System and method for providing online customer support

Abstract

The invention discloses a system and a method for implementing a customer support interface. The method includes generating the customer support interface for a customer to perform a chat session with a chat robot, wherein an option to perform a chat session with an artificial agent is made unavailable on the customer support interface. The method further includes receiving, by the chat robot, a first query from the customer; and determining, by the chat robot, a first reply to the first query. The first reply is generated for display on the customer support interface. After generating the first reply, the method uses a prediction engine to determine an overall probability value that provides the customer with the option to make a chat session with a human agent will improve the likelihood that the customer will obtain a positive result, and determines whether the overall probability value satisfies a predetermined condition.

Description

Technical Field

[0001] The present invention generally relates to managing customer support resources and providing customer support. In particular, the present invention relates to systems and methods for implementing an online customer support interface. Background Art

[0002] Live chat systems have become an essential communication tool for businesses to interact with online customers and resolve their queries or complaints in real time. However, due to the high cost of hiring customer service agents to provide live chat services, most businesses cannot afford to retain enough staff 24 / 7 to always immediately attend to every customer using the live chat service. This inevitably leads to dissatisfied customers who have to wait a long time for a customer service agent to attend to them.

[0003] As artificial intelligence (AI) technology becomes more prevalent, businesses are increasingly looking to supplement their live chat services with automated customer support systems managed by chatbots to reduce the need for live chat services. Chatbots can be programmed to perform a wide variety of tasks, from answering common questions to making appointments. Chatbots help reduce or divert the need for live chat services by providing customers with self-service options to get answers or perform simple transactions without having to interact with a customer service agent.

[0004] Ideally, chatbots serve as the first level of customer engagement and support, used to divert customers with simple queries (e.g., FAQs) or basic requests (e.g., making reservations) from live chat services, thereby allowing customer service agents to focus on customers with complex queries that chatbots cannot adequately resolve. However, customer service chatbots also have some disadvantages. In one example, chatbots lack understanding of human emotions and intent, and may not be able to provide meaningful responses to customers' queries (e.g., questions or requests). In another example, the answers or responses that chatbots can provide to customers are often limited by the information stored inside the system that interfaces with the chatbot. If the stored information is insufficient, the chatbot may not be able to properly assist the customer. This can cause frustration and dissatisfaction among customers when interacting with the chatbot. Therefore, many customers choose to bypass chatbot interactions and contact customer service agents directly via live chat services at the beginning of (or during) a chat session, regardless of whether their queries or requests can be resolved by the chatbot. This prevents chatbots from achieving their intended purpose. In addition, this will put more pressure on customer service agents with limited resources. Therefore, a balance needs to be struck when using chatbots and customer service agents to handle customer queries. Therefore, there is a need to minimize the number of interactions between customers and customer service agents without compromising the customer’s chances of a positive outcome.

[0005] Therefore, there is a need for an improved system and / or method for providing online customer support that solves the above-mentioned problems by selectively hiding the live chat option at the beginning of an automated chat session, and selectively presenting (or displaying) the hidden live chat option only when it is predicted that human intervention will increase the likelihood of a positive outcome to the chat session, thereby optimizing the use of automated chatbot system and human agent resources. Moreover, other desirable features and characteristics will become apparent from the subsequent detailed description and the appended claims, taken in conjunction with the accompanying drawings and this background of the disclosure. Summary of the invention

[0006] According to a first aspect of the present invention, a method for implementing a customer support interface is disclosed. The method includes generating the customer support interface for a customer to conduct a chat session with a chatbot, wherein an option to conduct a chat session with a human agent is made unavailable on the customer support interface. The method also includes: receiving, by the chatbot, a first query from the customer; and determining, by the chatbot, a first reply to the first query and generating the first reply for display on the customer support interface. After generating the first reply, the method uses a prediction engine to determine an overall probability value that providing the customer with the option to conduct a chat session with a human agent will increase the likelihood that the customer will obtain a positive outcome, and determines whether the overall probability value satisfies a predetermined condition. In response to determining that the overall probability value satisfies the predetermined condition, the method makes an optional interface element representing the option to conduct a chat session with a human agent available on the customer support interface, wherein selecting the optional interface element transfers the customer from the chat session with the chatbot to the chat session with the human agent.

[0007] According to some aspects of the subject technology, the method includes determining that providing the customer with the option to conduct a chat session with the human agent will increase the likelihood that the customer will obtain a positive outcome, including using the prediction engine to analyze one or more feature data associated with the customer and calculate a first conditional probability value and a second conditional probability value, wherein the first conditional probability value indicates the likelihood that the customer will obtain the positive outcome with human intervention and the second conditional probability value indicates the likelihood that the customer will obtain the positive outcome without human intervention. The method also includes calculating the overall probability value by subtracting the second conditional probability value from the first conditional probability value.

[0008] According to some aspects of the subject technology, the method includes analyzing the one or more characteristic data associated with the customer including analyzing the first response to determine a degree of effectiveness of the first response based on the first query.

[0009] According to some aspects of the subject technology, the prediction engine implements a boosting model to determine the overall probability value.

[0010] According to some aspects of the subject technology, the method includes determining whether the overall probability value satisfies a predetermined condition including determining whether the overall probability value is greater than a threshold.

[0011] According to some aspects of the subject technology, if the overall probability value is not greater than the threshold, the option to conduct a chat session with a human agent remains unavailable to the customer.

[0012] According to some aspects of the subject technology, the option to conduct a chat session with a human agent is made unavailable on the customer support interface by visually hiding the selectable interface element from the customer.

[0013] According to some aspects of the subject technology, the option to conduct a chat session with a live agent is made unavailable on the customer support interface by disabling the selectable interface element.

[0014] According to some aspects of the subject technology, the first reply is displayed and the selectable interface element is simultaneously made available on the customer support interface.

[0015] According to some aspects of the subject technology, the selectable interface element is made available on the customer support interface after or shortly after displaying the first reply.

[0016] According to some aspects of the subject technology, the selectable interface element is made available as part of the first reply.

[0017] According to some aspects of the subject technology, the method also includes executing a rule-based system prior to initiation of the chat session to determine whether to make a live chat option available to the customer at the initiation of the chat session, and using the rule-based system to determine to make the option to conduct a chat session with a live agent unavailable on the customer support interface.

[0018] According to some aspects of the subject technology, the method also includes executing the rule-based system after generating the first reply, wherein the rule-based system includes a condition that depends on a number of interactions between the customer and the chatbot in the chat session.

[0019] According to some aspects of the subject technology, the rule-based system obtains data from a customer account database and / or a historical chat database.

[0020] According to some aspects, the present invention provides a system comprising one or more computers and one or more storage devices storing computer-readable instructions, which when executed by the one or more computers cause the one or more computers to perform the steps of any of the above methods. The same applies to one or more non-transitory computer storage media storing instructions, which when executed by one or more computers cause the one or more computers to perform the steps of any of the above methods. Therefore, the method can be easily transferred to other systems and executed there. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings, in which like reference numerals refer to the same or functionally similar elements throughout the individual views and which are incorporated into and constitute a part of the specification together with the following detailed description, are merely used to illustrate various embodiments by way of example and to explain various principles and advantages according to the present embodiments.

[0022] Figure 1 A simplified diagram of a system implementing an online customer support interface is shown in accordance with some embodiments.

[0023] Figure 2a to Figure 2c A flow chart of a method for implementing an online customer support interface according to some embodiments is shown.

[0024] Figure 3a to Figure 3c An exemplary implementation of an online customer support interface according to certain embodiments is shown.

[0025] Skilled technicians will understand that the elements in the figures are illustrated for simplicity and clarity and are not necessarily drawn to scale. For example, the size of some elements in the block diagram or the steps in the flow chart may be exaggerated relative to other elements to help improve the understanding of the present embodiment. DETAILED DESCRIPTION

[0026] The following detailed description is exemplary only and is not intended to limit the invention or the application and uses of the invention. Furthermore, there is no intention to be bound by any theory presented in the preceding background of the invention or the following detailed description.

