciated by those skilled in the art that yet other modifications could be made to the provided invention without deviating from its spirit and scope as claimed.