GENERATING FEATURELESS SERVICE PROVIDER MATCHES
The present disclosure relates to systems, non-transitory computer-readable media, and methods that utilize featureless supervised or unsupervised machine learning to generate service provider match predictions based on raw communication data. In one or more embodiments, the disclosed systems utilize a trained service provider match neural network to generate a service provider match prediction based on raw communication data associated with an incoming audio, text, or video communication. In response to the generated service provider match prediction, the disclosed systems can route the incoming communication to the matched service provider.
This application is a continuation of U.S. patent application Ser. No. 16/549,186 filed on Aug. 23, 2019. The contents of which are hereby incorporated by reference.
BACKGROUNDRecent years have seen significant improvement in hardware and software platforms for communications routing. Indeed, conventional systems identify attributes or profiles associated with an incoming communication and utilize those attributes or profiles to determine how the communication should be routed. For example, in response to receiving an incoming text message, conventional systems can analyze the text message to identify a geographical origin of the text message, and can then route the text message to a service provider (e.g., a customer service agent) who is located within the same geographical area.
Although conventional systems can route communications, this attribute-based approach gives rise to a number of technological problems in accuracy, flexibility, and efficiency. For example, the attribute-based routing methodology utilized by conventional systems relies on a host of arbitrary assumptions and often leads to inaccurate service provider matches. In such circumstances, conventional systems fail to accurately route a communication to the most appropriate service provider simply because attributes of the communication happened to correspond with a different service provider. This approach can lead to routing a communication to a wrong service provider, leading to multiple routing attempts, which results in an inefficient use of communication resources.
Moreover, conventional systems are rigid and inflexible. Indeed, as just mentioned, conventional systems are rigidly tied to predefined connections between attributes and profiles associated with an incoming communication and attributes and profiles associated with a particular service provider. Because of this inflexibility, conventional systems often fail to route communications in such a way that will maximize target metrics that are customized to a particular organization or that will increase the accuracy and efficiency of routing a communication to the most appropriate service provider available.
In addition to shortcomings with regard to accuracy and flexibility, conventional systems are also inefficient. As an initial matter, conventional systems expend significant resources in analyzing commonalties in attributes and profiles between incoming communications and service providers to route the incoming communications. This leads to inefficient utilization of computational resources by routing the incoming communications in a way that is not truly optimized to the specific goals and metrics of an organization.
Thus, there are several technical problems with regard to conventional communication routing systems.
SUMMARYOne or more embodiments provide benefits and/or solve one or more of the foregoing or other problems in the art with systems, methods, and non-transitory computer readable storage media that utilize featureless supervised or unsupervised machine learning to accurately, flexibly, and efficiently route incoming communications to matched service providers so as to optimize one or more metrics customized to a particular organization. In particular, an unsupervised matching system, as described herein, generates matches between an incoming communication and service providers without relying on any attribute or profile extracted from or associated with either the communication or the service provider.
For example, in one or more embodiments, the disclosed systems train a service provider match neural network with training inputs that include raw communication training data that is untethered from any additional information associated with the communications themselves. By applying the trained service provider match neural network to raw communication data associated with a new incoming communication, the unsupervised matching system can generate a service provider match prediction that will accurately, flexibly, and efficiently utilize computing resources in optimizing one or more target metrics associated with the incoming communication.
Additional features and advantages of one or more embodiments of the present disclosure will be set forth in the description that follows, and in part will be obvious from the description, or may be learned by practice of such example embodiments.
The detailed description is described with reference to the accompanying drawings in which:
One or more embodiments of the present disclosure includes a featureless matching system that utilizes machine learning (e.g., supervised learning, unsupervised learning, trained neural networks) to accurately generate service provider match predictions such that an incoming communication can be efficiently routed to the predicted service provider. For example, the unsupervised matching system can train a service provider match neural network with raw communication data to generate service provider match predictions. Upon training, and when provided with a new incoming communication, the service provider match neural network can generate a service provider match prediction that is based on raw communication data, rather than attribute or profile information associated with the communication or the predicted service provider. Thus, the unsupervised matching system avoids the pitfalls of conventional systems by accurately, flexibly, and efficiently generating service provider matches and routing incoming communications without relying on arbitrary and pre-defined rules and connections.
To illustrate, in one or more embodiments, the unsupervised matching system relies on raw communication data. For example, conventional systems may extract attributes associated with an incoming communication such as a location of origin (e.g., determined based on a phone number area code), a user profile (e.g., associated with a user ID), a time of day (e.g., taked from an email header), and so forth. A conventional system may then utilize these extracted attributes to route the communication to a service provider who satisfies a pre-defined rule associated with those extracted attributes.
Instead of this attribute-based approach, various embodiments of the unsupervised matching system matches and routes an incoming communication based on raw communication data associated with that communication. For example, if the incoming communication is a telephone call, the unsupervised matching system utilizes raw audio data associated with that telephone call to route the incoming communication. Similarly, if the incoming communication is a video communication or a text communication, the unsupervised matching system utilizes the raw video data or raw text data associated with the communication to route the incoming communication.
As mentioned, in at least one embodiment, the unsupervised matching system provides the raw communication data associated with an incoming communication to a trained service provider match neural network to generate a service provider match prediction. In one or more embodiments, the unsupervised matching system trains the service provider match neural network to generate service provider match predictions utilizing raw communication training data and corresponding service provider match markers. Specifically, the unsupervised matching system applies the service provider match neural network to input vectors associated with raw communication training data. Upon receiving a service provider match prediction from the service provider match neural network, the unsupervised matching system compares the service provider match prediction to the service provider match marker associated with the raw communication training data. The unsupervised matching system can then modify parameters of the service provider match neural network based on the comparison.
After training the service provider match neural network, the unsupervised matching system can apply the service provider match neural network to new incoming communications in order to generate service provider match predictions. With each prediction, the unsupervised matching system can either route the communication to the predicted service provider, or can provide the service provider match prediction to a third-party customer request system for further use. In at least one embodiment, the unsupervised matching system may determine that a predicted service provider is not available to receive an incoming communication. In that event, the unsupervised matching system can modify one or more parameters of the service provider match neural network in order to generate a second service provider match prediction so that the unsupervised matching system routes the communication to the best available service provider based on the raw communication data.
The unsupervised matching system provides many advantages and benefits over conventional systems and methods. For example, by utilizing raw communication data to generate service provider match predictions, the unsupervised matching system improves accuracy relative to conventional systems. Specifically, the unsupervised matching system can accurately generate a service provider match prediction for an incoming communication, even when no additional attributes are known about incoming communication. Rather, the unsupervised matching system generates service provider match predictions based wholly on the raw communication data associated with the communication.
