Techniques for automatic generation of contact center system embeddings

The automatic generation of contact center system embeddings using machine learning addresses the challenge of agent qualification and queue assignment, enhancing call routing and new agent integration efficiency.

JP7441954B6Active Publication Date: 2024-12-03GENESYS CLOUD SERVICES INC
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2022543016
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-12-31
Filing Date
2021-10-11
Publication Date
2024-12-03
Estimated Expiration
2041-10-11

AI Technical Summary

Technical Problem

Contact centers face challenges in ensuring that calls are routed to the most qualified agent, especially when the most qualified agent is unavailable, and new agents are often assigned to virtual queues where they lack the necessary skills, leading to unproductivity and extended ramp-up times.

Method used

A method and system for automatic generation of contact center system embeddings using machine learning to determine similarities between agents, their skills, and virtual queues, enabling efficient routing of calls and optimal assignment of new agents based on generated embeddings.

Benefits of technology

Enhances the ability to route calls to the most suitable agents and efficiently assign new agents to appropriate virtual queues, improving productivity and reducing ramp-up times by leveraging agent and queue embeddings.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007441954000001
    Figure 0007441954000001
  • Figure 0007441954000002
    Figure 0007441954000002
  • Figure 0007441954000003
    Figure 0007441954000003
Patent Text Reader

Abstract

A method for automatically generating contact center system embeddings according to one embodiment includes: determining, by a computing system, contact center system agents, contact center system agent skills, and / or contact center system virtual queue experiences; generating, by the computing system, a matrix representation based on the contact center system agents, contact center system agent skills, and / or contact center system virtual queue experiences; generating, by the computing system, one of a contact center system agent identifier, a contact center system agent skill identifier, and / or a contact center system virtual queue identifier based on the matrix representation; and associating, by the computing system, the contact center system agent identifier, the contact center system agent skill identifier, and / or the contact center system virtual queue identifier with the contact center system agent. and training, by a computing system, the contact center system agent embeddings, contact center system agent skill embeddings, and / or contact center system virtual queue embeddings by applying machine learning to obtain final weights for the contact center system agent embeddings, contact center system agent skill embeddings, and / or contact center system virtual queue embeddings.
Need to check novelty before this filing date? Find Prior Art

Description

[Background technology]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Non-provisional Patent Application No. 17 / 139,033, filed December 31, 2020, which is incorporated by reference in its entirety.

[0002] Contact center agents may have a particular set of skills and may serve a variety of different end users. Contact centers may often receive more end user telephony calls than there are agents available to manage such calls. End users may want their calls answered by the agent who is most qualified to answer their question and / or an agent who has previously answered their question. However, the most qualified agent and / or the same agent may not always be available to answer the end user's call because the agent may be unavailable for a variety of different reasons (e.g., on a call with another end user, etc.). The problem is that it may be difficult to find another agent who is most similar to the unavailable agent to route the call to.

[0003] Additionally, after a new agent is hired into the contact center, it may not be known which virtual queue within the contact center the new agent should be assigned to, or what type of end-users the new agent should take calls from. The problem is that assigning the new agent to a virtual queue in which the new agent does not have the skills to be successful can result in the new agent being unproductive and / or can result in a long ramp-up time for the new agent to become productive. Summary of the Invention

[0004] According to one embodiment, a method for automatic generation of contact center system embeddings includes determining, by a computing system, at least one of one or more contact center system agents, one or more contact center system agent skills, and one or more contact center system virtual queue experiences; generating, by the computing system, a matrix representation based on the at least one of the one or more contact center system agents, the one or more contact center system agent skills, and the one or more contact center system virtual queue experiences; generating, by the computing system, at least one of one or more contact center system agent identifiers, one or more contact center system agent skill identifiers, and one or more contact center system virtual queue identifiers based on the matrix representation; converting at least one of the contact center system virtual queue identifiers into at least one of the one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings, where weights of at least one of the one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings are randomly initialized; and applying machine learning by the computing system to obtain final weights of at least one of the one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings;and training at least one of one or more contact center system agent skill embeddings and one or more contact center system virtual queue embeddings.

[0005] In some embodiments, the method may further include determining, by the computing system, a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings based on comparing final weights of at least one of the one or more of the contact center system agent embeddings, the one or more of the contact center system agent skill embeddings, and the one or more of the contact center system virtual queue embeddings.

[0006] In some embodiments, determining the similarity of at least one of the two or more contact center system agent embeddings, the two or more contact center system agent skill embeddings, and the two or more contact center system virtual queue embeddings may include utilizing at least one of cosine similarity, principal component analysis, or t-distribution stochastic neighbor embeddings.

[0007] In some embodiments, the method may further include at least one of: routing the telephony call to a contact center system agent in response to determining a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings; and assigning the contact center system agent to a virtual queue in response to determining a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings.

[0008] In some embodiments, at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings may include a k-dimensional vector.

[0009] In some embodiments, the k-dimensional vector may be in a 5-dimensional space or a 10-dimensional space.

[0010] In some embodiments, at least one of the converting and training may be based on a neural network.

