[0001] Electronic devices, such as imaging devices, computing devices, network peripherals, audio and video devices, and other electronic or mechanical goods, are peripherals commonly used in home and office environments for carrying out a wide range of processes. Each electronic device may have certain functionalities to carry out the wide range of processes. To assist a user in resolving issues related to the functioning of an electronic device or their parts, support systems, such as customer support centers may be established and managed by manufacturers. Such customer support centers may employ remote agents for responding to troubleshoot queries from users of the electronic devices for resolving the issues with the electronic devices.

BRIEF DESCRIPTION OF DRAWINGS

[0002] The detailed description is described with reference to the accompanying figures. It should be noted that the description and figures are merely examples of the present subject matter and are not meant to represent the subject matter itself.

[0003] Figure 1 illustrates a system for providing remote agent support, according to an example implementation of the present subject matter.

[0004] Figure 2 illustrates a computing environment with the system for providing remote agent support, according to an example implementation of the present subject matter.

[0005] Figure 3 illustrates a method for providing remote agent support, according to an example implementation of the present subject matter.

[0006] Figure 4 illustrates a method for training a machine learning model for providing remote agent support, according to another example implementation of the present subject matter. [0007] Figure 5 illustrates a method for providing remote agent support, according to an example implementation of the present subject matter.

[0008] Figure 6 illustrates a network environment having a non-transitory computer readable medium for providing remote agent support, according to an example implementation of the present subject matter.

[0009] Throughout the drawings, identical reference numbers designate similar, but not necessarily identical, elements. The figures are not necessarily to scale and the size of some parts may be exaggerated to more clearly illustrate the example shown. Moreover, the drawings provide examples and/or implementations consistent with the description; however, the description is not limited to the examples and/or implementations provided in the drawings.

DETAILED DESCRIPTION

[0010] Companies utilize customer support centers to provide assistance to users for resolving an issue related to user devices sold or maintained by the company. Examples of such user devices include, but are not limited to, imaging devices, computing devices, network peripherals, audio and video devices, and other electronic or mechanical goods. At these customer support centers, multiple remote agents, such as human or virtual support agents, answer telephone calls or respond to text messages from users looking for support. During such a conversation, the user may explain the issue with the user device to the remote agent to help the remote agent diagnose and resolve the issues with the user devices. In some cases, the remote agent may ask the user to execute a series of instructions to aid in the diagnosis and resolution.

[0011] Typically, when a user contacts a remote agent with an issue, the remote agent has no prior knowledge of the status or health of the user device. In many cases, the user also may not be able to correctly communicate the issue with the device to the remote agent. For instance, the user may explain the issue vaguely and in broad terms that may make it difficult for the remote agent to understand the issue. The remote agent may either take a lot of time to resolve the issue or may not be able to correctly identify the issue and provide correct instructions to the user for resolving the issue. If the issue is not resolved by the remote agent, field agents may have to be sent for resolving the issues. Further, during such field visits, the field agents may sometimes ascertain that the actual issue diagnosed for the user device is different from the issue described by the user. In such cases it may be possible that the field agent may not have carried the correct parts for replacement or repair and a second field visit may have to be undertaken to service the user device. In some cases, it may also be possible that the actual issue could have been remotely resolved by the remote agent and the field visit could have been avoided if the issue had been correctly identified. Thus, inaccurate diagnosis of the issue by the remote agent, based on incorrect or incomplete description of the issue by the user, may lead to extra and unnecessary field visits by the field agents and wastage of efforts, resources, and time of the remote agent, the field agent, and the user.

[0012] The present invention relates to remote agent support for accurately and efficiently resolving issues with user devices. In one example implementation, a machine learning model may be used to recommend probable resolution steps to the remote agent for resolving the issues. The machine learning model may use device data corresponding to the user device to determine a device health index (DHI) and a part health index (PHI) for a plurality of parts to identify a cause of the issue with the user device. In one example, the device data may include one of telemetry data, audio data, image data, text data from call log, device status, and recent service and maintenance logs corresponding to the user device. The machine learning model may further determine the probability of resolving the issue by the remote agent to ascertain if the issue can be remotely resolved. If it is ascertained that a field visit by a field agent may be useful, the machine learning model may also provide recommendations for parts that may be carried by the field agent for replacement or repair. Thus, unnecessary field visits or long calls with the remote agent may be avoided.