Moreover, by avoiding the attribute-based approach of conventional systems, the unsupervised matching system improves flexibility relative to conventional systems. For example, as discussed above, conventional systems are rigidly tied to attribute-based rules and pre-defined connections in routing incoming communications. As such, conventional systems are unable to route incoming communications with no associated attributes, or simply routes the communication based on no logic. Conversely, the unsupervised matching system flexibly matches and routes a communication based on that communication's raw communication data. Furthermore, by generating service provider match predictions based on raw communication data, the unsupervised matching system improves efficiency relative to conventional systems. Specifically, the unsupervised matching system does not waste system resources in routing communications to service providers who are poor matches, which causes additional transfers and/or additional communication based on the poor match.
As illustrated by the foregoing discussion, the present disclosure utilizes a variety of terms to describe features and advantages of the unsupervised matching system. Additional detail is now provided regarding the meaning of such terms. For example, as used herein, the term “communication” refers to any type of communication that the unsupervised matching system receives from a contacting user. For example, a communication can be a phone call (e.g., VoIP, cellular, POTS), a chat message, a text message (e.g., SMS), video call data (e.g., a video stream), an instant message (e.g., via social media), or any other type of electronic communication.
As mentioned above, the unsupervised matching system utilizes raw communication data associated with a communication. As used herein, the term “raw communication data” refers to non-attribute based information associated with a communication. For example, raw communication data extracted from a phone call communication can include a digital recording of at least a portion of the audio of the phone call. Similarly, raw communication data extracted from a text message can include digital information representing the textual contents of the message. Similarly, raw communication data can include an analysis of video data and/or image data.
Additionally, as used herein, the term “contacting user” refers to a user that contacts the unsupervised matching system. A contacting user can refer to a user that is experiencing a problem or issue with a service system and that contacts the unsupervised matching system in an attempt to resolve the problem or to inform others of the problem. For example, a contacting user can include a caller, a text chat user, a video chat user, or a text message user that sends a communication to the unsupervised matching system. As used herein, the term “service provider” refers to a user who provides a specific type of service via phone, text, and/or video. For example, a service provider may be a customer service representative who provides technical support to contacting users over the telephone. In another example, a service provider may be a personal shopper who curates clothing choices for a contacting user over a video chat.
As mentioned above, the unsupervised matching system can utilize various machine learning techniques to generate service provider match predictions. For example, in one embodiment, the unsupervised matching system may not be unsupervised. For instance, the matching system may utilize supervised machine learning to generate provider match predictions from raw data. To illustrate, in supervised machine learning, the matching system utilizes known inputs and known outputs (e.g., monitored raw data and match predictions collected over time) to “learn” a mapping function such that the matching system can predict new outputs (e.g., new provider match predictions) when presented with new inputs (e.g., new raw data).
Alternatively, the matching system may be unsupervised. For example, the matching system may utilize known inputs with no corresponding outputs in order to model underlying functions such that when presented with new inputs (e.g., new raw data), the model can generate provider match predictions. In some embodiments, this type of unsupervised learning is also called “deep learning” and may include one or more neural networks. Although as described herein, the unsupervised matching system utilizes one or more neural networks to generate provider match predictions, it will be understood that in alternative embodiments, the matching system can utilize other types of machine learning to generate provider match predictions, as discussed above.
As used herein, the term “neural network” refers to a machine learning model that can be tuned (e.g., trained) based on inputs to approximate unknown functions. In particular, the term neural network can include a model of interconnected artificial neurons (or layers) that communicate and learn to approximate complex functions and generate outputs based on a plurality of inputs provided to the model. In particular, a neural network includes a computer-implemented algorithm that implements deep learning techniques to analyze input (e.g., training input encoded as a neural network input vector) to make predictions and that improves in accuracy by comparing generated predictions against ground truth data and modifying internal parameters for subsequent predictions. In some embodiments, a neural network can employ supervised learning, while in other embodiments a neural network can employ unsupervised learning or reinforced learning. Examples of neural networks include deep convolutional neural networks, generative adversarial neural networks, and recurrent neural networks.
Relatedly, the term “train” refers to utilizing information to tune or teach a neural network. The term “training” (used as an adjective or descriptor, such as “training data”) refers to information or data utilized to tune or teach the neural network. In some embodiments, the unsupervised matching system trains the service provider match neural network to generate accurate service provider match predictions based on respective training data.
As mentioned above, the unsupervised matching system trains the service provider match neural network with raw communication training data and corresponding service provider match markers. As used herein, “raw communication training data” refers to raw communication data associated with a communication that has previously been matched with and/or routed to a service provider with optimized results. Also as used herein, a “service provider match marker” refers to a ground truth associated with the corresponding raw communication data. To illustrate, a service provider match marker can include a service provider ID and/or information associated with the service provider to whom the communication associated with the corresponding raw communication training data was previously routed.
Additional detail regarding the communication management system will now be provided with reference to the figures. For example,
As shown in
As mentioned, the environment includes different user devices such as the client devices 108a-108d and the service provider devices 112a-112d. The client devices 108a-108d can be a one of a variety of computing devices, including a smartphone, tablet, smart television, desktop computer, laptop computer, virtual reality device, augmented reality device, or other computing device as described in relation to
As shown, each of the client devices 108a-108d include a communication application (e.g., the communication applications 110a, 110b, 110c, and 110d, respectively). In particular, the communication applications 110a-110d may be web applications, native applications installed on the client devices 108a-108d (e.g., mobile applications, desktop applications, etc.), or cloud-based applications where part of the functionality is performed by the server(s) 106 and/or the third party server(s) 120. The communication applications 110a-110d can present or display information to a contacting user, including information relating to options for creating and processing a communication to help a contacting user accomplish a desired purpose. A contacting user can interact with one of the communication applications 110a-110d to provide a communication that, for example, includes a customer service request.
As also mentioned, the environment includes the service provider devices 112a, 112b, 112c, and 112d. Like the client devices 108a-108d, the service provider devices 112a-112d can each be one of a variety of computing devices, including a smartphone, tablet, smart television, desktop computer, laptop computer, virtual reality device, augmented reality device, or other computing device as described in relation to
As shown, the service provider devices 112a-112d include an agent application (e.g., the agent applications 114a, 114b, 114c, and 114d, respectively). In particular, each of the agent applications 114a-114d may be a web application, a native application installed on an agent device (e.g., a mobile application, a desktop application, etc.), or a cloud-based application where part of the functionality is performed by the server(s) 106 and/or the third party server(s) 120. The agent applications 114a-114d can present or display information to a service provider, including a user interface for interacting with a contacting user and presenting information pertaining to the contacting user (e.g., a routed communication, raw communication data, etc.). A service provider can interact with one of the agent applications 114a-114d to, for example, help a contacting user reschedule a flight or replace a defective product.