[0027] Certain embodiments are intended to teach systems and methods for implementing a customer support interface. The customer support interface is provided by a business entity (e.g., an e-commerce platform, a financial services company, a gaming company, etc.) as a means of providing customer support to its online customers, including but not limited to responding to customer queries in real time or providing information to customers. The customer support interface is configured to receive customer input queries (e.g., questions or requests) from customers seeking customer support (e.g., customer support requests) from the business entity, and transmit the customer input queries to an automated chatbot system for generating responses. For example, the customer support interface includes a chat feature that allows an online customer to enter a query, and the chatbot will provide a response to the query.

[0028] In some embodiments, the chatbot is configured to initiate a conversation session (or chat session) with an online customer, while retaining (i.e., making unavailable) the option (e.g., a live chat option) for the customer to transfer the customer's chat session to a customer service agent (hereinafter referred to as a "human agent") when generating and displaying a customer support interface. After receiving a first customer input query from a customer, the chatbot generates a first reply to the first customer input query (without human intervention) for display on the customer support interface. After generating the first reply, the chatbot obtains a first input data set and feeds the first input data set to a prediction engine to determine the overall probability (or likelihood) that human intervention (e.g., assistance from a human agent) can increase the likelihood that the customer will obtain a positive result. For example, a positive result refers to the resolution of the customer's query. The prediction engine may consider the degree of effectiveness of the machine-generated first reply when determining the overall probability.

[0029] In some embodiments, when the prediction engine determines that the overall probability (human intervention can improve the likelihood that the customer will receive a positive outcome) is greater than a threshold, an option to transfer the customer to a human agent will be provided on the customer support interface. For example, the live chat option (via a selectable interface element) may be displayed simultaneously with the machine-generated response, or shortly after the machine-generated response is displayed, or thereafter. At this point, if the customer is not satisfied with the first machine-generated response, the customer can select the live chat option and be transferred to a human agent for further assistance.

[0030] In some embodiments, when the prediction engine determines that the overall probability (human intervention can improve the likelihood of a positive outcome for the customer) is not greater than a threshold, the option to transfer the customer (or chat session) to a human agent continues to be reserved for the customer. For example, the first machine-generated reply is displayed, and the live chat option remains unavailable (e.g., hidden or disabled) on the customer support interface, so that the customer cannot select it. At this point, the customer can continue to interact with the chatbot by providing a second customer input query. The chatbot receives the second customer input query from the customer and generates a second reply (without human intervention) for display on the customer support interface. After generating the second reply, the chatbot obtains a second input data set and feeds the second input data set to the prediction engine to determine the overall probability (or likelihood) that human intervention can improve the likelihood of a positive outcome for the customer, and the cycle repeats each time the chatbot generates a reply to the corresponding query. The prediction engine may consider the degree of effectiveness of the second machine-generated reply when determining the overall probability.

[0031] By retaining the live chat option for customers from the beginning of the chat session, the present invention prevents customers from bypassing chatbots and ensures that chatbots effectively serve as first-level customer engagement and support while saving limited human agent resources. At the same time, the present invention monitors the interactions of customers with chatbots in real time and uses these real-time interactions as input features to enrich the input data set for the prediction engine to determine the overall probability that human intervention can improve the likelihood of a positive outcome for the customer. For example, during the process of determining the overall probability that human intervention can improve the likelihood of a positive outcome for the customer, the prediction engine can consider the effectiveness of the machine-generated reply based on the corresponding customer input query. The prediction engine is configured to selectively provide the live chat option only when the overall probability is greater than a threshold or within a threshold range. Therefore, the live chat option is available to customers (or customer groups) only when it is predicted that they will benefit most from interacting with human agents. This optimizes the allocation of available human agents (which are limited resources) and also increases the chance or rate of customers obtaining positive results.

[0032] On the other hand, when it is predicted that other customers will have little or no benefit from interacting with human customer service, the live chat option is reserved for other customers to encourage other customers to interact with the chatbot to resolve their queries. Therefore, the chatbot's engagement rate (e.g., the percentage of customers who interact with the chatbot) and turn rate (e.g., the percentage of customer queries resolved by the chatbot without any human intervention) are improved. In addition, by configuring the live chat option to be displayed simultaneously with (or shortly after or after) displaying the machine-generated reply, the machine-generated reply has the opportunity to resolve the customer's query before the customer selects (or is presented with) the live chat option. This can further increase the chatbot's turn rate. The present invention also advantageously balances increasing the turn rate with increasing customer satisfaction by running a prediction engine during a chat session to monitor the customer's interaction with the chatbot and dynamically determining when the live chat option should be provided to the customer (e.g., when it is predicted that human intervention can improve the customer's results).

[0033] Figure 1 A simplified diagram of a system 100 implementing an online customer support interface according to some embodiments is shown. The system 100 may include one or more components, modules, or layers, which may be implemented using software and / or hardware, optionally across multiple locations or using multiple devices or units. Figure 1 An overview of the system 100 is described. Thereafter, a more detailed description of the system 100 is provided with respect to the remaining figures.

[0034] The system 100 may be implemented as a single system, multiple systems, a distributed system, or any other form. Each component of the system 100 may be any type of computing device suitable for performing the functions required by one or more embodiments of the present invention and capable of communicating with each other directly or through a network 110. In some embodiments, the system 100 includes a client device 120, an agent device 130, and a server 140 that are communicatively coupled to each other via the network 110.

[0035] Figure 1 The network 110 shown in FIG. 1 may include any number of suitable networks and / or network types. For example, the network 110 may include, but is not limited to, a virtual private network (VPN), a local area network (LAN), a wide area network (WAN) (such as the Internet), a wireless wide area network (WWAN), a metropolitan area network (MAN), a Wi-Fi network, network, One or more of a (Bluetooth) communication network, a peer-to-peer network, a cellular telephone network, or any combination of such networks. The network 110 may include any combination of connections (e.g., wired, wireless, or fiber optic connections) and protocols that will support communication between each component in the system 100 according to one or more embodiments of the present invention. For example, data transmitted by the network 110 may include packetized data (e.g., Internet Protocol (IP) data packets) that are routed to a designated computing device coupled to the network 110. The network 110 may include networking software and / or hardware that interconnects the computing devices of the system 100. For example, the network 110 may include a packet switching device that routes data packets to various computing devices based on information included in the headers of the data packets.

[0036] Data exchanged over the network 110 may be represented using techniques and / or formats including Hypertext Markup Language (HTML), Extensible Markup Language (XML), JavaScript Object Notation (JSON), Comma Separated Values ​​(CSV), Java Database Connectivity (JDBC), Open Database Connectivity (ODBC), etc. In addition, data exchanged over the network may be encrypted using encryption techniques, such as Secure Sockets Layer (SSL), Secure HTTP (HTTPS), and / or Virtual Private Network (VPN) or Internet Protocol Security (IPsec).

[0037] It should be appreciated that custom and / or dedicated data communications techniques may be used in addition to or in addition to the foregoing to perform the functionality required by one or more embodiments of the present invention.

[0038] Figure 1 The server 140 shown in the figure may take the form of one or more web-based server computers, such as web servers, application servers, database servers, etc., which can communicate with client devices 120 and agent devices 130 through network 110. For example, server 140 may be embodied in a single server computer, or distributed as multiple interconnected server computers, and operate in a cloud computing environment. Server 140 may include a processor and a memory. The memory may be a non-transient computer-readable medium of volatile or non-volatile type, such as a magnetic medium, an optical medium, a random access memory (RAM), a read-only memory (ROM), a removable medium, or any other suitable local or remote memory unit. The memory may include digital data accessible to the processor. The processor may be any type of device or multiple devices capable of manipulating or processing information from a non-transient computer-readable medium (e.g., a memory). For example, the processor includes a central processing unit (CPU) for processing computer executable program instructions to perform one or more processes.