[0013] In one example, the machine learning model may originally be trained using historical device data and past solutions provided for solving issues faced by the user devices. The machine learning model may further generate mapping tables for being used in a prediction phase for recommending solutions to the issues faced by the users in real time. The mapping tables may indicate, for each part of the device, a mapping between an issue, historical device data, and past solutions provided for solving the issue. In one example, the mapping tables may include a PHI mapping table, a resolution mapping table, and a part replacement mapping table.

[0014] In one example implementation of the present subject matter, when the user of the user device communicates with the remote agent for obtaining a solution for an issue with the user device, a call log may be created to capture the communication between the remote agent and the user. In one example, the call log may include the issue indicated by the user and notes that capture conversation between the remote agent and the user.

[0015] The machine learning model may then use the device data and the PHI mapping table to determine the PHI for a plurality of parts of the user device and the DHI for the user device. In one example implementation, the machine learning model may use entire or a combination of device data available for the user device. The machine learning model may then recommend probable resolution steps to the remote agent for resolving the issue based on the DHI, the PHI for a plurality of parts of the user device, and the resolution mapping table.

[0016] In one example, to provide the recommendations for probable resolution steps, a probability of remotely resolving the issue may be initially determined by the machine learning engine based on the resolution mapping table, the DHI, and the PHI. If it is determined that the probability is less than a threshold, a field visit by a field agent may be suggested for resolving the issue. If it is determined that the probability is greater than the threshold, the remote agent may be suggested to continue interacting with the user for remotely resolving the issue. In case a field visit is suggested, a recommendation may be provided for possible resolution and parts to be serviced and carried by the field agent to repair the user device. The parts to be serviced may be determined based on the PHI for the plurality of parts of the user device, the device data, and a part replacement mapping table. [0017] Using the device data and the PHI mapping table to determine the

DHI and the PHI for a plurality of parts of the user device, the health of the user device and parts of the user device may be monitored to correctly identify the issue faced by the user. The DHI, the PHI, the device data, and the resolution mapping table may further be used to recommend probable resolution steps to the remote agent for resolving the issue. The probable resolution steps may suggest whether the issue could be remotely resolved by the remote agent or a field visit by the field agent would be suggested. In one example, the recommended probable resolution steps may further suggest probable actions that the remote agent may suggest to the user for resolving the issue. The remote agent may also be provided with a rationale for the recommendation to provide reasonings and basis for determining the probable cause and recommending the probable resolution steps. The rationale for the recommendation may further help the remote agent in providing solution and suggestions to the user for resolving the issue.

[0018] Thus, using the device data and the mapping tables, the health or abnormal behavior of the overall device or device parts of the device may be monitored and analyzed. The remote agent may now be able to correctly identify the issue with the user device and recommend probable resolution steps to resolve the identified problem. It may further be determined whether the issue could be solved by the remote agent remotely or a field visit by the field agent would be necessary, thereby, preventing unnecessary field visits for the field agents. By recommending possible resolutions and parts to be serviced or carried by the field agent for repair of the user device, it may further be ensured that the field agent visits the user’s location with correct parts, thereby, saving repeated or extra visits for the field agent.

[0019] Further, training the machine learning model on multiple data types, such as telemetry data, audio data, call logs, and image data facilitates in ensuring that the probable solution steps are recommended even if entire device data is not available. For instance, if any type of data, such as the audio data or the image data is not available, the machine learning model may use the available device data, such as call log or telemetry data to determine the DHI, the PHI, and the probable solution steps.

[0020] The present subject matter is further described with reference to

Figures 1 to 6. It should be noted that the description and figures merely illustrate principles of the present subject matter. Various arrangements may be devised that, although not explicitly described or shown herein, encompass the principles of the present subject matter. Moreover, all statements herein reciting principles, aspects, and examples of the present subject matter, as well as specific examples thereof, are intended to encompass equivalents thereof.