As illustrated in
As further illustrated in
Although
In some embodiments, though not illustrated in
As discussed above, the unsupervised matching system 102 can generate service provider match predictions based on raw communication data. For instance,
As shown in
Based on the output of the service provider match neural network, the unsupervised matching system 102 performs the act 206 of routing the communication to the matched service provider. For example, in one embodiment, the unsupervised matching system 102 routes a communication directly to the matched service provider. In that case, the unsupervised matching system 102 can provide the communication to the matched service provider via the agent application (e.g., one of the agent applications 114a-114d shown in
In response to receiving the communication 302, the unsupervised matching system 102 performs an act 304 of extracting raw communication data. In one or more embodiments, the unsupervised matching system 102 extracts the raw communication data directly from the received communication. For example, the unsupervised matching system 102 can sample a portion of a live analog audio stream and can convert the sample to a digital file. More specifically, the unsupervised matching system 102 can convert the same to a digital file by mapping the height of the analog sound waves to a set of positive and negative integers that replicate the wave in digital form.
Additionally or alternatively, rather than extracting the raw communication data directly from the received communication, the unsupervised matching system 102 can utilize the communication from a client device to identify stored raw communication data associated with the client device. For example, in at least one embodiment, the communication management system 104 and/or the customer request system 116 stores raw communication data associated with client devices in a central repository (e.g., the customer database 118). In that embodiment, upon receiving the communication from the client device, the unsupervised matching system 102 can determine a user identifier (e.g., a phone number, a profile name, an account number) associated with the client device. The unsupervised matching system 102 can then utilize the user identifier to extract the corresponding raw communication data from the repository.
With the raw communication data associated with the communication extracted, the unsupervised matching system 102 provides the raw communication data to the service provider match neural network 306. As shown in
After generating the input vector, the service provider match neural network 306 performs the act 310 of processing the input vector. For example, the service provider match neural network 306 can process the input vector by providing the input vector to an input neural network layer, and causing the subsequent layer output to be provided to subsequent processing layers within the service provider match neural network 306 (e.g., LSTM layers, BiMax pooling layers, etc.).
Next, the service provider match neural network 306 performs the act 312 of generating a service provider match prediction. As mentioned above, the one or more processing layers of the service provider match neural network 306 ultimately output a service provider match prediction. In one or more embodiments, service provider match neural network 306 generates a service provider match prediction including a service provider identifier (e.g., an employee number, a service provider device ID). Additionally or alternatively, the service provider match neural network 306 can generate a service provider match prediction including a service provider's name, or characteristics of the service provider, as will be explained in detail below.
With the service provider match prediction, the unsupervised matching system 102 can perform the act 314 of routing the communication. As mentioned above, the unsupervised matching system 102 can route the communication directly to the predicted service provider. Additionally or alternatively, the unsupervised matching system 102 can route the communication by providing the service provider match prediction to the customer request system 116 hosted on the third-party server 120.
As illustrated in
In one or more embodiments, the unsupervised matching system 102 may store the collected raw communication data in various ways. For example, the unsupervised matching system 102 may store raw communication data based on user IDs. For instance, a client's user ID may be a telephone number (e.g., a telephone number associated with the client's client device if the client device is capable of making phone calls). Additionally or alternatively, a client's user ID may be an account number, a profile name, an email address, a screen name, a date of birth, or any other unique identifier.
In at least one embodiment, the unsupervised matching system 102 determines a user ID associated with a communication by analyzing data included in and/or associated with the communication. For example, if the communication is an email, the unsupervised matching system 102 may determine the user ID associated with the email by analyzing header information included with the email to extract an email address associated with the sender. In another example, if the communication is an incoming video stream, the unsupervised matching system 102 may determine the user ID associated with the video stream by analyzing data packet information associated with the video stream to extract an IP address, an account number, or a screen name. Moreover, in some embodiments, if the communication includes visual data (e.g., such as with a digital image or video), the unsupervised matching system 102 may determine the user ID associated with the communication by performing image analysis techniques (e.g., facial recognition) in connection with the communication.
Once the unsupervised matching system 102 has determined a user ID associated with the received communication, the unsupervised matching system 102 can perform the act 406 of querying raw communication associated with the user ID from the repository of raw communication data. For example, as shown in
In one or more embodiments, the unsupervised matching system 102 queries a predetermined unit of raw communication data from the customer database 118 that is associated with the user ID. For example, if the received communication is a phone call, the unsupervised matching system 102 may query ten seconds of raw audio data associated with the calling client from the customer database 118. In another example, if the received communication is an SMS text message, the unsupervised matching system 102 may query 50 characters of raw textual data associated with the calling client from the customer database 118.
After querying the raw communication data from the data repository, the unsupervised matching system 102 performs the act 410 of generating an input vector. As discussed above, the unsupervised matching system 102 generates an input vector for the service provider match neural network 306 that is configured or packaged in a manner that is customized to the input layer of the service provider match neural network 306. It should be noted that, at this point, the unsupervised matching system 102 no longer relies on the determined user ID in generating the service provider match prediction. Rather, as discussed above, the service provider match neural network 306 generates the service provider match prediction based on the raw communication data included in the input vector. As such, the unsupervised matching system 102 next performs the acts 412 and 414 of applying the service provider match neural network 306 to the generated input vector and generating the service provider match prediction, as discussed above.
For example, as illustrated in
At this point, the unsupervised matching system 102 performs the act 424 of determining whether the matched service provider is available to receive the communication. In one or more embodiments, the unsupervised matching system 102 makes this determination by providing the generated service provider match predication to the customer request system 116 and then receiving an indication from the customer request system 116 that the matched service provider is unavailable. Additionally or alternatively, the unsupervised matching system 102 can make this determination by contacting the matched service provider directly and receiving an indication from the service provider that he or she is unavailable. Additionally or alternatively, the unsupervised matching system 102 can make this determination by contacting the matched service provider directly and receiving no response from the matched service provider after a threshold amount of time. Additionally or alternatively, the unsupervised matching system 102 can make this determination by accessing information (e.g., timekeeping information, human resources information, etc.) that indicates whether a matched service provider is out sick, on vacation, or on a scheduled break.