[0011] Some embodiments may further include converting the matrix representation to a single row representation for training.

[0012] In some embodiments, applying machine learning to obtain the final weights may include determining a dot product and ground truth of at least one of one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings.

[0013] In some embodiments, the one or more contact center system agent skills may include at least one of one or more language skills, one or more product skills, or one or more instructional skills.

[0014] According to one embodiment, a system for automated generation of contact center system embeddings may include at least one processor and at least one memory including a plurality of instructions stored thereon, which, upon execution by the at least one processor, cause the system to: determine, by the computing system, at least one of one or more contact center system agents, one or more contact center system agent skills, and one or more contact center system virtual queue experiences; generate, by the computing system, a matrix representation based on the at least one of the one or more contact center system agents, the one or more contact center system agent skills, and the one or more contact center system virtual queue experiences; generate, by the computing system, at least one of one or more contact center system agent identifiers, one or more contact center system agent skill identifiers, and one or more contact center system virtual queue identifiers based on the matrix representation; and converting, by the computing system, at least one of the one or more contact center system agent identifiers, the one or more contact center system agent skill identifiers, and the one or more contact center system virtual queue identifiers into at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings, wherein a weight of at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings is randomly initialized; and converting, by the computing system, the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings,and training at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings by applying machine learning to obtain final weights for at least one of the one or more contact center system virtual queue embeddings.

[0015] In some embodiments, the instructions may further cause the system to determine a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings based on comparing, by the computing system, a final weighting of at least one of the one or more of the contact center system agent embeddings, the one or more of the contact center system agent skill embeddings, and the one or more of the contact center system virtual queue embeddings.

[0016] In some embodiments, determining the similarity of at least one of the two or more contact center system agent embeddings, the two or more contact center system agent skill embeddings, and the two or more contact center system virtual queue embeddings may include utilizing at least one of cosine similarity, principal component analysis, or t-distribution stochastic neighbor embeddings.

[0017] In some embodiments, the instructions may further cause the system to route the telephony call to a contact center system agent in response to determining a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings, or to assign a contact center system agent to a virtual queue in response to determining a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings.

[0018] In some embodiments, at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings may include a k-dimensional vector.

[0019] In some embodiments, the k-dimensional vector may be in a 5-dimensional space or a 10-dimensional space.

[0020] In some embodiments, at least one of the converting and training may be based on a neural network.

[0021] In some embodiments, the instructions may further cause the system to convert the matrix representation into a single row representation for training.

[0022] In some embodiments, applying machine learning to obtain the final weights may include determining a dot product and ground truth of at least one of one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings.

[0023] In some embodiments, the one or more contact center system agent skills may include at least one of one or more language skills, one or more product skills, or one or more instructional skills.

[0024] This Summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to be used as an aid in limiting the scope of the claimed subject matter. Further embodiments, forms, features, and aspects of the present application will become apparent from the description and figures provided herein. [Brief description of the drawings]

[0025] The concepts described herein are illustrated by way of example, and not by way of limitation, in the accompanying drawings. For simplicity and clarity of illustration, elements illustrated in the figures have not necessarily been drawn to scale. Where considered appropriate, reference labels have been repeated among the figures to indicate corresponding or analogous elements.

[0026] [Figure 1] 1 is a simplified flow diagram of at least one embodiment of a method for automatic generation of contact center system embeddings.

[0027] [Diagram 2] FIG. 1 illustrates at least one embodiment of a matrix representation based on contact center system agent skills and contact center system virtual queue experience that a contact center system agent may have.

[0028] [Diagram 3] FIG. 1 illustrates at least one embodiment of a neural network that can be utilized for the automated generation of contact center system embeddings.

[0029] [Figure 4]FIG. 3 illustrates at least one embodiment of a single row representation of the matrix representation of FIG. 2 for neural network training.

[0030] [Diagram 5] 2 is a simplified flow diagram of at least one embodiment of a method for routing calls to available agents based on automatic generation of contact center system embeddings using the method of FIG. 1.

[0031] [Figure 6] FIG. 2 is a simplified flow diagram of at least one embodiment of a method for assigning newly hired agents to virtual queues in a contact center system based on automatic generation of contact center system embeddings using the method of FIG. 1.

[0032] [Figure 7] FIG. 1 is a simplified block diagram of at least one embodiment of a computing system. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0033] While the concepts of the present disclosure are susceptible to various modifications and alternative forms, specific embodiments have been shown by way of example in the drawings and are described in detail herein, it being understood, however, that there is no intention to limit the concepts of the present disclosure to the particular forms disclosed, but on the contrary, the intention is to cover all modifications, equivalents, and alternatives consistent with the scope of the present disclosure and the appended claims.