[0039] Server 140 may include one or more storage devices that store data structures and / or computer instructions for execution by a processor. For example, server 140 may include one or more computer-readable data storage devices to support the functions required by one or more embodiments of the present invention. Various types of computing devices may be used for server 140. For example, the computer may be a mainframe, a workstation, and other types of processing devices. The memory may include digital data accessible to the processor. Server 140 and / or the data storage device may be coupled to various databases or storage devices in the form of any suitable computer-readable medium, such as a hard drive, a memory device, a flash drive, or an optical drive.

[0040] In some embodiments, an e-commerce platform (not shown) may be implemented in whole or in part on server 140. The e-commerce platform may provide merchants with online resources and facilities to manage their business. Merchants may utilize the e-commerce platform to manage commerce with customers, such as by implementing an e-commerce experience with customers via an online store. An online store may refer to a multi-tenant storefront collection (e.g., for multiple merchants) or a storefront of a separate merchant (e.g., an online store of a merchant) supported by an e-commerce platform. For example, server 140 may support a single integrated website (also referred to as a multi-merchant website) that serves as a virtual shopping mall. In some embodiments, the e-commerce platform provides online customer support to customers, such as by implementing an automatic chatbot system and a real-time chat system. In some embodiments, the e-commerce platform provides an online store to customers who have registered an account on the e-commerce platform, as will be described in more detail below.

[0041] Figure 1 The client devices 120 shown in the figure represent any number of client devices that may be included in the system 100. Each client device 120 is a computing device with data processing and communication capabilities. In some embodiments, the client device 120 may include a processor (e.g., CPU), memory (e.g., RAM, ROM, etc.), a power supply, a network interface, and / or other software and / or hardware components, such as a display, a graphics processor, a wireless transceiver, a keyboard, a speaker, a camera, a sensor, firmware, an operating system, a driver, various physical connection interfaces (e.g., USB, HDMI, etc.). For example, the client device 120 may take the form of one or more personal computers, tablet computers, smart phones, home entertainment computers, network computing devices, mobile computing devices, personal digital assistants, virtual machines, servers, other computing devices, or any combination of such computing devices.

[0042] In some embodiments, the client device 120 may be associated with a client 124 of the e-commerce platform. For example, the client 124 may have a client account registered on the e-commerce platform. The client 124 interacts (or interfaces) with the system 100 using the client device 120 associated with the particular client. Figure 1 The customer 124 shown in FIG. 1 represents any number of customers that may be included in the system 100. The customer device 120 includes a display for displaying the customer support interface 122, as further described herein and with respect to FIG. Figure 2a-2c and Figure 3a-3c described.

[0043] Figure 1 The proxy device 130 shown in the figure represents any number of proxy devices that may be included in the system 100. Each proxy device 130 is a computing device with data processing and communication capabilities. In some embodiments, the proxy device 130 may include a processor (e.g., virtual, physical, etc.), a memory, a power supply, a network interface, and / or other software and / or hardware components, such as a display, a graphics processor, a wireless transceiver, a keyboard, a speaker, a camera, a sensor, firmware, an operating system, a driver / driver, various physical connection interfaces (e.g., USB, HDMI, etc.). For example, the proxy device 130 may take the form of one or more personal computers, tablet computers, smart phones, home entertainment computers, network computing devices, mobile computing devices, personal digital assistants, virtual machines, servers, other computing devices, or any combination of such computing devices.

[0044] In some embodiments, the agent device 130 may be associated with a human customer service agent (or human agent) 134 employed by the e-commerce platform. For example, the human agent 134 operates a real-time chat service provided to the e-commerce platform's customers 124. The human agent 134 interacts (or interfaces) with the system 100 using the agent device 130 associated with the particular human agent. Figure 1 The human agent 134 shown in the figure represents any number of human agents that may be included in the system 100. The agent device 130 includes a display for displaying a real-time agent console 132. The real-time agent console 132 provides an interface for the human agent 134 to communicate with the customer 124. For example, the real-time agent console 132 can be communicatively coupled to the customer support interface 122 via the network 110.

[0045] The server 140 may include a customer account database 142, a historical chat database 144, an automatic chatbot system 146, a real-time chat system 148, a prediction engine 150, a rule-based system 152, and a feedback module 154. The customer account database 142 and the historical chat database 144 are both data storage warehouses that store previously obtained data. The customer account database 142 stores information and attributes of each registered customer 124 (i.e., a customer who has registered an account on the e-commerce platform), including but not limited to age, gender, race, nationality, email, purchase history, product interests, residential address, geographic location, and any other information corresponding to the customer 124. For example, the information and attributes of each customer 124 stored in the customer account database 142 can be obtained from the process of registering an account on the e-commerce platform, from the activities of the registered customer 124 on the e-commerce platform, and from tracking the IP address of the customer 124.

[0046] The historical chat database 144 stores previous chat transcripts of conversations between each registered customer 124 and the chatbot and / or human agent 134 from chat sessions conducted on the customer support interface 122. The historical chat database 144 can count the number of past chat sessions conducted for each registered customer 124. The historical chat database 144 can receive and store customer feedback and human agent feedback associated with each chat session. For example, the customer feedback and human agent feedback are transmitted to the historical chat database 144 from the feedback module 154.

[0047] The automated chatbot system 146 provides a chatbot and a graphical user interface for customers 124 to communicate with the chatbot and / or human agents 134. For example, the automated chatbot system 146 may receive customer input (e.g., text messages) transmitted from the customer device 120 and the agent device 130, and display the customer input query on a graphical user interface. The automated chatbot system 146 also transmits customer input received from the customer device 120 to the chatbot, and then receives corresponding machine-generated responses (without human intervention) from the chatbot for display on the graphical user interface. For example, the graphical user interface is the customer support interface 122, as further described herein and with respect to Figure 2a-2c and Figure 3a-3c The automated chatbot system 146 determines whether the customer input is part of a new conversation session or part of an existing conversation session. The automated chatbot system 146 determines whether the conversation session is complete based on a completion signal (e.g., interface window closure) from the customer device 120 and / or the agent device 130.

[0048] A chatbot is an automatic computer program that conducts a conversation via an auditory or text method. The chatbot is configured to simulate the behavior of a human as a conversation partner. The chatbot receives a text message from a client device 120 and parses keywords or phrases within the text message and / or uses natural language processing to determine the subject and intent of the text message. The chatbot then generates a reply based on the identified subject and intent. The chatbot then transmits the generated response to an automatic chatbot system 146 for display on a customer support interface 122. In some embodiments, after generating a response, the chatbot is configured to obtain an input data set including the generated response, and feed the input data set to a prediction engine 150 to determine whether the customer 124 should be enabled to use the option of having a chat session with an artificial agent 134. Based on the determination of the prediction engine 150, the automatic chatbot system 146 then determines whether the customer 124 should be enabled to use the option of having a chat session with an artificial agent 134, as will be described in more detail below.

[0049] The real-time chat system 148 may be configured to communicatively connect the customer 124 with the human agent 134 for real-time interaction (e.g., real-time chat). The real-time chat system 148 may receive a request from the automated chatbot system 148 to initiate a real-time chat between the customer 124 of the client device 120 and the real-time agent 134. The request for the real-time chat may include customer information (e.g., customer name, email address, Internet Protocol (IP) address, etc.), information about the type of service requested for the real-time chat (e.g., services related to accounts, purchases, pending orders, technical issues, etc.), and a history of the conversational conversation between the customer 124 and the chatbot. For example, the real-time chat may be initiated during an existing chat session with the chatbot, and the request includes a history of the customer 124's interactions with the chatbot in the existing chat session.