[0021] Figure 1 illustrates a system 102, according to an example implementation of the present subject matter. In one example of the present subject matter, the system 102 may provide support to a remote agent for resolving issues with user devices. The issue may indicate a problem that the user may encounter while operating the electronic device. In one example implementation, the system 102 may be used to recommend probable resolution steps to the remote agent for resolving the issues. The system 102 may further determine the probability of remotely resolving the issue by the remote agent. Examples of the system 102 may include, but are not limited to, a stand-alone system or a distributed computing system having one or more physical computing systems geographically distributed at same or different locations.

[0022] In one example implementation, the system 102 may include a query engine 104 to obtain device data corresponding to a user device of a user communicating with a remote agent for obtaining solution for an issue with the user device. In one example, the device data may include at least one of telemetry data, audio data, image data, call log, device status, and recent service and maintenance logs corresponding to the user device. In one example, the telemetry data may include various error codes and part related information generated by the user device. For example, the telemetry data may indicate error codes that may occur when a part of the user device may stop functioning according to its defined functioning, fail, or encounter any error. In one example, the telemetry data may indicate an error that may occur when a display unit of the user device starts to flicker. In this case, the error code may indicate an issue with the display unit of the user device. The audio data may include an audio file received from the user device. The audio file may provide an audio sample representing noise generated by different parts of the user device 202 that may indicate whether a part is performing normally. The image data may include images received from the user by the remote agent. In one example, the images may include a scanned image of an output of the user device, such as a scanned image of a printed page from a print device. The image data may also include an image of a part of the user device or the user device itself.

[0023] The call log may include notes, created by the remote agent, describing issue faced by the user of the user device, preliminary actions suggested by the remote agent to the user to overcome the issue, and response from the user for the preliminary actions provided by the remote agent. The device status may represent information about the status of the user device. Examples of the device status include, but are not limited to, power ON state, power OFF state, ERROR state, network connection state. The recent service and maintenance logs may provide information about services done in past and part replaced and repaired for the user device. In one example, the query engine 104 may obtain the device data using a device ID of the user device. In another example, the query engine 104 may obtain some parts of the device date using the device ID and may obtain the other parts of the device data from the remote agent.

[0024] The system 102 may further include a machine learning engine 106 to recommend probable resolution steps for resolving the issue based on the determining. The system 102 may initially compute a part health index (PHI) for a plurality of parts of the user device and a device health index (DHI) for the user device based on the device data and a PHI mapping table. In one example, the PHI mapping table may indicate, for each part of the user device, a mapping between the issue, historical device data, past solutions provided for solving the issue, and health index values. [0025] Further, the machine learning engine 106 may determine a probability of remotely resolving the issue by the remote agent based on the device data, the DHI, the PHI for the plurality of parts of the user device, and a resolution mapping table. In one example, the resolution mapping table may indicate a mapping between the issue, the historical device data, the past solutions provided for solving the issue, and remote resolution probability score. The machine learning engine 106 may subsequently recommend probable resolution steps for resolving the issue based on the probability of remotely resolving the issue by the remote agent. In one example, the probable solution steps may include suggestions for the remote agent to continue interaction with the user for remotely resolving the issue, actions to be suggested to the user by the remote agent to resolve the issue, a field visit by the field agent, and parts to be replaced or repaired by the field agent.

[0026] In one example implementation, the machine learning engine 106 may compare the probability with a threshold to determine if the issue may be remotely resolved by the remote agent. If it is determined that the probability is less than the threshold, the machine learning engine 106 may recommend a field visit by a field agent for resolving the issue. If it is determined that the probability is greater than the threshold, the machine learning engine 106 may recommend the remote agent to continue interacting with the user for remotely resolving the issue.

[0027] Figure 2 illustrates a computing environment 200 implementing the system 102 for providing remote agent support for resolving issues with user devices, according to an example implementation of the present subject matter. The computing environment 200 may include the system 102, a plurality of user devices 202-1 and 202-2, and remote agent systems 204 connected to each other over a communication network 206. The user devices 202-1 and 202-2 may be individually referred to as user device 202 and collectively referred to as user devices 202.