[0034] References such as "one embodiment," "one embodiment," "exemplary embodiment," and the like indicate that the described embodiment may include a particular feature, structure, or characteristic, but that all embodiments may or may not necessarily include the particular feature, structure, or characteristic. Moreover, such phrases are not necessarily referring to the same embodiment. It should be further understood that references to "preferred" components or features may indicate the desirability of a particular component or feature with respect to one embodiment, but that the present disclosure is not so limited with respect to other embodiments in which such component or feature may be omitted. Moreover, when a particular feature, structure, or characteristic is described in connection with an embodiment, the implementation of such feature, structure, or characteristic in connection with other embodiments, whether or not explicitly described, is within the knowledge of one of ordinary skill in the art. Furthermore, it should be understood that an item in a list of the form "at least one of A, B, and C" can mean (A), (B), (C), (A and B), (B and C), (A and C), or (A, B, and C). Similarly, it should be understood that an item listed in the format "at least one of A, B, or C" can mean (A)(B), (C), (A and B), (B and C), (A and C), or (A, B, and C). Additionally, with respect to the claims, use of words and phrases such as "a," "an," "at least one," and / or "at least one portion" should not be construed as limiting to only one of such elements, unless specifically stated to the contrary, and use of phrases such as "at least a portion" and / or "a portion" should be construed to encompass both embodiments including only a portion of such an element and embodiments including the entirety of such an element, unless specifically stated to the contrary.

[0035] The disclosed embodiments may, in some cases, be implemented in hardware, firmware, software, or a combination thereof. The disclosed embodiments may also be implemented as instructions executed by or stored on one or more transitory or non-transitory machine-readable (e.g., computer-readable) storage media that may be read and executed by one or more processors. A machine-readable storage medium may be embodied as any storage device, mechanism, or other physical structure for storing or transmitting information in a form readable by a machine (e.g., a volatile or non-volatile memory, a media disk, or other media device).

[0036] In the figures, certain structural or method features may be shown in a particular arrangement and / or order. However, it should be understood that such a particular arrangement and / or ordering may not be required. Rather, in some embodiments, such features may be arranged in a different manner and / or order than that shown in the illustrative figures, unless indicated to the contrary. Furthermore, the inclusion of a structural or method feature in a particular figure does not imply that such feature is required in all embodiments, and in some embodiments, it may not be included or may be combined with other features.

[0037] 1, in use, a computing system 700 may execute the method 100 for automatic generation of contact center system embeddings. In some embodiments, the contact center system embeddings may include at least one of one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings. It should be understood that certain blocks of the method 100 are shown by way of example, and that such blocks may be combined or divided, added or removed, and / or reordered in whole or in part, depending on the particular embodiment, unless stated to the contrary.

[0038] A contact center system may be embodied as any system capable of providing contact center services (e.g., call center services) to end users and otherwise performing the functions described herein. It should be understood that depending on the particular embodiment, the contact center system may be located on the premises of the organization utilizing the contact center system and / or may be located remotely relative to the organization (e.g., in a cloud-based computing environment). In some embodiments, parts of the contact center system may be located on the premises / campus of the organization while other parts of the contact center system are located remotely relative to the premises / campus of the organization. As such, it should be understood that the contact center system may be deployed in equipment dedicated to the organization or its third party service provider and / or in a remote computing environment such as, for example, a private or public cloud environment with an infrastructure to support multiple contact centers for multiple enterprises. In some embodiments, the contact center system includes resources (e.g., workers, computers, and telecommunications equipment) to enable the provision of services via telephone and / or other communication mechanisms. Such services may include, for example, technical support, help desk support, emergency response, and / or other contact center services depending on the particular type of contact center.

[0039] A contact center system agent may have a particular set of skills. The value attributed to such skills may depend on the contact center system to which the agent is assigned or the assistance the agent provides to end users. In some embodiments, agent skills may include language skills, product skills, or education skills. Language skills may be skills related to the type of language the agent can speak and / or understand. In some embodiments, language skills may be at least one of English, Spanish, French, German, Russian, Portuguese, Chinese, or Hindi. For example, if a contact center system supports end users who speak English, Spanish, French, and German, an agent may have one or more language skills if such an agent speaks one or more of those languages. It should be understood that language skills may include skills related to any spoken language. Product skills may be skills related to the type of product the agent can support (i.e., provide support to end users regarding inquiries related to such products). For example, if a contact center system of company A supports company A's products X, Y, and Z, an agent who is familiar with products X and Z may possess product skills X and Z. Educational skills may be skills related to educational qualifications. It should be further appreciated that agent skills may include any variety of skills that may be necessary or useful to support a contact center system.

[0040] Contact center system virtual queues may also be used to describe agents. In some embodiments, an agent may have one or more contact center system virtual queue experiences. For example, when an end user makes a call to speak with an agent of the contact center system, the call may be placed in a virtual queue of the contact center system. An agent may be assigned to a particular virtual queue and have expertise on the subject matter represented by such virtual queue or the type of end user to whom the agent may be taking calls from such virtual queue. Unlike language skills or teaching skills, virtual queues may be more dynamic than agent skills because the virtual queues to which an agent may be assigned may change often. For example, an agent may be assigned to a virtual queue for "billing inquiries" and then reassigned a week later to another virtual queue such as "new order inquiries."