[0050] The prediction engine 150 includes a program or algorithm (e.g., a machine learning algorithm) for automatically processing an input data set given by a chatbot when executed by the system 100 to determine an overall probability that providing a live chat option (or human intervention) will increase the likelihood that a customer 124 will obtain a positive outcome (e.g., resolution of a customer query). The input data set of the prediction engine 150 may include the latest machine-generated replies (or chatbot replies) and corresponding customer queries. In some embodiments, the prediction engine 150 is configured to process the machine-generated replies included in the input data set and determine the degree of effectiveness of the replies based on the corresponding queries. The determined degree of effectiveness of the replies is then used as feature data together with one or more other feature data in the input data set for the prediction engine 150 to determine the overall probability. Other feature data include, for example, information and attributes stored in a customer account database and / or a historical chat database associated with a particular customer 124.

[0051] The prediction engine 150 can implement a boost model. The boost model can be defined by the following equation:

[0052]

[0053] Where P represents the probability value, i represents a specific customer, and Y i =1 indicates that customer i gives positive feedback at the end of the chat session (e.g., via the feedback module 154), W i = 1 indicates that customer i receives the treatment (e.g., the live chat option is made available to customer i), W i = 0 means that customer i did not receive the treatment (e.g., customer i is unable to use the live chat option), X i A vector representing variables associated with customer i (e.g., feature data stored in the customer account database 142 and / or the historical chat database 144 associated with customer i), P(W i =1|X i ) means that at a given X i The probability of processing under the condition of i =0|X i ) means that at a given X i The probability of not taking action when i =1,W i =1|X i ) / P(W i =1|X i ) means that at a given X i The conditional probability that customer i will give positive feedback if the live chat option is made available to customer i is P(Y i =1,W i =0|X i) / P(W i =0|X i ) means that at a given X i The conditional probability that customer i will give positive feedback if the live chat option is not available to customer i.

[0054] In some embodiments, the prediction engine utilizes a lift model to process multiple feature data in the input data set, including the determined degree of effectiveness of the chatbot's responses, to determine an overall probability that human intervention will improve the likelihood that the customer 124 will receive a positive outcome. For example, if the overall probability satisfies a predetermined condition (e.g., exceeds a threshold or is within a threshold range), the prediction engine 150 will determine that the customer's outcome is likely to be improved by human intervention. The term "lift" refers to an increase in the chance that the customer will receive a positive outcome with human intervention compared to the chance that the customer will receive a positive outcome without human intervention.

[0055] The rule-based system 152 applies predefined rules to determine whether the real-time chat option should be made available to the customer 124. For example, the rule-based system 152 is communicatively coupled to the automatic chatbot system 146. The rule-based system 152 may be executed before and / or after the start of a chat session to determine whether the predefined rules are met. A rule typically includes one or more conditions and one or more actions to be taken when the conditions are met. It should be understood that a rule may include multiple conditions, and there may be complex interrelationships between the conditions. The predefined rules may include making only a group of selected customers 124 available to use the real-time chat option. For example, the rule-based system 152 may identify customers 124 who have expressed dissatisfaction (e.g., given negative feedback) with the chatbot interaction in a previous chat session, and make the identified customers 124 available to use the real-time chat option to increase the possibility of these customers 124 obtaining positive results. Other specific implementations of the rule-based system 152 may also be useful. In some embodiments, the rule-based system 152 is configured to be executed before and after the start of a chat session, as described in more detail below. For example, the rule-based system 152 may be executed once or multiple times during a chat session.

[0056] The feedback module 154 is configured to solicit feedback from the customer 124 at the end of the chat session. In some embodiments, the feedback module 154 generates a rating interface (not shown) at the end of the chat session and displays the rating interface on the customer device 120. The rating interface may include any suitable rating options for rating a particular chat session. The customer 124 may operate the rating options or metrics on the rating interface to provide feedback about, for example, an interaction with a chatbot or a human agent 132. The results from the feedback are stored in the historical chat database 144 for future use in the processes further described herein. For example, the feedback results may be supplied to the rule-based system 152 and / or the prediction engine 150 to determine whether the customer 124 should be made available for the real-time chat option.

[0057] It should be understood that Figure 1 The system 100 shown in the is an exemplary system according to one or more embodiments, and a variety of different system environments and configurations are contemplated and are within the scope of the present disclosure. In addition, the system 100 may include: automatic speech recognition (ASR) functionality (not shown), which may enable verbal / voice input (e.g., via VOIP) and / or requests; and text-to-speech (TTS) functionality, which may enable the system 100 to verbally output to an end user (or users) of the system 100 (such as a human agent 134 or a customer 124). However, it should be appreciated that customer input may use text and the system 100 may provide output via text.

[0058] Figure 2a to Figure 2c A flow chart describing a method 200 for implementing an online customer support interface 122 according to certain embodiments is shown. The method 200 is explained in conjunction with the system 100 previously described by previous figures of the present disclosure and the same or similar processes and elements may not be described in detail. Figure 2a to Figure 2c The steps identified in the description are exemplary and may include various alternatives, equivalents or derivatives thereof, including but not limited to the order of their execution. Figure 2a to Figure 2c The steps of (and various alternatives thereof) may be embodied in hardware or software, including a computer-readable storage medium comprising instructions executable by a processor or the like in a computing device.

[0059] In step 202, the method generates a customer support interface 122. The customer support interface 122 may be provided by a business entity as a means of providing assistance to its online customers. For example, a customer 124 may launch the customer support interface 122 from a web browser or other application installed on an associated customer device 120, wherein the web browser (or other application) is connected to an automatic chatbot system 146 associated with the business entity via an IP address. In some embodiments, the customer support interface 122 may be provided on an e-commerce platform accessed by the customer device 120 of the customer 124. The customer support interface 122 may include or represent an interface window of the automatic chatbot system 146. For example, the customer support interface 122 may be configured as an overlay window (or floating window) on a portion of a web page of a website. Alternatively, the customer support interface 122 may be provided by a front-end system (such as a mobile application) installed on the customer device 120. For example, the customer support interface 122 may be configured as a page (e.g., a customer support page) within a mobile application. The automated chatbot system 146 hosts a chatbot that is configured to generate automated conversation conversations with customers via the customer support interface 122 and receive customer input queries from the customer device 120 via the customer support interface 122 in response to the automated conversation conversations.

[0060] In step 204, the method executes the rule-based system 152 to determine whether to make the live chat option available. For example, the rule-based system 152 may be executed prior to the start of a chat session to determine whether to make the live chat option available to the customer 124 on the customer support interface 122 at the start of a particular chat session. In some embodiments, the rules used by the rule-based system 152 may include, but are not limited to, a predefined set of conditional rules, rules learned based on customer behavior and attributes, or a combination thereof. For example, one or more rules may be learned based on customer behavior data and / or customer attribute data obtained from the customer account database 142, the historical chat database 144, and the automated chatbot system 146.

[0061] To illustrate an exemplary rule-based system 152, a first rule may include the following condition: if the customer 124 is new to the customer support interface 122 or has a low chatbot usage rate (or a low frequency of contact with the chatbot) according to the data in the historical chat database 144 and the customer account database 142, then determine not to make the live chat option available on the customer support interface 122 to encourage the customer 124 to interact with the chatbot or increase interaction with the chatbot. A second rule may include the following condition: if the customer 124 is in a relatively high tier (e.g., the highest loyalty tier) of a tiered loyalty program according to the data in the customer account database 142, then determine to make the live chat option available on the customer support interface 122 as a reward provided by the loyalty program. A third rule may include the following condition: if the customer 124 has interacted with the chatbot a predefined number of times within the same chat session, then a determination is made that a live chat option is to be made available on the customer support interface 122 in order to provide the customer 124 with the option to interact with a human agent 134 because the customer 124 has not received sufficient assistance from the chatbot (e.g., to preempt the customer from abandoning the chat session due to lack of human assistance). Other similar rules include any number of conditions that depend on customer behavior and / or customer attribute data obtained from the historical chat database 144 and the customer account database 142 to cause the method to determine whether to make the live chat option available on the customer support interface 122. In addition, in some embodiments, if any of the rules used by the rule system 152 are not satisfied (e.g., evaluated to true), a default determination may be made not to make the live chat option available on the customer support interface 122.