[0028] Examples of the user device 202 may include, but are not limited to, imaging devices, such as printers and scanner; computing devices; network peripherals; personal digital assistant (PDA); mobile phones; and other electronic or mechanical goods. The communication network 206 may be a wireless network, a wired network, or a combination thereof. The communication network 206 may also be an individual network or a collection of many such individual networks, interconnected with each other and functioning as a single large network, e.g., the Internet or an intranet. Examples of such individual networks include local area network (LAN), wide area network (WAN), the internet, Global System for Mobile Communication (GSM) network, Universal Mobile Telecommunications System (UMTS) network, Personal Communications Service (PCS) network, Time Division Multiple Access (TDMA) network, Code Division Multiple Access (CDMA) network, Next Generation Network (NGN), Public Switched Telephone Network (PSTN), and Integrated Services Digital Network (ISDN). Depending on the technology, the communication network includes various network entities, such as transceivers, gateways, and routers. In an example, the communication network 206 may include any communication network that uses any of the commonly used protocols, for example, Hypertext Transfer Protocol (HTTP), and Transmission Control Protocol/Intemet Protocol (TCP/IP).

[0029] The user may use the user device, such as the user device 202-1 to communicate with a customer support center using the communication network 206. In one example, the customer support center may have multiple remote agents to answer telephone calls or respond to text messages with troubleshoot queries from the users looking for support with the user device 202. In one example, the remote agents may include human support agents, chatbots, voice assistants, and virtual agents. In one example, a remote agent may communicate with the user of the user device 202 using the remote agent systems 204. The remote agent may further communicate with the system 102 for receiving support and recommendations for resolving the issue.

[0030] The system 102 may include interface(s) 208 and memory 210. The interface(s) 208 may include a variety of machine readable instructions-based interfaces and hardware interfaces that allow interaction with a user and with other communication and computing devices, such as network entities, web servers, networked computing devices, external repositories, and peripheral devices. The interface(s) 208 may also include a variety of interfaces, for example, interfaces for data input and output devices, referred to as I/O (input/output) devices, storage devices, network devices, and the like. The interface(s) 208 may facilitate communication between the system 102, the remote agent systems 204, and various other computing devices operating in the computing environment 200. The interface(s) 208 may also provide a communication pathway for one or more components of the system 102. Examples of such components may include, but are not limited to, input devices, such as keyboards and a touch enabled graphical user interface.

[0031] The memory 210 may store one or more computer-readable instructions, which may be fetched and executed to provide print interfaces to users for providing print instructions. The memory 210 may include any non-transitory computer-readable medium including, for example, volatile memory such as Random Access Memory (RAM), or non-volatile memory such as Erasable Programmable Read-only Memory (EPROM), flash memory, and the like. The system 102 further includes engine(s) 212 and data 214.

[0032] The engine(s) 212 may be implemented as a combination of hardware and programming (for example, programmable instructions) to implement one or more functionalities of the system 102. In examples described herein, such combinations of hardware and programming may be implemented in several different ways. For example, the programming for the engine(s) 212 may include processor executable instructions stored on a non-transitory machine- readable storage medium and the hardware for the engine(s) 212 may include a processing resource to execute such instructions. In one example, the engine(s) 212 may further be coupled to processors of the system 102 to execute the functionalities of the engine(s) 212. The processor(s) may include microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, and/or any other devices that manipulate signals and data based on computer-readable instructions. [0033] In the present examples, the machine-readable storage medium may store instructions that, when executed by the processing resource, may implement the engine(s) 212. In such examples, the system 102 may include the machine- readable storage medium storing the instructions and the processing resource to execute the instructions, or the machine-readable storage medium may be separate but accessible to the system 102 and the processing resource. In other examples, engine(s) 212 may be implemented by electronic circuitry. The engine(s) 212 may further include circuitry and hardware for monitoring operations.

[0034] The data 214 may include data that is either stored or generated as a result of functionalities implemented by any of the engine(s) 212. Further, the engine(s) 212 of the system 102 may include a query engine 104, a machine learning engine 106, and other engine(s) 216. The other engine(s) 216 may implement functionalities that supplement applications or functions performed by the engine(s) 212. Further, the data 214 may include user device data 218, mapping tables 220, and other data 222.