[0041] An embedding may be a mapping of a continuous number of discrete entities to a vector. A contact center system agent may be a discrete entity. An embedding may be a low-dimensional, learned continuous vector representation of a discrete entity (i.e., an agent). An embedding may reduce the dimensionality of the discrete entity and meaningfully represent variables in the transformed space. It should be appreciated that an embedding may have several purposes, including, for example, finding nearest neighbors in the embedding space, which may be used to make recommendations based on cluster categories. Additionally, an embedding may be used as an input to a machine learning model for supervised tasks. Additionally, an embedding may be used to visualize relationships between concepts and categories. Once in vector form, two agents may be compared to determine how similar such agents are to one another. Vectors may be added or subtracted to yield newer vectors. The process of converting discrete entities, such as agents, into vectors may generally involve training such data using large amounts of data and machine learning. In some embodiments, a neural network approach may be utilized to generate (e.g., automatically) a contact center system embedding.

[0042] Computing system 700 may include and / or utilize artificial intelligence, adaptive learning, bots, cognitive computing, and / or other automation techniques. For example, in some embodiments, computing system 700 may utilize one or more neural network algorithms, regression algorithms, instance-based algorithms, regularization algorithms, decision tree algorithms, Bayesian algorithms, clustering algorithms, association rule learning algorithms, deep learning algorithms, dimensionality reduction algorithms, rule-based algorithms, ensemble algorithms, artificial intelligence, and / or other suitable machine learning algorithms, artificial intelligence algorithms, techniques, and / or mechanisms.

[0043] It should be appreciated that the techniques described herein may enable leveraging an agent's skills and / or virtual queues to generate an agent profile (i.e., embedding). The agent profile / embedding may enable agents to be grouped together with similar skills and / or virtual queue experience, which may help provide a better end-user experience. For example, if an agent is not available to receive a call, the call may be routed to the most similar available agent. Similarly, as newly hired agents join the contact center system, the agent profile may be used to group agents based on their skills and / or virtual queue experience, allowing agents to become productive sooner.

[0044] It should be appreciated that prior to execution of the method 100, a user may interact with the computing system 700 via a graphical user interface of an application to upload an agent's resume, job application, and / or other information that lists the agent's skills and / or work history (i.e., virtual queues the agent has previously worked in). The exemplary method 100 may begin at block 102, where the computing system 700 may determine that the agent possesses one or more contact center system agent skills. For example, the computing system 700 may determine that the agent speaks English and Spanish (i.e., the agent has language skills) and supports products X and Y (i.e., the agent has product skills). At block 104, the computing system 700 may determine that the agent has one or more virtual queue experiences. For example, the computing system 700 may determine that the agent has experience working in virtual queues for "billing."

[0045] In block 106, the computing system 700 may generate a matrix representation based on at least one of one or more contact center system agents, one or more contact center system agent skills, and / or one or more virtual queue experiences. For example, FIG. 2 illustrates a matrix representation based on one or more contact center system agents, one or more contact center system agent skills, and / or one or more virtual queue experiences. In the example illustrated in FIG. 2, the table can be viewed as a matrix in which agents are represented as rows and agent skills and / or virtual queue experiences are represented as columns. For example, “Language Skill 1”, “Language Skill 2”, and “Language Skill 3” represent language skills. “Product Skill 1”, “Product Skill 2”, and “Product Skill 3” represent product skills, and “Virtual Queue 1”, “Virtual Queue 2”, and “Virtual Queue 3” represent virtual queue experiences. Similarly, for example, “Agent 1”, “Agent 2”, and “Agent N” represent contact center system agents. If an agent has a skill, the table is marked with “1”. If an agent does not have a skill, the table is marked with “0”. The numbers "1" and "0" in the matrix representation shown in Figure 2 are sparse because each agent may have few skills compared to the total number of skills shown in the table. For example, "Agent 1" may have "Language Skill 1" (e.g., German) and "Language Skill 3" (e.g., French), but not "Language Skill 2" (e.g., Hindi). It should be understood that the matrix representation may include any number of contact center system agents, contact center system agent skills, and desired virtual queue experiences, including, for example, tens to tens of thousands of entries (or more).

[0046] Referring to FIG. 1, at block 108, the computing system 700 may generate at least one of one or more contact center system agent identifiers, one or more contact center system agent skill identifiers, and one or more contact center system virtual queue identifiers based on the matrix representation generated at block 106. Referring back to FIG. 2, identifiers 1, 2, and 3 are generated for three different agents (e.g., “Agent 1”, “Agent 2”, “Agent 3”). If there are “N” agents, the generated identifiers may be 1 to “N”. Similarly, as shown in FIG. 2, identifiers 1, 2, and 3 are generated for three different language skills (e.g., “Language Skill 1”, “Language Skill 2”, and “Language Skill 3”), three different product skills (e.g., “Product Skill 1”, “Product Skill 2”, “Product Skill 3”), and three different virtual queue experiences (e.g., “Virtual Queue 1”, “Virtual Queue 2”, and “Virtual Queue 3”). If there are “M” agent skills and virtual queues, the generated identifiers may be 1 to “M”.