If the unsupervised matching system 102 determines that the matched service provider is available (e.g., “Yes”), the unsupervised matching system 102 performs the act 426 of sending the service provider ID (e.g., indicated by the generated service provider match prediction) to the customer request system 116. As discussed above, in the event that the unsupervised matching system 102 does not contact the service providers directly, the unsupervised matching system 102 can provide match and routing information to the customer request system 116 hosted by the third-party server(s) 120.
If the unsupervised matching system 102 determines that the matched service provider is not available (e.g., “No”), the unsupervised matching system 102 again performs the act 422 of applying the service provider match neural network to the generated input vector. For example, in at least one embodiment, the service provider match neural network 306 includes latent memory features that “remember” matched service providers for a threshold amount of time. As such, when the unsupervised matching system 102 applies the service provider match neural network 306 to the same input vector within a predetermined period of time (e.g., 5 seconds), the service provider match neural network can remove the previously matched service provider from consideration. In that scenario, the service provider match neural network 306 can generate a service provider match prediction that features a different service provider. In this way, the service provider match neural network 306 avoids unnecessary lag.
Additionally or alternatively, in at least one embodiment, the unsupervised matching system 102 can provide service provider available information as part of the input vector. For example, the unsupervised matching system 102 can generate the input vector (e.g., in the act 420) to include raw communication data associated with the received communication as well as human resources information indicating the service providers currently available. In at least one embodiment, the unsupervised matching system 102 can update the input vector each time the service provider match neural network 306 is applied to include the most up-to-date availability information. In one or more embodiments, the unsupervised matching system 102 can continue to apply the service provider match neural network to the input vector until the unsupervised matching system 102 determines that the matched service provider is available in act 424.
As discussed above, the unsupervised matching system 102 can train a service provider match neural network 306 to generate service provider match predictions. For instance,
As shown in
Each instance of raw communication match data 504 corresponds to a service provider match marker 506. For example, every sample of recorded audio corresponds to a service provider match marker 506 that includes a ground truth (e.g., a service provider ID, or attributes of the predicted service provider) associated with that sample of recorded audio. Similarly, every portion of textual data and/or video data in the raw communication match data 504 corresponds to a service provider match marker 506 including the ground truth associated with that sample.
To begin training the service provider match neural network 306, the unsupervised matching system 102 provides an instance of raw communication match data 504 to the service provider match neural network 306. As discussed above, the service provider match neural network 306 may include the functionality to generate an input vector from received raw communication data. Accordingly, in response to receiving the instance of raw communication match data 504, the service provider match neural network 306 may generate an input vector, and pass the generated input vector to an input layer of the service provider match neural network 306.
Depending on the structure of the service provider match neural network 306, the service provider match neural network 306 may process the generated input vector in a number of ways. For example, in one embodiment, the service provider match neural network 306 may pass latent feature vectors between sequential processing layers of the service provider match neural network 306. Additionally, the service provider match neural network 306 may pool prediction results (e.g., in a max pooling layer), and generate a service provider match prediction.
After the service provider match neural network 306 generates a service provider match prediction, the unsupervised matching system 102 continues to train the service provider match neural network 306 by performing the act 508 of comparing the service provider match prediction to a corresponding service provider match marker. For example, as discussed above, for each instance of raw communication match data 504, the match prediction training data 502 includes a corresponding service provider match marker 506. Each service provider match marker 506 includes a ground truth indicating a service provider to whom a corresponding communication was successfully routed.
As mentioned above, a communication is “successfully” routed when a positive metric results from the communication being routed to a particular service provider. For example, depending on the type of service provided by a service provider, a positive metric may be a successfully resolved customer service issue, a successful sale, a positive product review, a first-time service scheduling, a re-scheduling of a service, a re-booking of a ticket, and so forth.
As such, the act 508 can involve the unsupervised matching system 102 determining whether the service provider match prediction generated by the service provider match neural network 306 is the same as the service provider match marker 506 corresponding to the raw communication match data 504 provided to the service provider match neural network 306. The unsupervised matching system 102 can determine that the generated prediction is the same as the corresponding marker by utilizing a direct character comparison (e.g., via OCR). In additional embodiments, the unsupervised matching system 102 can determine that the generated prediction is the same as the corresponding marker if the generated prediction is “close enough.” For example, if the generated prediction and the corresponding marker both include service provider attributes (e.g., nationality, age, sex) rather than a specific service provider ID, the unsupervised matching system 102 can determine that the prediction and the marker are the same if a threshold number of attributes match.
To continue training the service provider match neural network 306, the unsupervised matching system 102 performs the act 510 of modifying parameters of the service provider match neural network 306 based on the comparison performed in the act 508. For example, based on this comparison, the unsupervised matching system 102 can modify parameters of one or more layers of the service provider match neural network 306 to reduce the measure of loss.
In one or more embodiments, the unsupervised matching system 102 continues training the service provider match neural network 306 until either the match prediction training data 502 is exhausted or the measure of loss is minimized and stable over a threshold number of training cycles. The unsupervised matching system 102 may periodically retrain the service provider match neural network 306 in the same manner illustrated in
As described in relation to
Each of the components 602-608 of the unsupervised management system 102 can include software, hardware, or both. For example, the components 602-608 can include one or more instructions stored on a computer-readable storage medium and executable by processors of one or more computing devices, such as a client device or server device. When executed by the one or more processors, the computer-executable instructions of the unsupervised management system 102 can cause the computing device(s) to perform the methods described herein. Alternatively, the components 602-608 can include hardware, such as a special-purpose processing device to perform a certain function or group of functions. Alternatively, the components 602-608 of the unsupervised management system 102 can include a combination of computer-executable instructions and hardware.
Furthermore, the components 602-608 of the unsupervised management system 102 may, for example, be implemented as one or more operating systems, as one or more stand-alone applications, as one or more modules of an application, as one or more plug-ins, as one or more library functions or functions that may be called by other applications, and/or as a cloud-computing model. Thus, the components 602-608 may be implemented as a stand-alone application, such as a desktop or mobile application. Furthermore, the components 602-608 may be implemented as one or more web-based applications hosted on a remote server. The components 602-608 may also be implemented in a suite of mobile device applications or “apps.”
As mentioned above, and as shown in
As further shown in
As mentioned above, and as shown in
Furthermore, as mentioned above and as shown in
Although the unsupervised matching system 102 is described herein as generating service provider match predictions utilizing a service provider match neural network, in additional embodiments, the unsupervised matching system 102 can generate service provider match predictions in other ways. For example, in one alternative embodiment, the unsupervised matching system 102 can generate service provider match predictions using data segmentation and topic modeling. To illustrate, based on a defined optimization goal (e.g., sales), the unsupervised matching system 102 can develop clusters of raw communication data and corresponding service provider match predictions. The unsupervised matching system 102 can then identify one or more characteristics of each data cluster. Then when a new communication is received, the unsupervised matching system 102 can determine the best data cluster with which to associated the new communication based on one or more characteristics of the raw communication data associated with the new communication, and can generate a service provider match prediction based on the service provider match predictions within the identified cluster.