[0062] In some embodiments, when it is determined that the live chat option should be made available to the customer 124 at the beginning of the chat session, the method proceeds from step 204 to step 240 .

[0063] In step 240, the method generates an initial message for display on the customer support interface 122. The initial message is generated by the chatbot and displayed on the customer support interface 122 to engage the customer 124 in an electronic conversation and mark the beginning of the chat session. The initial message is, for example, a welcome message. The method continues to step 242.

[0064] In step 242, the method makes a live chat option available on the customer support interface 122. The live chat option can be represented by a selectable interface element located on the customer support interface 122. For example, the live chat option can be an selectable interface element presented via a chatbot message. In another example, the live chat option can be displayed as an selectable interface element that occupies a dedicated position on the customer support interface 122. The customer 124 can select the live chat option (e.g., select a "Contact a Live Agent" option represented by the selectable interface element) to conduct a chat session with a human agent 134.

[0065] In some embodiments, when the initial message is displayed on the customer support interface 122, the live chat option is available at the same time. For example, the initial message may include the live chat option as part of the message. In some alternative embodiments, the live chat option is available after (e.g., shortly after) the initial message is displayed on the customer support interface 122. For example, the live chat option does not form part of the initial message. The initial message is used to encourage the customer 124 to communicate with the chatbot rather than use the available live chat option. Therefore, the customer 124 may be given the choice of interacting with the chatbot or interacting with the human agent 134 via the live chat option at the beginning of the chat session.

[0066] In other embodiments, when it is determined that the live chat option should not be made available to the customer 124 at the beginning of the chat session, the method proceeds from step 204 to step 206. The live chat option may be represented by an optional interface element located on the customer support interface 122. For example, if the live chat option is an optional interface element presented via a chatbot message, the live chat option may be made unavailable by omitting the real-time chat option from the chatbot message (e.g., a welcome message). In another example, if the live chat option is an optional interface element that occupies a dedicated position on the customer support interface 122, the real-time chat option may be made unavailable by visually hiding the optional interface element or by disabling the optional interface element (e.g., the optional interface element is grayed out to indicate that it is in an inactive / disabled state).

[0067] In step 206, the method executes the prediction engine 150 to determine whether to make the live chat option available. For example, the prediction engine 150 may be executed before the chat session begins and in sequence with the rule-based system 152 to determine whether to make the live chat option available to the customer 124 on the customer support interface 122 at the beginning of a particular chat session. In some embodiments, after generating the customer support interface 122, the chatbot obtains an initial input data set for the prediction engine 150 and feeds the initial input data set to the prediction engine 150. The initial input data set may include customer behavior data and / or customer attribute data obtained from the historical chat database 144 and the customer account database 142. The method utilizes the prediction engine 150 to determine the overall probability or likelihood that providing the live chat option to the customer 124 will increase the likelihood that the customer 124 will obtain a positive outcome.

[0068] In some embodiments, the prediction engine 150 utilizes a lift model to process multiple feature data in a given input data set (e.g., an initial input data set), and calculates a first and a second conditional probability to determine an overall probability value that human intervention (e.g., assistance from an artificial agent 134) will increase the likelihood that the customer 124 will obtain a positive result. For example, the first conditional probability indicates the likelihood that the customer's query will be resolved through human intervention, and the second conditional probability indicates the likelihood that the customer's query will be resolved without human intervention. The overall probability value is calculated by subtracting the second conditional probability value from the first conditional probability value. In some embodiments, when the overall probability value satisfies a predetermined condition (e.g., exceeds a threshold), it is determined that the human intervention will increase the likelihood that the customer 124 will obtain a positive result. The overall probability may represent an estimated lift value for the customer 124 due to receiving assistance from the artificial agent 134.

[0069] The probability value can be a value between 0 and 1, where 1 indicates that a given result is predicted to occur, and 0 indicates that a given result is predicted not to occur. For example, a higher probability value indicates a greater likelihood of the result occurring, and a lower probability value indicates a lesser likelihood of the result occurring. In this case, the predetermined condition is that the overall probability value (or the estimated lift value) exceeds a threshold value of 0. In some exemplary implementations, when the first conditional probability value is 0.7 and the second conditional probability value is 0.3, the overall probability value is equal to 0.4 and the predetermined condition is met; when the first conditional probability value is 0.5 and the second conditional probability value is 0.8, the overall probability value is equal to -0.3 and the predetermined condition is not met; when the first conditional probability value is the same as the second conditional probability value, the overall probability value is equal to 0 and the predetermined condition is not met.

[0070] Table 1 below is provided to illustrate the calculation of the overall probability value using the above exemplary implementation, but does not limit the present invention. Referring to Table 1, the leftmost column shows an exemplary customer, the second column from the left shows an exemplary first conditional probability value associated with the customer, the third column from the left shows an exemplary second conditional probability value associated with the customer, and the rightmost column shows the overall probability value calculated based on the corresponding first and second conditional probability values.

[0071] Table 1. Exemplary calculation of overall probability values

[0072]

[0073] Although the method is described as performing steps 204 and 206 separately, it should be understood that the method may also be configured to incorporate step 204 into step 206. For example, the rule-based system 152 may be incorporated into the prediction engine 150. In an alternative embodiment, the prediction engine 150 sequentially implements the rule-based system 152 and the boosting model to determine whether to make the live chat option available on the customer support interface 122. For example, the prediction engine 150 first implements the rule-based system 152 to make a decision, and if the rule-based system 152 determines not to make the live chat option available on the customer support interface 122, the method continues by implementing the boosting model to make a decision.

[0074] In some embodiments, when the prediction engine 150 determines that the overall probability value satisfies a predetermined condition (e.g., the overall probability value is greater than a threshold value of 0), the method proceeds from step 206 to step 240. For example, based on the overall probability value, the overall probability that the human intervention will increase the likelihood that the customer 124 will obtain a positive outcome is predicted.

[0075] In other embodiments, when the prediction engine 150 determines that the overall probability value does not satisfy the predetermined condition (e.g., the overall probability value is less than or equal to a threshold value of 0), the method proceeds from step 206 to step 208. For example, based on the overall probability value, it is predicted that manual intervention will not increase the likelihood that the customer 124 will obtain a positive outcome.

[0076] Continuing to step 208, the method generates an initial message for display on the customer support interface 122. For example, the initial message is generated by the chatbot and displayed on the customer support interface 122 at the client device 120 to engage the customer 124 in an electronic conversation and mark the beginning of the chat session. In some embodiments, the chatbot's initial message is a welcome message for initiating a conversation with the customer 124. For example, the initial message may be "Hello, how can I help?". Other suitable messages for initiating a conversation with the customer 124 may also be useful. In step 208, since the live chat option is not available on the customer support interface (e.g., the initial message does not include the live chat option), the customer 124 cannot bypass the chatbot by requesting to chat with a human agent 134, and is forced to interact with the chatbot to obtain customer support services.

[0077] It should be appreciated that the chatbot's initial message may also be configured to include one or more selectable chatbot options representing common questions or requests to facilitate interaction with the customer 124 .