[0047] Referring back to FIG. 1 , in block 110, computing system 700 may convert at least one of one or more contact center system agent identifiers, one or more contact center system agent skill identifiers, and one or more contact center system virtual queue identifiers into at least one of one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings. As described above, constructing the embeddings may include converting discrete entities (e.g., agents) into vectors. For example, computing system 700 may convert one or more contact center system agent identifiers generated in block 108 (e.g., “Agent 1”, “Agent 2”, “Agent 3”) into one or more contact center system agent embeddings. Further, for example, computing system 700 may convert one or more contact center system agent skill identifiers generated in block 108 (e.g., “Language Skill 1”, “Language Skill 2”, etc., or “Product Skill 1”, “Product Skill 2”, etc.) into one or more contact center system agent skill embeddings. Further, for example, computing system 700 may convert one or more contact center system virtual queue identifiers generated in block 108 (e.g., “Virtual Queue 1”, “Virtual Queue 2”, etc.) into one or more contact center system virtual queue embeddings. In some embodiments, the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings may be k-dimensional vectors. In some embodiments, the k-dimensional vectors are in a five-dimensional space (i.e., K=5). In some embodiments, the k-dimensional vectors are in a ten-dimensional space (i.e., K=10).For example, in an embodiment including contact center system agents, computing system 700 may convert “agent 1,” “agent 2,” and “agent 3” into k-dimensional vectors for each agent, where K=5, each agent may be represented as a k-dimensional vector as shown below:

[0048] [1]=[0.1,0.07,-0.43,0.68,0.01]

[0049] [2]=[-0.7,0.65,0.03,-0.05,0.38]

[0050] [N]=[0.1,0.08,-0.39,0.75,0.05]

[0051] "Agent 1" may be represented as [1]. "Agent 2" may be represented as [2]. "Agent N" may be represented as [N]. Each decimal may represent a different dimension and specific weight of the three contact center system agent embeddings shown in this example. It should be appreciated that one or more contact center system agent skill embeddings and / or one or more contact center system virtual queue embeddings may similarly be represented as k-dimensional vectors in a five dimensional space, as shown in this example.

[0052] FIG. 3 illustrates a neural network that may be utilized for the automated generation of contact center system embeddings. In an exemplary embodiment, agents and skills / virtual queues (e.g., “VQ”) are two parts of a neural network. Further, agents may represent rows and skills / virtual queues (e.g., “VQ”) may represent columns in the matrix representation generated in block 106. The K arrows from each agent identifier may represent an agent embedding. Similarly, the K arrows from each skill / virtual queue may represent a skill / virtual queue embedding. The agent embedding layer and the skill / VQ embedding layer each include five boxes that represent a five-dimensional space. The computing system 700 may first generate randomly initialized weights for the five-dimensional space (i.e., the K arrows may include randomly initialized weights). For example, the computing system 700 may generate randomly initialized weights for one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings.

[0053] Referring back to FIG. 1 , in block 112, the computing system 700 may utilize machine learning to train at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings to obtain final weights for the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings by the computing system. For example, the computing system 700 may utilize backpropagation learning to obtain the final weights for the contact center system embeddings. The neural network may learn a k-dimensional vector for each of the one or more contact center system agents, the one or more contact center system agent skills, and the one or more contact center system virtual queue experiences.

[0054] Referring now to FIG. 4, computing system 700 can convert the matrix representation generated in block 106 into a single row representation for neural network training. For example, computing system 700 can transfer each agent-skill / VQ pair from the matrix representation generated in block 106 into a single row representation. The single row representation can include input data for the neural network including, for example, “Agent_id” and “Skill_id”. Referring back to FIG. 1, computing system 700 can analyze the dot product and ground truth (i.e., a0 or a1) of one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings to determine final weights for such embeddings. By completing multiple iterations of training, reliable final weights can be obtained for the contact center system agents and contact center system agent skills.

[0055] In block 114, the computing system 700 may determine / compare the similarity of at least one of the two or more of the contact center system agent embeddings, the two or more of the contact center system agent skill embeddings, and the two or more of the contact center system virtual queue embeddings based on the final weighting of at least one of the one or more of the contact center system agent embeddings, the one or more of the contact center system agent skill embeddings, and the one or more of the contact center system virtual queue embeddings by the computing system. It should be appreciated that the computing system 700 may utilize various visualization techniques to determine the similarity of the embeddings. In some embodiments, the computing system 700 may utilize at least one of cosine similarity, principal component analysis (e.g., PCA), or t-distributed stochastic neighbor embeddings (e.g., t-SNE) to determine the similarity of the embeddings. For example, referring back to FIG. 3, the computing system 700 may utilize visualization techniques to determine that Agent 1 is similar to Agent 2 and Agent 3, but is most similar to Agent 2. In an exemplary embodiment, this similarity determination may be stored in one or more databases, data structures, and / or data storage devices that may store data in computing system 700. Similarly, computing system 700 may determine that skill / VQ 1 is similar to skill / VQ 2 and skill / VQ M, but is most similar to skill / VQ 2. For example, computing system 700 may determine that an agent with language skill 2 may also have language skill 1, and the two skills may be related (e.g., adjacent languages ​​German and Polish). This similarity determination may also be stored in one or more databases, data structures, and / or data storage devices that may store data in computing system 700.