As mentioned,
As shown in
Also as shown in
The series of acts 700 also includes an act 730 of generating a service provider match prediction by providing the input vector to an unsupervised matching system. For example, the act 730 can involve generating a service provider match prediction by: providing the input vector to an unsupervised matching system, wherein the unsupervised matching system is trained to generate a service provider match from raw communication training data.
Additionally, the series of acts 700 includes an act 740 of providing the service provider match prediction to a customer request system. For example, the act 740 can involve, in response to generating the service provider match prediction, providing the service provider match prediction to a customer request system.
In one or more embodiments, the series of acts 700 further includes, prior to receiving the communication from the client device associated with the contacting user, training the unsupervised matching system to generate the service provider match. For example, training the unsupervised matching system can include: generating a plurality of training input vectors based on raw communication training data, wherein each training input vector is associated with a corresponding service provider match marker, and for each of the plurality of training input vectors: providing the training input vector to the unsupervised matching system to generate a service provider match prediction, and comparing the service provider match prediction with the corresponding service provider match marker to modify parameters of the unsupervised matching system.
In one or more embodiments, the series of acts 700 further includes: receiving, from the customer request system, an indication that the service provider match prediction cannot be accommodated, in response to receiving the indication, modifying one or more parameters of the unsupervised matching system, and generating a second service provider match prediction utilizing the unsupervised matching system.
Embodiments of the present disclosure may comprise or utilize a special purpose or general-purpose computer including computer hardware, such as, for example, one or more processors and system memory, as discussed in greater detail below. Embodiments within the scope of the present disclosure also include physical and other computer-readable media for carrying or storing computer-executable instructions and/or data structures. In particular, one or more of the processes described herein may be implemented at least in part as instructions embodied in a non-transitory computer-readable medium and executable by one or more computing devices (e.g., any of the media content access devices described herein). In general, a processor (e.g., a microprocessor) receives instructions, from a non-transitory computer-readable medium, (e.g., a memory, etc.), and executes those instructions, thereby performing one or more processes, including one or more of the processes described herein.
Computer-readable media can be any available media that can be accessed by a general purpose or special purpose computer system. Computer-readable media that store computer-executable instructions are non-transitory computer-readable storage media (devices). Computer-readable media that carry computer-executable instructions are transmission media. Thus, by way of example, and not limitation, embodiments of the disclosure can comprise at least two distinctly different kinds of computer-readable media: non-transitory computer-readable storage media (devices) and transmission media.
Non-transitory computer-readable storage media (devices) includes RAM, ROM, EEPROM, CD-ROM, solid state drives (“SSDs”) (e.g., based on RAM), Flash memory, phase-change memory (“PCM”), other types of memory, other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store desired program code means in the form of computer-executable instructions or data structures and which can be accessed by a general purpose or special purpose computer.
A “network” is defined as one or more data links that enable the transport of electronic data between computer systems and/or modules and/or other electronic devices. When information is transferred or provided over a network or another communications connection (either hardwired, wireless, or a combination of hardwired or wireless) to a computer, the computer properly views the connection as a transmission medium. Transmissions media can include a network and/or data links which can be used to carry desired program code means in the form of computer-executable instructions or data structures and which can be accessed by a general purpose or special purpose computer. Combinations of the above should also be included within the scope of computer-readable media.
Further, upon reaching various computer system components, program code means in the form of computer-executable instructions or data structures can be transferred automatically from transmission media to non-transitory computer-readable storage media (devices) (or vice versa). For example, computer-executable instructions or data structures received over a network or data link can be buffered in RAM within a network interface module (e.g., a “NIC”), and then eventually transferred to computer system RAM and/or to less volatile computer storage media (devices) at a computer system. Thus, it should be understood that non-transitory computer-readable storage media (devices) can be included in computer system components that also (or even primarily) utilize transmission media.
Computer-executable instructions comprise, for example, instructions and data which, when executed at a processor, cause a general-purpose computer, special purpose computer, or special purpose processing device to perform a certain function or group of functions. In some embodiments, computer-executable instructions are executed on a general-purpose computer to turn the general-purpose computer into a special purpose computer implementing elements of the disclosure. The computer executable instructions may be, for example, binaries, intermediate format instructions such as assembly language, or even source code. Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the described features or acts described above. Rather, the described features and acts are disclosed as example forms of implementing the claims.
Those skilled in the art will appreciate that the disclosure may be practiced in network computing environments with many types of computer system configurations, including, personal computers, desktop computers, laptop computers, message processors, hand-held devices, multi-processor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, mobile telephones, PDAs, tablets, pagers, routers, switches, and the like. The disclosure may also be practiced in distributed system environments where local and remote computer systems, which are linked (either by hardwired data links, wireless data links, or by a combination of hardwired and wireless data links) through a network, both perform tasks. In a distributed system environment, program modules may be located in both local and remote memory storage devices.
Embodiments of the present disclosure can also be implemented in cloud computing environments. In this description, “cloud computing” is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources. For example, cloud computing can be employed in the marketplace to offer ubiquitous and convenient on-demand access to the shared pool of configurable computing resources. The shared pool of configurable computing resources can be rapidly provisioned via virtualization and released with low management effort or service provider interaction, and then scaled accordingly.
A cloud-computing model can be composed of various characteristics such as, for example, on-demand self-service, broad network access, resource pooling, rapid elasticity, measured service, and so forth. A cloud-computing model can also expose various service models, such as, for example, Software as a Service (“SaaS”), Platform as a Service (“PaaS”), and Infrastructure as a Service (“IaaS”). A cloud-computing model can also be deployed using different deployment models such as private cloud, community cloud, public cloud, hybrid cloud, and so forth. In this description and in the claims, a “cloud-computing environment” is an environment in which cloud computing is employed.
In particular embodiments, processor(s) 802 includes hardware for executing instructions, such as those making up a computer program. As an example, and not by way of limitation, to execute instructions, processor(s) 802 may retrieve (or fetch) the instructions from an internal register, an internal cache, memory 804, or a storage device 806 and decode and execute them.