[0078] In step 210, the method receives a customer input query and transmits the customer input query to the chatbot. In some embodiments, the chatbot includes a natural language processing (NLP) system. The NLP system may be configured to perform semantic processing on a customer input query (e.g., a message) received from a client device 120 to determine the meaning or context of the customer input query. The semantic processing may receive any relevant information as input, including but not limited to a message text from a customer 124, information about the customer 124 (e.g., an IP address, customer gender, information about the entry point from which the customer initiates the customer support interface 122), or a customer identifier. In some embodiments, information from an e-commerce platform server, such as information about the customer 124 or information about an e-commerce platform service (e.g., a promotional activity), may be used during semantic processing. Semantic processing may be used to select the intent of the customer 124 input query from a specified set of intents. For example, a chatbot may be configured with a set of intents related to services provided by an e-commerce platform, such as technical support, shipping status, or common problems related to a product or service.

[0079] In step 212, the method determines a reply to the customer input query and generates a reply for display on the customer support interface 122. In some embodiments, after the chatbot receives the customer 124 input query, semantic processing is used to understand the meaning of the customer 124 input query and then generate a machine-generated reply. For example, the chatbot generates relevant information, available operations, and / or options for display on the customer support interface 122. In an exemplary chatbot interaction, the customer 124's first input query is a question or request, and the chatbot provides a first reply that immediately solves the customer 124's first input query (without human intervention). In another exemplary chatbot interaction, more information is needed from the customer 124, and the machine-generated first reply requires the customer 124 to provide additional information. The reply (or response) that the chatbot can provide to the customer 124 may be limited by the information stored in the system 100. If the stored information is insufficient, the chatbot may not be able to provide a sufficiently effective response to provide a positive result to the customer 124 (e.g., solve the customer's query). The method continues to step 214.

[0080] In step 214, the method executes the rule-based system 152 to determine whether to make the live chat option available. For example, the rule-based system 152 may be executed before and after the start of a chat session to determine whether to make the live chat option available to the customer 124 on the customer support interface 122 at the start or during a particular chat session. As described herein (e.g., in step 204), the rule-based system 152 may include the following condition: if the customer 124 interacts with the chatbot a predefined number of times within the same chat session, then determine to make the live chat option available on the customer support interface 122 to provide the customer 124 with the option to interact with a human agent 134. In some embodiments, the automatic chatbot system 146 is configured to update a chatbot interaction counter to monitor the number of interactions between the chatbot and the customer 124 within the same chat session. The term "interaction" refers to a query-reply pair. For example, each time the chatbot generates a reply to a customer input query in a chat session, the counter value of the chatbot interaction counter is immediately or almost immediately increased by 1.

[0081] In some embodiments, at step 214, the rule-based system 152 determines whether the counter value of the chatbot interaction counter has reached an interaction threshold (e.g., "5"). For example, after updating the chatbot interaction counter, the rule-based system 152 determines whether the counter value of the chatbot interaction counter has reached (e.g., equal to or greater than) the interaction threshold. In some embodiments, when it is determined that the real-time chat option is to be made available (e.g., the counter value of the chatbot interaction counter has reached the interaction threshold), the method proceeds from step 214 to step 250.

[0082] In step 250, the method displays the chatbot reply and the live chat option on the customer support interface 122. In one embodiment, the machine-generated reply and the live chat option are displayed on the customer support interface 122 simultaneously. For example, the live chat option is displayed as part of the machine-generated reply. In another example, the live chat option may be displayed as an optional interface element that occupies a dedicated position on the customer support interface 122. In an alternative embodiment, the live chat option is made available after a short delay period (e.g., 2 seconds or 3 seconds later) after the machine-generated reply is displayed. For example, the live chat option is displayed in the form of a message on the customer support interface 122 shortly after the machine-generated reply is displayed (e.g., 1, 2, or 3 seconds later). Displaying the live chat option after a short delay period after the machine-generated reply is displayed advantageously encourages the customer to consider whether the machine-generated reply has resolved the query before selecting the live chat option, such as when the machine-generated reply presents two or more possible options (or solutions) to respond to the customer's query. This can help further improve the chatbot's turnaround rate and save limited human agent 134 resources.

[0083] In other embodiments, when it is determined that the real-time chat option is not to be made available (eg, the counter value of the chatbot interaction counter is less than the interaction threshold), the method proceeds from step 214 to step 216 .

[0084] In step 216, the method executes the prediction engine 150 to determine whether to make the live chat option available. For example, the prediction engine 150 may be executed before or after the chat session begins and in sequence with the rule-based system 152 to determine whether to make the live chat option available to the customer 124 on the customer support interface 122 at the beginning or during a particular chat session. In some embodiments, after generating a reply to the input query of the customer 124, the chatbot obtains an input data set (e.g., a first input data set) for the prediction engine 150 and feeds the input data set to the prediction engine 150. The input data set may include the most recently generated reply and its corresponding query. The method utilizes the prediction engine 150 to determine the overall probability or likelihood that human intervention will increase the likelihood of the customer 124 obtaining a positive outcome.

[0085] In some embodiments, the prediction engine 150 analyzes the chatbot responses to determine the degree of validity of the responses based on their corresponding customer input queries. For example, the prediction engine 150 compares the responses to the corresponding queries to determine the degree of semantic match between the responses provided by the chatbot and the corresponding customer queries (e.g., based on semantic analysis of the queries and responses). In one exemplary scenario, the degree of validity may be high when the customer 124 submits a query to check the status of an order, where the query may be: "What is the status of order #123?", and the chatbot response includes the status of the order, where the chatbot reply may be "Order #123 has been shipped". In another exemplary scenario, the degree of validity may be low when the customer 124 submits a query to dispute the status of an online order, where the query may be: "The status of order #123 shows as shipped, but I never received it", and the chatbot response includes a suggested action not mentioned in the query, where the chatbot reply may be "You may want to request a refund". The determined degree of validity of the response is then used as feature data together with one or more other feature data in the input data set for processing by the prediction engine 150.

[0086] In some embodiments, the prediction engine 150 utilizes the boost model to process multiple feature data in a given input data set (e.g., the first input data set), and calculates the first and second conditional probabilities to determine an overall probability value that human intervention (e.g., assistance from the human agent 134) will increase the likelihood that the customer 124 will obtain a positive result, as described in step 206. For example, if the prediction engine determines the effectiveness level of the machine-generated first response based on its corresponding first input query in step 216, the calculated overall probability value takes into account the determined effectiveness level of the machine-generated first response. For example, a high effectiveness level (e.g., within an upper threshold range) can increase the overall probability value and a low effectiveness level (e.g., within a lower threshold range) can decrease the overall probability value.

[0087] Although the method is described as performing steps 214 and 216 separately, it should be understood that the method may also be configured to incorporate step 214 into step 216. For example, the rule-based system 152 may be incorporated into the prediction engine 150. In this case, the prediction engine 150 may be configured to first implement the rule-based system 152, and if the rule-based system 152 determines not to make the live chat option available on the customer support interface 122, the method continues by implementing the prediction engine (e.g., the boosting model) to determine whether to make the live chat option available on the customer support interface 122.

[0088] In some embodiments, when the prediction engine 150 determines to make the live chat option available, the method proceeds from step 216 to step 250. For example, the overall probability value (taking into account the determined degree of effectiveness of the machine-generated response) satisfies a predetermined condition (e.g., the overall probability value is greater than a threshold of 0). Therefore, based on the overall probability value, it is predicted that the overall probability that human intervention will increase the likelihood that the customer 124 will obtain a positive outcome.

[0089] In other embodiments, when the prediction engine 150 determines not to make the live chat option available, the method proceeds from step 216 to step 218. For example, the overall probability value (taking into account the determined degree of effectiveness of the machine-generated response) does not satisfy the predetermined condition (e.g., the overall probability value is less than or equal to a threshold of 0). Therefore, based on the overall probability value, it is predicted that human intervention will not increase the likelihood of the customer 124 obtaining a positive outcome.