[0056] Although blocks 102-114 are described relatively sequentially, it should be understood that in some embodiments, various blocks of method 100 may be performed in parallel.

[0057] 5, during use, a computing system 700 may execute a method 500 for routing calls to available agents based on the automatic generation of contact center system embeddings. It should be understood that certain blocks of the method 500 are illustrated by way of example, and that such blocks may be combined or divided, added or removed, and / or reordered in whole or in part, depending on the particular embodiment, unless stated to the contrary. The exemplary method 500 may begin at block 502, where the computing system 700 may receive a call from an end user. At block 504, the computing system 700 may determine that the most qualified agent to answer the call from the end user is unavailable (e.g., the first agent). For example, the computing system 700 may determine that Agent 1 of FIG. 3 is unavailable. At block 506, the computing system 700 may determine that the first agent embedding is most similar to the second agent embedding (e.g., Agent 1 and Agent 2 of FIG. 3 may be most similar). For example, computing system 700 may perform method 100 (or a portion thereof) of FIG. 1 to determine embedding similarity (e.g., see block 114 of FIG. 1). At block 508, computing system 700 may route the call to an available agent (e.g., a second agent) that is best qualified to answer the call from the end user based on the determination made in block 506. For example, computing system 700 may route the call to Agent 2 of FIG. 3.

[0058] Although blocks 502-508 are described relatively sequentially, it should be understood that in some embodiments, various blocks of method 500 may be performed in parallel.

[0059] Referring now to FIG. 6, in use, a computing system 700 may perform a method 600 for assigning a newly hired agent to a virtual queue in a contact center system based on an automatic generation of contact center system embeddings. It should be understood that certain blocks of the method 600 are illustrated by way of example, and such blocks may be combined or divided, added or removed, and / or reordered in whole or in part, depending on a particular embodiment, unless stated to the contrary. The exemplary method 600 may begin at block 602, where the computing system 700 may determine that the newly hired agent possesses one or more contact center system agent skills (see, e.g., block 102 of FIG. 1). In block 604, the computing system 700 may determine that the agent has one or more virtual queue experiences (see, e.g., block 104 of FIG. 1). At block 606, the computing system 700 may generate a contact center system agent embedding for the newly hired agent and at least one of one or more contact center system agent skill embeddings and one or more contact center system virtual queue embeddings for the newly hired agent (e.g., see method 100 of FIG. 1). At block 608, the computing system 700 may determine that the newly valid agent embedding is most similar to a cluster of agent embeddings in the first virtual queue. For example, the computing system 700 may perform method 100 of FIG. 1 (or a portion thereof) to determine the embedding similarity (e.g., see block 114 of FIG. 1). At block 610, the computing system 700 may assign the newly hired agent to the first virtual queue based on the determination made at block 608.

[0060] Although blocks 602-610 are described relatively sequentially, it should be understood that in some embodiments, various blocks of method 600 may be performed in parallel.

[0061] It should be appreciated that while method 100 of FIG. 1 and method 600 of FIG. 6 may be applied to a contact center system, in one embodiment, such methods may be applied to a general enterprise. For example, when a newly hired account manager joins an enterprise, a profile may be created for that account manager based on skills and past work history derived from resumes, job applications, and / or other information. In such an embodiment, computing system 700 may assign the account manager to a particular product division within the enterprise based on similarity between the profiles of other peers in that product division.

[0062]

[0023] Referring now to Figure 7, there is shown a simplified block diagram of at least one embodiment of a computing system 700. The exemplary computing system 700 is capable of performing the method 100 of Figure 1, the method 500 of Figure 5, and / or the method 600 of Figure 6, as described in detail above. Depending on the particular embodiment, the computing system 700 may be embodied as a server, a desktop computer, a laptop computer, a tablet computer, a notebook, a netbook, an Ultrabook™, a mobile phone, a mobile computing device, a smartphone, a wearable computing device, a personal digital assistant, an Internet of Things (IoT) device, a processing system, a wireless access point, a router, a gateway, and / or any other computing, processing, and / or communication device capable of performing the functions described herein.

[0063] The computing system 700 includes a processing device 702 that executes algorithms and / or processes data in accordance with operational logic 708, an input / output device 704 that enables communication between the computing system 700 and one or more external devices 710, and a memory 706 that stores data received from the external device 710 via, for example, the input / output device 704.

[0064] The input / output devices 704 enable the computing system 700 to communicate with external devices 710. For example, the input / output devices 704 may include a transceiver, a network adapter, a network card, an interface, one or more communication ports (e.g., a USB port, a serial port, a parallel port, an analog port, a digital port, VGA, DVI, HDMI, FireWire, CAT5, or any other type of communication port or interface), and / or other communication circuitry. The communication circuitry of the computing system 700 may be configured to perform such communication using any one or more communication technologies (e.g., wireless or wired communication) and associated protocols (e.g., Ethernet, Bluetooth, Wi-Fi, WiMAX, etc.), depending on the particular computing system 700. The input / output devices 704 may include hardware, software, and / or firmware suitable for performing the techniques described herein.