The computing device 800 includes memory 804, which is coupled to the processor(s) 802. The memory 804 may be used for storing data, metadata, and programs for execution by the processor(s). The memory 804 may include one or more of volatile and non-volatile memories, such as Random-Access Memory (“RAM”), Read Only Memory (“ROM”), a solid-state disk (“SSD”), Flash, Phase Change Memory (“PCM”), or other types of data storage. The memory 804 may be internal or distributed memory.
The computing device 800 includes a storage device 806 includes storage for storing data or instructions. As an example, and not by way of limitation, storage device 806 can comprise a non-transitory storage medium described above. The storage device 806 may include a hard disk drive (HDD), flash memory, a Universal Serial Bus (USB) drive or a combination of these or other storage devices.
The computing device 800 also includes one or more input or output (“I/O”) devices/interfaces 808, which are provided to allow a user to provide input to (such as user strokes), receive output from, and otherwise transfer data to and from the computing device 800. These I/O devices/interfaces 808 may include a mouse, keypad or a keyboard, a touch screen, camera, optical scanner, network interface, modem, other known I/O devices or a combination of such I/O devices/interfaces 808. The touch screen may be activated with a writing device or a finger.
The I/O devices/interfaces 808 may include one or more devices for presenting output to a user, including, but not limited to, a graphics engine, a display (e.g., a display screen), one or more output drivers (e.g., display drivers), one or more audio speakers, and one or more audio drivers. In certain embodiments, devices/interfaces 808 is configured to provide graphical data to a display for presentation to a user. The graphical data may be representative of one or more graphical user interfaces and/or any other graphical content as may serve a particular implementation.
The computing device 800 can further include a communication interface 810. The communication interface 810 can include hardware, software, or both. The communication interface 810 can provide one or more interfaces for communication (such as, for example, packet-based communication) between the computing device and one or more other computing devices 800 or one or more networks. As an example, and not by way of limitation, communication interface 810 may include a network interface controller (NIC) or network adapter for communicating with an Ethernet or other wire-based network or a wireless NIC (WNIC) or wireless adapter for communicating with a wireless network, such as a WI-FI. The computing device 800 can further include a bus 812. The bus 812 can comprise hardware, software, or both that couples components of computing device 800 to each other.
In the foregoing specification, the invention has been described with reference to specific example embodiments thereof. Various embodiments and aspects of the invention(s) are described with reference to details discussed herein, and the accompanying drawings illustrate the various embodiments. The description above and drawings are illustrative of the invention and are not to be construed as limiting the invention. Numerous specific details are described to provide a thorough understanding of various embodiments of the present invention.
The present invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The described embodiments are to be considered in all respects only as illustrative and not restrictive. For example, the methods described herein may be performed with less or more steps/acts or the steps/acts may be performed in differing orders. Additionally, the steps/acts described herein may be repeated or performed in parallel with one another or in parallel with different instances of the same or similar steps/acts. The scope of the invention is, therefore, indicated by the appended claims rather than by the foregoing description. All changes that come within the meaning and range of equivalency of the claims are to be embraced within their scope.
This disclosure contemplates any suitable network 904. As an example, and not by way of limitation, one or more portions of the network 904 may include an ad hoc network, an intranet, an extranet, a virtual private network (VPN), a local area network (LAN), a wireless LAN (WLAN), a wide area network (WAN), a wireless WAN (WWAN), a metropolitan area network (MAN), a portion of the Internet, a portion of the Public Switched Telephone Network (PSTN), a cellular telephone network, or a combination of two or more of these. The network 904 may include one or more networks 904.
Links may connect the contacting user device 902, the communication management system 104, the agent device 908, and the third-party system 906 to communication network 904 or to each other. This disclosure contemplates any suitable links. In particular embodiments, one or more links include one or more wireline (such as for example Digital Subscriber Line (DSL) or Data Over Cable Service Interface Specification (DOCSIS)), wireless (such as for example Wi-Fi or Worldwide Interoperability for Microwave Access (WiMAX)), or optical (such as for example Synchronous Optical Network (SONET) or Synchronous Digital Hierarchy (SDH)) links. In particular embodiments, one or more links each include an ad hoc network, an intranet, an extranet, a VPN, a LAN, a WLAN, a WAN, a WWAN, a MAN, a portion of the Internet, a portion of the PSTN, a cellular technology-based network, a satellite communications technology-based network, another link, or a combination of two or more such links. Links need not necessarily be the same throughout network environment 900. One or more first links may differ in one or more respects from one or more second links.
In particular embodiments, the contacting user device 902 and/or agent device 908 may be an electronic device including hardware, software, or embedded logic components or a combination of two or more such components capable of carrying out the appropriate functionalities implemented or supported by the contacting user device 902 and/or agent device 908. As an example, and not by way of limitation, a contacting user device 902 or agent device 908 may include any of the computing devices discussed above in relation to
In particular embodiments, the agent device 908 is part of (e.g., housed within, either entirely or in part) the third-party system 906. For example, the third-party system 906 may include a service system which maintains one or more agent users, each associated with a separate agent device 908. Accordingly, the communication management system 104 provides various functionality, user interfaces, electronic communications, and other information to the agent device 908 via the third-party system 906. In other embodiments, the agent device 908 is not associated with or part of a third-party system 906. Rather, the agent device 908 refers to a client device associated directly with the communication management system 104 or a different third-party and that communicates with the communication management system 104 without interaction via the third-party system 906.
In particular embodiments, the contacting user device 902 and/or the agent device 908 may include a web browser, such as MICROSOFT INTERNET EXPLORER, GOOGLE CHROME or MOZILLA FIREFOX, and may have one or more add-ons, plug-ins, or other extensions, such as TOOLBAR or YAHOO TOOLBAR. A user at contacting user device 902 and/or agent device 908 may enter a Uniform Resource Locator (URL) or other address directing the web browser to a particular server (such as server, or a server associated with a third-party system 906), and the web browser may generate a Hyper Text Transfer Protocol (HTTP) request and communicate the HTTP request to server. The server may accept the HTTP request and communicate to the contacting user device 902 (or the agent device 908) one or more Hyper Text Markup Language (HTML) files responsive to the HTTP request. The contacting user device 902 and/or agent device 908 may render a webpage based on the HTML files from the server for presentation to the user. This disclosure contemplates any suitable webpage files. As an example, and not by way of limitation, webpages may render from HTML files, Extensible Hyper Text Markup Language (XHTML) files, or Extensible Markup Language (XML) files, according to particular needs. Such pages may also execute scripts such as, for example and without limitation, those written in JAVASCRIPT, JAVA, MICROSOFT SILVERLIGHT, combinations of markup language and scripts such as AJAX (Asynchronous JAVASCRIPT and XML), and the like. Herein, reference to a webpage encompasses one or more corresponding webpage files (which a browser may use to render the webpage) and vice versa, where appropriate.