[0090] In step 218, the method displays the chatbot response on the customer support interface 122 without making the live chat option available. For example, after the first interaction with the chatbot (e.g., the first query-response pair), the live chat option continues to be reserved for the customer 124. After displaying the chatbot response on the customer support interface 122, the method returns to step 210. For example, the method continues to generate a response (without human intervention) in response to each subsequent input query of the customer on the customer support interface 122, and uses each of the customer's live interactions with the chatbot as feature data for the prediction engine 150 to determine, at each interaction, an updated overall probability value that human intervention will increase the likelihood that the customer 124 will obtain a positive outcome.

[0091] It is contemplated that the customer's input query may be resolved by the chatbot without the assistance of a human agent 134, or that the customer 124 may abandon the chat session before the live chat option is made available through the rule-based system 152 or the prediction engine 150. Thus, in some embodiments, the results of the chat session and the chat transcript of the chat session may be stored in the historical chat database 144 as customer classification data. For example, the customer classification data may include, but is not limited to, a determination of whether the customer's query has been resolved, a determination of whether to abandon the chat session if the live chat option is unavailable, customer feedback on the interaction with the chatbot and / or human agent 134, human agent feedback to the customer (e.g., input query, behavior, etc.), or a combination thereof. The acquired customer classification data may then be used to train the prediction engine (e.g., improve the model) to make better predictions and dynamically adapt the prediction engine to customer needs that may change over time.

[0092] As described above, the present invention is configured to use a more cost-effective chatbot to handle a customer's query and conserve limited human agent 134 resources until it is predicted that providing a live chat option will increase the likelihood that the customer will obtain a positive outcome. In other words, the live chat option is selectively presented to the customer only when it is determined that the customer may benefit most from interacting with a human agent 134. In this way, the allocation of available human agents is optimized and the chatbot's turn rate is improved. Therefore, the present invention advantageously balances increasing the turn rate with increasing customer satisfaction by running a predictive engine during a chat session to monitor the customer's interaction with the chatbot and dynamically determine when the live chat option should be provided to the customer (e.g., when it is predicted that human intervention may improve the customer's outcome).

[0093] Figure 3a to Figure 3c An example of a chat session displayed on a customer support interface 300 according to some embodiments is shown. The chat session and customer support interface 300 are explained in conjunction with the system 100 and method 200 as previously described by the previous figures of the present disclosure and the same or similar processes and elements may not be described in detail. Specifically, Figure 3a to Figure 3c An exemplary diagram of a customer support interface 300 is shown, which may be viewed at Figure 1 The system 100 and Figure 2a to Figure 2c Some specific implementations of the methods described in the present invention are presented on a display screen of the client device 120.

[0094] In some embodiments, the customer support interface 300 includes a live text-based communication area 302 and a message input field (not shown). Messages received at the customer support interface 300 may be displayed in the live text-based communication area 302 (hereinafter referred to as the "communication area"). In some embodiments, the communication area 302 displays a conversation between the customer 124 and the chatbot of the e-commerce platform in real time. For example, the customer 124 and the chatbot may exchange separate messages, where the separate messages are collectively referred to as "conversation conversations". For example, the messages are displayed in text bubbles. Other types of message containers may also be used to present messages.

[0095] exist Figure 2a to Figure 2c In an exemplary implementation of the system 100 and method described in Figure 3a The customer support interface 300 is shown from the customer's perspective. As shown, the chatbot can generate and display a welcome message 342 in the communication area 302 to indicate the start of the chat session. For example, when the customer 124 selects the chatbot icon on the web page of the e-commerce platform, the welcome message 342 is displayed.

[0096] In some embodiments, the customer support interface 300 does not make the real-time chat option available at the beginning of the chat session. For example, the rule-based system 152 is executed before the welcome message 342 is generated, and it has been determined that the real-time chat option is not available to the customer 124 at the beginning of the chat session. The real-time chat option is an option for real-time chat with a human agent 134. In some embodiments, the customer support interface 300 is not configured with a dedicated real-time chat button, and the real-time chat option is selectively provided only through a chatbot message. In an alternative embodiment, the customer support interface 300 may be configured to include a dedicated real-time chat button that is disabled (e.g., the button turns gray to indicate that it is in an invalid state) or visually hidden at the beginning of the chat session. In this exemplary specific implementation, the real-time chat option is selectively provided only through a chatbot message. Since the real-time chat option is not presented in the welcome message 342, the customer 124 is forced to interact with the chatbot.

[0097] Figure 3b The customer support interface 300 is shown receiving a first query 362 from a customer 124. The customer may submit a query to the chatbot using any suitable technique, such as typing or speaking. For example, the query may be submitted as a text message or as audio of the customer 124 speaking the message. When the customer 124 speaks the message to the customer device 120, the customer device 120 or another component of the system 100 may perform speech recognition to convert the customer's spoken words into text.

[0098] The chatbot receives the customer's first query 362 and automatically processes the query 362 to determine a response to the first reply without the involvement of the human agent 134. For example, the chatbot performs semantic processing on the text of the customer's query to understand the meaning of the query, and then automatically generates a reply for display on the customer support interface 300. Figure 3b , the chatbot has processed a first query 362 received from customer 124, determined that the intent of the first message is related to the status of a particular order (e.g., order #123) associated with customer 124, and generated a first reply 344 indicating the current status of the particular order.

[0099] After generating the first reply 344, the chatbot updates the chatbot interaction counter on the automated chatbot system 146 and communicates with the rule-based system 152. For example, the chatbot may cause the rule-based system 152 to be executed during the chat session to determine whether the counter value of the chatbot interaction counter has reached (e.g., is equal to or greater than) an interaction threshold. Figure 3bIn the example of FIG. 1 , the rule-based system 152 has determined that the counter value of the chatbot interaction counter has not reached the interaction threshold. Thereafter, the chatbot obtains a first input data set and feeds the first input data set to the prediction engine 150. The first input data set includes a first reply 344 and a first query 362.

[0100] The prediction engine 150 processes the first input data set to determine a degree of effectiveness of the first reply based on the first query (e.g., based on a semantic analysis of the query and the reply). The prediction engine 150 then uses the determined degree of effectiveness of the first reply as feature data along with one or more other feature data in the first input data set as input data for further processing. For example, the prediction engine 150 utilizes a boost model to further process the given input data, which includes the determined degree of effectiveness of the first reply, and determines an overall probability that providing the live chat option to the customer 124 will increase the likelihood that the customer 124 will obtain a positive outcome. Figure 3b , the prediction engine 150 has determined that the overall probability value is less than or equal to 0 (e.g., the likelihood that the customer's first query is resolved without human intervention is equal to or greater than the likelihood that the customer's first query is resolved with human intervention). Therefore, the overall probability value does not meet the predetermined condition for displaying the live chat option (e.g., exceeds the threshold of 0).

[0101] After determining the overall probability, the prediction engine 150 sends an instruction to the chatbot (or automated chatbot system 146) to continue to retain the live chat option. After receiving the instruction from the prediction engine 150, the chatbot then transmits a first reply 344 for display on the customer support interface 300 without making the live chat option available. After receiving the first reply 344, the customer support interface 300 displays the first reply 344 in the communication area 302.

[0102] Figure 3c The customer support interface 300 is shown displaying a second query 364 received from the customer 124. The chatbot automatically processes the second query 364 to determine a second response to the second query 364 without the involvement of the human agent 134. Figure 3c , the chatbot has processed a second query 364 received from a customer 124 , determined that the second query 364 is intended to disagree with (or raise objections to) a machine-generated first reply 344 , and generated a second reply 346 that includes presenting one or more suggested actions that the customer 124 can take.