[0065] External device 710 may be any type of device that allows data to be input or output from computing system 700. For example, in various embodiments, external device 710 may be embodied as another computing device, a switch, a diagnostic tool, a controller, a printer, a display, an alarm, a peripheral device (e.g., a keyboard, a mouse, a monitor / display, a touch screen display, speakers, headphones, etc.), and / or any other computing, processing, and / or communication device capable of performing the functions described herein. Additionally, it should be understood that in some embodiments, external device 710 may be integrated into computing system 700.

[0066] The processing device 702 may be embodied as any type of processor capable of performing the functions described herein. In particular, the processing device 702 may be embodied as one or more single or multi-core processors, microcontrollers, or other processors or processing / control circuits. For example, in some embodiments, the processing device 702 may include or be embodied as an arithmetic logic unit (ALU), a central processing unit (CPU), a digital signal processor (DSP), and / or another suitable processor. The processing device 702 may be of a programmable type, a dedicated hardwired state machine, or a combination thereof. A processing device 702 having multiple processing units may utilize distributed processing, pipeline processing, and / or parallel processing in various embodiments. Furthermore, the processing device 702 may be dedicated to performing only the operations described herein or may be utilized in one or more additional applications. In an exemplary embodiment, the processing device 702 is programmable and executes algorithms and / or processes data according to operating logic 708 defined by programming instructions (e.g., software or firmware) stored in memory 706. Additionally or alternatively, the operating logic 708 of the processing device 702 may be defined at least in part by hardwired logic or other hardware. Further, the processing device 702 may include one or more components of any type suitable for processing signals received from the input / output device 704, or from other components or devices, and providing a desired output signal. Such components may include digital circuits, analog circuits, or a combination thereof.

[0067] The memory 706 may be one or more types of non-transitory computer readable media, such as solid state memory, electromagnetic memory, optical memory, or combinations thereof. Additionally, the memory 706 may be volatile and / or non-volatile, and in some embodiments, some or all of the memory 706 may be portable, such as disk, tape, memory stick, cartridge, and / or other suitable portable memory. In operation, the memory 706 may store various data and software used during operation of the computing system 700, such as an operating system, applications, programs, libraries, and drivers. It should be appreciated that the memory 706 may store data that is manipulated by the operating logic 708 of the processing device 702, e.g., data representing signals received from and / or transmitted to the input / output device 704, in addition to or instead of storing programming instructions that define the operating logic 708. As shown in FIG. 7, the memory 706 may be included in and / or coupled to the processing device 702, depending on the particular embodiment. For example, in some embodiments, the processing device 702, memory 706, and / or other components of the computing system 700 may form part of a system on a chip (SoC) and be integrated into a single integrated circuit chip.

[0068] In some embodiments, various components of computing system 700 (e.g., processing device 702 and memory 706) may be communicatively coupled via an input / output subsystem, which may be embodied as circuits and / or components for facilitating input / output operations with processing device 702, memory 706, and other components of computing system 700. For example, the input / output subsystem may be embodied as or may include a memory controller hub, an input / output control hub, firmware devices, communications links (i.e., point-to-point links, bus links, wires, cables, light guides, printed circuit board traces, etc.), and / or other components and subsystems for facilitating input / output operations.

[0069] Computing system 700 may, in other embodiments, include other or additional components, such as those typically found in a typical computing device (e.g., various input / output devices and / or other components). It should be further understood that one or more of the components of computing system 700 described herein may be distributed across multiple computing devices. In other words, the techniques described herein may be employed by a computing system including one or more computing devices. Furthermore, while only a single processing device 702, I / O device 704, and memory 706 are illustratively shown in FIG. 7, it should be understood that in other embodiments, a particular computing system 700 may include multiple processing devices 702, I / O devices 704, and / or memory 706. Furthermore, in some embodiments, multiple external devices 710 may be in communication with computing system 700.

Claims

1. 1. A method for automated generation of contact center system embeddings, the method comprising: determining, by a computing system, at least one of one or more contact center system agents, one or more contact center system agent skills, and one or more contact center system virtual queue experiences; generating, by the computing system, a matrix representation based on the at least one of the one or more contact center system agents, the one or more contact center system agent skills, and the one or more contact center system virtual queue experiences; generating, by the computing system and based on the matrix representation, at least one of one or more contact center system agent identifiers, one or more contact center system agent skill identifiers, and one or more contact center system virtual queue identifiers; converting, by the computing system, the at least one of the one or more contact center system agent identifiers, the one or more contact center system agent skill identifiers, and the one or more contact center system virtual queue identifiers into at least one of one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings, wherein a weight of at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings is randomly initialized; training, by the computing system, the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings by applying machine learning to obtain final weights for the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings; A method comprising:

2. 2. The method of claim 1, further comprising: determining, by the computing system, a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings based on comparing the final weights of the at least one of the one or more of the contact center system agent embeddings, the one or more of the contact center system agent skill embeddings, and the one or more of the contact center system virtual queue embeddings.