In particular embodiments, the communication management system 104 may be a network-addressable computing system that can host an online communication network. The communication management system 104 may generate, store, receive, and send messaging data, such as, for example, text data, audio data, and video data. The communication management system 104 can further generate, store, receive, and send other data such as user-profile data, location data, or other suitable data related to the online communication network. The communication management system 104 may be accessed by the other components of network environment 900 either directly or via network 904. In particular embodiments, the communication management system 104 may include one or more servers. Each server may be a unitary server or a distributed server spanning multiple computers or multiple datacenters. Servers may be of various types, such as, for example and without limitation, a web server, a communication server, a news server, a mail server, a message server, an advertising server, a file server, an application server, an exchange server, a database server, a proxy server, or another server suitable for performing functions or processes described herein, or any combination thereof. In particular embodiments, each server may include hardware, software, or embedded logic components or a combination of two or more such components for carrying out the appropriate functionalities implemented or supported by server.
In particular embodiments, the communication management system 104 may include one or more data stores. Data stores may be used to store various types of information. In particular embodiments, the information stored in data stores may be organized according to specific data structures. In particular embodiments, each data store may be a relational, columnar, correlation, or other suitable database. Although this disclosure describes or illustrates particular types of databases, this disclosure contemplates any suitable types of databases. Particular embodiments may provide interfaces that enable a contacting user device 902, a communication management system 104, an agent device 908, or a third-party system 906 to manage, retrieve, modify, add, or delete, the information stored in data store.
In particular embodiments, the communication management system 104 may store one or more correspondences in one or more data stores. In particular embodiments, a correspondence may include electronic messages—which may include digital audio, digital video, and/or digital text (e.g., as received in an electronic message or as transcribed from digital audio). The communication management system 104 may provide users (e.g., contacting users) of the online communication network the ability to communicate and interact with other users (e.g., agent users). In particular embodiments, agent users may join the online communication network via the communication management system 104, and the communication management system 104 may maintain connects for the agent users, including current and historical information relating to contacting users with whom the agent users interact.
In particular embodiments, the communication management system 104 may provide agent users with the ability to take actions on various types of items or objects, supported by communication management system 104. As an example, and not by way of limitation, the items and objects may include groups or individual contacting users communicating with the communication management system 104, computer-based applications that a user may use, transactions that allow agent users to buy, return, or sell items or services, indications that allow agent users to report events, or other interactions with various operations that an agent user may perform to assist a contacting user.
In particular embodiments, the communication management system 104 may be capable of linking a variety of entities. As an example, and not by way of limitation, the communication management system 104 may enable users such as contacting users and agent users to interact with each other as well as receive content from third-party systems 906 or other entities, or to allow users to interact with these entities through an application programming interfaces (API) or other communication channels.
In particular embodiments, a third-party system 906 may include one or more types of servers, one or more data stores, one or more interfaces, including but not limited to APIs, one or more web services, one or more content sources, one or more networks, or any other suitable components, e.g., that servers may communicate with. A third-party system 906 may be operated by a different entity from an entity operating the communication management system 104. In particular embodiments, however, the communication management system 104 and third-party systems 906 may operate in conjunction with each other to provide communication services to users of the communication management system 104 or third-party systems 906. In this sense, the communication management system 104 may provide a platform, or backbone, which other systems, such as third-party systems 906, may use to provide communication services and functionality to users across the Internet or other network.
In particular embodiments, a third-party system 906 may include a third-party service system. A third-party content service system may include one or more sources of products or services that are used by contacting user. As an example, and not by way of limitation, a service system can include a traffic monitoring system, a social networking system, an emergency reporting system, a home repair system, a weather monitoring system, an internet service provider system, an airline system, a transportation provider system, a news media system, a travel accommodation system, an electronic smart appliance distribution or monitoring system, or a power distribution or monitoring system.
In particular embodiments, the communication management system 104 may include a variety of servers, sub-systems, programs, modules, logs, and data stores. In particular embodiments, the communication management system 104 may include one or more of the following: a web server, action logger, API-request server, notification controller, action log, inference module, search module, advertisement-targeting module, user-interface module, user-profile store, third-party content store, or location store. The communication management system 104 may also include suitable components such as network interfaces, security mechanisms, load balancers, failover servers, management-and-network-operations consoles, other suitable components, or any suitable combination thereof. In particular embodiments, the communication management system 104 may include one or more user-profile stores for storing user profiles (e.g., for contacting users). A user profile may include, for example, contacting user information, biographic information, demographic information, behavioral information, social information, or other types of descriptive information, such as work experience, educational history, hobbies or preferences, interests, affinities, or location.
The web server may include a mail server or other messaging functionality for receiving and routing messages between the communication management system 104 and one or more contacting user devices 902 or agent devices 906. An API-request server may allow a third-party system 906 to access information from the communication management system 104 by calling one or more APIs. An action logger may be used to receive communications from a web server about a user's actions on or off the communication management system 104. Information may be pushed to a contacting user device 902 and/or agent device 908 as notifications, or information may be pulled from contacting user device 902 responsive to a request received from contacting user device 902 or agent device 908.
Claims
1. A method comprising:
- identifying match prediction training data, wherein the match prediction training data comprises a plurality of instances of raw communication match data and a plurality of service provider match markers, and further wherein each instance of raw communication match data of the plurality of instances of raw communication match data corresponds to a service provider match marker of the plurality of service provider match markers; and
- training a service provider neural network using the match prediction training data by, for each instance of raw communication data: passing the instance of raw communication data to the service provider match neural network; receiving a prediction from the service provider match neural network for the instance of raw communication data; comparing the prediction to the service provider match marker corresponding to the instance of raw communication match data; and
- modifying one or more parameters of one or more layers of the service provider match neural network based on the comparing.
2. The method of claim 1, wherein passing the instance of raw communication data to the service provider match neural network comprises:
- generating an input vector from the instance of raw communication data; and
- passing the generated input vector to the service provider match neural network.
3. The method of claim 1, wherein each instance of raw communication data comprises recorded audio, and each corresponding service provider match marker comprises one or more of a service provider identifier or a plurality of attributes of a predicted service provider.