[0103] After generating the second reply 346, the chatbot updates the chatbot interaction counter on the automated chatbot system 146 and communicates with the rule-based system 152. For example, the rule-based system is executed to determine whether the updated counter value of the chatbot interaction counter reaches the interaction threshold. Figure 3c In the example, the rule-based system 152 has determined that the counter value of the chatbot interaction counter has not reached the interaction threshold. Thereafter, the chatbot obtains a second input data set and feeds the second input data set to the prediction engine 150. The second input data set includes a second reply 346 and a second query 364.

[0104] Prediction engine 150 processes the second input data set to determine a degree of effectiveness of the second response based on the second query. Prediction engine 150 then uses the determined degree of effectiveness of the second response as feature data along with one or more other feature data in the second input data set as input data for further processing. For example, prediction engine 150 utilizes a boost model to further process the given input data, which includes the determined degree of effectiveness of the second response, and determines an overall probability that providing a live chat option to customer 124 will increase the likelihood that customer 124 will obtain a positive outcome. Figure 3c , prediction engine 150 has determined that the overall probability value is greater than 0 (e.g., the customer's second query is more likely to be resolved with human intervention than the customer's second query is resolved without human intervention). Therefore, the overall probability value satisfies the predetermined condition for displaying the live chat option (e.g., exceeds the threshold of 0).

[0105] After determining the overall probability, the prediction engine 150 sends an instruction to the chatbot (or automated chatbot system 146) to make the live chat option available. After receiving the instruction from the prediction engine 150, the chatbot then transmits the second reply 346 along with the live chat option 360 to the customer support interface 300. After receiving the second reply 346 and the live chat option 360, the customer support interface 300 displays the second reply 346 and the live chat option 360 in the communication area 302. Figure 3c As shown, real-time chat option 360 may be displayed as a selectable interface element within the text bubble containing second reply 346. For example, real-time chat option 360 is displayed simultaneously with second reply 346 and forms part of second reply 346. Alternatively, second reply 346 may be displayed at a different point in time than real-time chat option 360 (e.g., real-time chat option 360 may be displayed in a text bubble separate from the text bubble containing the second reply). For example, second reply 346 may be displayed first, and real-time chat option 360 may be displayed after a delay period after second reply 346 is displayed.

[0106] In an alternative implementation, if the live chat option is configured as an optional interface element that occupies a dedicated location (not shown) on customer support interface 300, and the live chat option is made unavailable at the start of a chat session by visually hiding the optional interface element or by disabling the optional interface element, the live chat option may be made available by displaying or enabling the optional interface element for selection by customer 124 when or after second reply 346 is displayed.

[0107] As described above, the present invention is configured to use a more cost-effective chatbot to resolve a customer's 124 query and conserve limited human agent 134 resources until it is predicted that providing a live chat option 360 will increase the likelihood that the customer will obtain a positive outcome. In other words, the live chat option 360 is selectively presented to the customer only when it is determined that the customer 124 may benefit most from an interaction with a human agent 134. In this way, the allocation of available human agents 134 is optimized and the chatbot's turn rate is improved. Therefore, the present invention advantageously balances increasing the turn rate with increasing customer satisfaction by running a predictive engine during a chat session to monitor the customer's interaction with the chatbot and dynamically determine when the live chat option 360 should be provided to the customer (e.g., when it is predicted that human intervention can improve the outcome for the customer).

[0108] The methods described herein are fully intended to be able to operate on a variety of devices or systems. Any hardware platform suitable for executing the methods described herein is suitable for the present invention. Unless otherwise explicitly stated, and as will be apparent from the following, it should be understood that throughout this specification, discussions of terms such as "receive", "identify", "initiate", "tag", "transmit", "run", "increment", "determine", "allocate", "approval", "select", "send", "calculate", "analyze", "replace", "generate", "initialize", "output", "train", etc. refer to the actions and processes of a computer system or similar electronic device that manipulates data represented as physical quantities within a computer system and converts the data into other data similarly represented as physical quantities within a computer system or other information storage, transmission or display device.

[0109] In this application, unless otherwise stated, the terms "comprises," "comprising," and grammatical variations thereof are intended to represent "open" or "inclusive" language such that they include the listed elements but also allow for the inclusion of additional, non-explicitly listed elements.

[0110] Obviously, after reading the above disclosure, various other modifications and adaptations of the present application will become apparent to those skilled in the art without departing from the spirit and scope of the present application, and all such modifications and adaptations are intended to be within the scope of the appended claims.

Claims

1. A method for implementing a customer support interface, the method comprising the following steps: generating the customer support interface for the customer to conduct a chat session with the chatbot, wherein an option to conduct a chat session with a human agent is not available on the customer support interface; receiving, by the chatbot, a first query from the customer; determining, by the chatbot, a first response to the first query and generating the first response for display on the customer support interface; After generating the first response, determining, using a prediction engine, an overall probability value that offering the customer the option to conduct a chat session with a human agent will increase the likelihood that the customer will obtain a positive outcome; Determining whether the overall probability value satisfies a predetermined condition; as well as In response to determining that the overall probability value satisfies a predetermined condition, making an optional interface element representing the option to conduct a chat session with a human agent available on the customer support interface, wherein selecting the optional interface element transfers the customer from the chat session with the chatbot to the chat session with the human agent.

2. The method of claim 1 , wherein determining the overall probability value that offering the customer the option to conduct a chat session with the human agent will increase the likelihood that the customer will obtain a positive outcome comprises: The prediction engine is used to analyze one or more characteristic data associated with the customer and calculate a first conditional probability value and a second conditional probability value, wherein The first conditional probability value indicates the likelihood that the client will obtain the positive result with human intervention, and The second conditional probability value indicates the possibility that the client obtains the positive result without human intervention; as well as The overall probability value is calculated by subtracting the second conditional probability value from the first conditional probability value. 3 . The method of claim 2 , wherein analyzing the one or more characteristic data associated with the customer comprises analyzing the first response to determine a degree of effectiveness of the first response based on the first query.

4. The method of any one of claims 1 to 3, wherein the prediction engine implements a boosting model to determine the overall probability value. 5 . The method according to claim 1 , wherein determining whether the overall probability value satisfies a predetermined condition comprises determining whether the overall probability value is greater than a threshold value.

6. The method of claim 5, wherein if the overall probability value is not greater than the threshold, the option to conduct a chat session with a human agent remains unavailable to the customer.

7. The method of any one of claims 1 to 6, wherein the option to conduct a chat session with a human agent is made unavailable on the customer support interface by visually hiding the selectable interface element from the customer.

8. The method of any one of claims 1 to 6, wherein the option to conduct a chat session with a human agent is made unavailable on the customer support interface by disabling the selectable interface element.

9. The method according to any one of claims 1 to 8, wherein displaying the first reply and simultaneously making the selectable interface element available on the customer support interface, or After or shortly after displaying the first reply, the selectable interface element is made available on the customer support interface.

10. The method of any one of claims 1 to 8, wherein the selectable interface element is made available as part of the first reply.

11. The method according to any one of claims 1 to 10, further comprising: executing a rules-based system prior to the initiation of said chat session; as well as The rule-based system is used to determine to make the option to conduct a chat session with a live agent unavailable on the customer support interface at the initiation of the chat session.

12. The method according to any one of claims 1 to 10, further comprising: The rule-based system is executed after generating the first reply, wherein the rule-based system includes a condition that depends on a number of interactions between the customer and the chatbot in the chat session.

13. The method according to claim 11 or 12, wherein the rule-based system obtains data from a customer account database and / or a historical chat database.

14. A system comprising one or more computers and one or more storage devices storing computer-readable instructions, which, when executed by the one or more computers, cause the one or more computers to perform one or more operations of the method according to any one of claims 1 to 13.

15. One or more non-transitory computer storage media storing instructions that, when executed by one or more computers, cause the one or more computers to perform one or more operations of the method of any one of claims 1 to 13.

Images