3. 3. The method of claim 2, wherein determining the similarity of the at least one of the two or more contact center system agent embeddings, the two or more contact center system agent skill embeddings, and the two or more contact center system virtual queue embeddings comprises utilizing at least one of cosine similarity, principal component analysis, or t-distribution stochastic neighbor embeddings.

4. routing the telephony call to a contact center system agent in response to determining the similarity of the at least one of the two or more of the contact center system agent embeddings, the two or more of the contact center system agent skill embeddings, and the two or more of the contact center system virtual queue embeddings; assigning contact center system agents to virtual queues in response to determining the similarity of the at least one of the two or more of the contact center system agent embeddings, the two or more of the contact center system agent skill embeddings, and the two or more of the contact center system virtual queue embeddings; The method of claim 2 , further comprising at least one of:

5. The method of claim 1 , wherein the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings comprises a k-dimensional vector.

6. The method of claim 5 , wherein the k-dimensional vector is in a five-dimensional space or a ten-dimensional space.

7. The method of claim 1 , wherein at least one of the converting and the training is based on a neural network.

8. The method of claim 1 , further comprising converting the matrix representation to a single row representation for training.

9. 2. The method of claim 1 , wherein applying machine learning to obtain the final weights comprises determining a dot product and ground truth of the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings.

10. The method of claim 1 , wherein the one or more contact center system agent skills include at least one of one or more language skills, one or more product skills, or one or more instructional skills.

11. 1. A system for automated generation of contact center system embeddings, the system comprising: At least one processor; at least one memory having a plurality of instructions stored therein, the instructions, upon execution by the at least one processor, providing the system with: determining, by a computing system, at least one of one or more contact center system agents, one or more contact center system agent skills, and one or more contact center system virtual queue experiences; generating, by the computing system, a matrix representation based on the at least one of the one or more contact center system agents, the one or more contact center system agent skills, and the one or more contact center system virtual queue experiences; generating, by the computing system, at least one of one or more contact center system agent identifiers, one or more contact center system agent skill identifiers, and one or more contact center system virtual queue identifiers based on the matrix representation; converting, by the computing system, the at least one of the one or more contact center system agent identifiers, the one or more contact center system agent skill identifiers, and the one or more contact center system virtual queue identifiers into at least one of one or more contact center system agent embeddings, one or more contact center system agent skill embeddings, and one or more contact center system virtual queue embeddings, wherein a weight of at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings is randomly initialized; at least one memory for causing the computing system to train the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings by applying machine learning to obtain final weights for the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings; A system comprising:

12. The instructions further include:

12. The system of claim 11, further comprising: causing the computing system to determine a similarity of at least one of two or more of the contact center system agent embeddings, two or more of the contact center system agent skill embeddings, and two or more of the contact center system virtual queue embeddings based on comparing the final weights of the at least one of the one or more of the contact center system agent embeddings, the one or more of the contact center system agent skill embeddings, and the one or more of the contact center system virtual queue embeddings.

13. 13. The system of claim 12, wherein determining the similarity of the at least one of the two or more contact center system agent embeddings, the two or more contact center system agent skill embeddings, and the two or more contact center system virtual queue embeddings comprises utilizing at least one of cosine similarity, principal component analysis, or t-distribution stochastic neighbor embeddings.

14. The instructions further include: routing a telephony call to a contact center system agent in response to determining the similarity of the at least one of the two or more of the contact center system agent embeddings, the two or more of the contact center system agent skill embeddings, and the two or more of the contact center system virtual queue embeddings; or causing a contact center system agent to be assigned to a virtual queue in response to determining the similarity of the at least one of the two or more of the contact center system agent embeddings, the two or more of the contact center system agent skill embeddings, and the two or more of the contact center system virtual queue embeddings. The system of claim 12.

15. 12. The system of claim 11, wherein the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings comprises a k-dimensional vector.

16. The system of claim 15 , wherein the k-dimensional vector is in a five-dimensional space or a ten-dimensional space.

17. The system of claim 11 , wherein at least one of the converting and the training is based on a neural network.

18. The instructions further include: The system of claim 11 , further comprising: converting the matrix representation into a single row representation for training.

19. 12. The system of claim 11, wherein applying machine learning to obtain the final weights comprises determining a dot product and ground truth of the at least one of the one or more contact center system agent embeddings, the one or more contact center system agent skill embeddings, and the one or more contact center system virtual queue embeddings.

20. The system of claim 11 , wherein the one or more contact center system agent skills include at least one of one or more language skills, one or more product skills, or one or more instructional skills.

Citation Information

Patent Citations

  • Device and method for automatically allocating call center agent to skill in call center

    JP1998224478A

  • Support system for product and support method for product

    JP2010218266A

  • Systems and methods for automating skill set addition

    JP2016507844A

  • Self-learning adaptive routing system

    JP2020520594A