4. The method of claim 1, wherein the prediction for the instance of raw communication data comprises a first plurality of attributes and the service provider match marker corresponding to the instance of raw communication data comprises a second plurality of attributes, and comparing the prediction to the service provider match marker corresponding to the instance of raw communication match data comprises:
- determining that more than a threshold number of attributes from the first plurality of attributes match attributes from the second plurality of attributes; and
- in response to the determination, determining that the prediction for the instance of raw communication data matches the service provider match marker corresponding to the instance of raw communication data.
5. The method of claim 1, further comprising:
- receiving a communication from a client device associated with a contacting user;
- extracting raw communication data associated with the contacting user from the received communication;
- providing the extracted raw communication data to the service provider match neural network;
- receiving a first service provider match prediction from the service provider match neural network based on the provided extracted raw communication data; and
- providing the first service provider match prediction to a customer request system.
6. The method of claim 5, further comprising:
- receiving, from the customer request system, an indication that the first service provider match prediction cannot be accommodated;
- in response to receiving the indication, modifying one or more parameters of one or more layers of the service provider match neural network;
- providing the extracted raw communication data to the service provider match neural network; and
- receiving a second service provider match prediction from the service provider match neural network based on the extracted raw communication data.
7. The method of claim 5, wherein providing the extracted raw communication data to the service provider match neural network comprises:
- generating an input vector based on the raw communication data; and
- providing the generated input vector to the service provider match neural network.
8. A system comprising:
- at least one processor; and
- at least one non-transitory computer-readable storage medium storing instructions thereon that, when executed by the at least one processor, cause the system to: identify match prediction training data, wherein the match prediction training data comprises a plurality of instances of raw communication match data and a plurality of service provider match markers, and further wherein each instance of raw communication match data of the plurality of instances of raw communication match data corresponds to a service provider match marker of the plurality of service provider match markers; and train a service provider neural network using the match prediction training data by, for each instance of raw communication data: passing the instance of raw communication data to the service provider match neural network; receiving a prediction from the service provider match neural network for the instance of raw communication data; comparing the prediction to the service provider match marker corresponding to the instance of raw communication match data; and modifying one or more parameters of one or more layers of the service provider match neural network based on the comparing.
9. The system of claim 8, wherein passing the instance of raw communication data to the service provider match neural network comprises:
- generating an input vector from the instance of raw communication data; and
- passing the generated input vector to the service provider match neural network.
10. The system of claim 8, wherein each instance of raw communication data comprises recorded audio, and each corresponding service provider match marker comprises one or more of a service provider identifier or a plurality of attributes of a predicted service provider.
11. The system of claim 8, wherein the prediction for the instance of raw communication data comprises a first plurality of attributes and the service provider match marker corresponding to the instance of raw communication data comprises a second plurality of attributes, and comparing the prediction to the service provider match marker corresponding to the instance of raw communication match data comprises:
- determining that more than a threshold number of attributes from the first plurality of attributes match attributes from the second plurality of attributes; and
- in response to the determination, determining that the prediction for the instance of raw communication data matches the service provider match marker corresponding to the instance of raw communication data.
12. The system of claim 8, further storing instructions thereon that, when executed by the at least one processor, cause the system to:
- receive a communication from a client device associated with a contacting user;
- extract raw communication data associated with the contacting user from the received communication;
- provide the extracted raw communication data to the service provider match neural network;
- receive a first service provider match prediction from the service provider match neural network based on the provided extracted raw communication data; and
- provide the first service provider match prediction to a customer request system.
13. The system of claim 12, further storing instructions thereon that, when executed by the at least one processor, cause the system to:
- receive, from the customer request system, an indication that the first service provider match prediction cannot be accommodated;
- in response to receiving the indication, modify one or more parameters of one or more layers of the service provider match neural network;
- provide the extracted raw communication data to the service provider match neural network; and
- receive a second service provider match prediction from the service provider match neural network based on the extracted raw communication data.
14. The system of claim 13, wherein providing the extracted raw communication data to the service provider match neural network comprises:
- generating an input vector based on the raw communication data; and
- providing the generated input vector to the service provider match neural network.
15. A non-transitory computer-readable medium storing instructions thereon that, when executed by at least one processor, cause a computer system to:
- identify match prediction training data, wherein the match prediction training data comprises a plurality of instances of raw communication match data and a plurality of service provider match markers, and further wherein each instance of raw communication match data of the plurality of instances of raw communication match data corresponds to a service provider match marker of the plurality of service provider match markers; and
- train a service provider neural network using the match prediction training data by, for each instance of raw communication data: passing the instance of raw communication data to the service provider match neural network; receiving a prediction from the service provider match neural network for the instance of raw communication data; comparing the prediction to the service provider match marker corresponding to the instance of raw communication match data; and modifying one or more parameters of one or more layers of the service provider match neural network based on the comparing.
16. The computer-readable medium of claim 15, wherein passing the instance of raw communication data to the service provider match neural network comprises:
- generating an input vector from the instance of raw communication data; and
- passing the generated input vector to the service provider match neural network.
17. The computer-readable medium of claim 15, wherein each instance of raw communication data comprises recorded audio, and each corresponding service provider match marker comprises one or more of a service provider identifier or a plurality of attributes of a predicted service provider.
18. The computer-readable medium of claim 15, wherein the prediction for the instance of raw communication data comprises a first plurality of attributes and the service provider match marker corresponding to the instance of raw communication data comprises a second plurality of attributes, and comparing the prediction to the service provider match marker corresponding to the instance of raw communication match data comprises:
- determining that more than a threshold number of attributes from the first plurality of attributes match attributes from the second plurality of attributes; and
- in response to the determination, determining that the prediction for the instance of raw communication data matches the service provider match marker corresponding to the instance of raw communication data.
19. The computer-readable medium of claim 15, further storing instructions thereon that, when executed by the at least one processor, cause the computer system to:
- receive a communication from a client device associated with a contacting user;
- extract raw communication data associated with the contacting user from the received communication;
- provide the extracted raw communication data to the service provider match neural network;
- receive a first service provider match prediction from the service provider match neural network based on the provided extracted raw communication data; and
- provide the first service provider match prediction to a customer request system.
20. The computer-readable medium of claim 19, further storing instructions thereon that, when executed by the at least one processor, cause the computer system to:
- receive, from the customer request system, an indication that the first service provider match prediction cannot be accommodated;
- in response to receiving the indication, modify one or more parameters of one or more layers of the service provider match neural network;
- provide the extracted raw communication data to the service provider match neural network; and
- receive a second service provider match prediction from the service provider match neural network based on the extracted raw communication data.
Type: Application
Filed: Aug 28, 2019
Publication Date: Feb 25, 2021
Inventor: Jafar Adibi (Los Angeles, CA)
Application Number: 16/553,462