[0035] As previously discussed, the system 102 may use a machine learning engine 106 to implement a machine learning model for recommending probable resolution steps to the remote agent for resolving the issues. In one example, the machine learning model may be a Deep Learning based model. As previously described, the machine learning model may use the device data as input and provide the recommendations, the probability of remotely resolving the issue by the remote agent, the DHI, the PHI for the plurality of parts of the user device, actions to be suggested to the user by the remote agent to resolve the issue, and parts to be replaced or repaired by the field agent. Further, as will be explain in detail, the machine learning model may pre-process the device data convert the device data to vectors and matrix for analysis along with the mapping tables.

[0036] As previously described, the issue may indicate a problem that the user may encounter while operating the electronic device. For example, a user may experience problems, such as boot failure, user error, hardware failure, software confliction, and programming error, while operating a computing device. Similarly, the user may experience problems, such as flickering, loose wire connection, no sound output, or screen blackout, while operating audio/video devices. The machine learning engine 106 may use device data corresponding to the user device and mapping tables to identify a cause of the issue with the user device and provide recommendations for resolving the issue. The machine learning engine 106 may further determine the probability of resolving the issue by the remote agent to ascertain if the issue can be remotely resolved. If it is ascertained that a field visit by a field agent may be useful, the machine learning engine 106 may also provide recommendations for parts that may be carried by the field agent for replacement or repair.

[0037] In one example, the machine learning engine 106 may be trained using historical device data and past solutions provided for solving issues faced by the same or similar devices as the user device 202, in the past. The historical device data may include at least one of historic telemetry data, historic audio data, historic image data, and historic call logs. In one example, the historic telemetry data may include a telemetry vector representation for various error codes and part related information generated by the user devices 202 in past. The historic audio data may include audio clips and related data obtained while resolving the issue with the user devices 202 in past. The historic image data may include image data obtained in past while resolving the issue with the user devices 202.

[0038] The historic call logs may include details of past calls received by remote agents for solving issues with the user devices 202. These details may include at least the issues and the past solutions provided for solving the issues. In one example, the remote agent may have created a call log when communicating with the user of the user device 202 in past, describing issues faced by the user, preliminary actions suggested by the remote agent to the user to overcome the issue, and response from the user for the preliminary actions provided by the remote agent. The machine learning engine 106 may obtain the historical device data from a central server, such as a central server 226. In one example, the central server 226 may be a remote server, a cluster of servers, or multiple servers that may be located at different geographic locations. The central server 226 may communicate with the user devices 202 and the system 102 using the communication network 206. In one example, the machine learning engine 106 may obtain historical device data from the user device data 218 and the other data 222.

[0039] The machine learning engine 106 may then be trained using the historical device data to determine cause for different issue(s), probability of remotely resolving the issue(s) by a remote agent and provide recommendations for solving the issue(s) that may be associated with different functional features of various user devices 202. In one example, the machine learning engine 106 may use deep learning to analyze the historical device data and create mappings between the past issues associated with multiple user device models, past solutions provided for solving the issues, remote solution probability generated in past for the issues, and list of parts replaced in past for solving the issue. These mappings may then be used to create mapping tables. In one example, the mapping tables may include a resolution mapping table, a PHI mapping table, and a part replacement mapping table. The PHI mapping table may indicate, for each part of the device, a mapping between the issue, historical device data, past solutions provided for solving the issue, and health index values. The machine learning engine 106 may analyze the issues and the past solutions, including whether parts were replaced or not to generate the mapping between the issue and the health index of parts.

[0040] The resolution mapping table may indicate a mapping between the issue, the historical device data, the past solutions provided for solving the issue, and remote resolution probability score. The part replacement mapping table may indicate a mapping between the issue, historical device data, and list of parts replaced in past for solving the issue. The machine learning engine 106 may analyze the issues and the past solutions, including whether a part was replaced, which parts were replaced or repaired to generate the mapping between the issue and the remote resolution probability score, and the mapping between the issue and the list of parts replaced in past.

[0041] Once the machine learning engine 106 is trained, the system 102 or a similar system may be implemented for providing remote agent support while resolving issues raised by users. In one example, when a user of the user device 202 encounters an issue with the user device 202, the user may initiate a communication with a remote agent of a customer support center to share a troubleshoot query indicating the device ID and the issue with the user device 202. In one example, the user may initiate a call from a communication device, such as the user device 202- 2 to interact with the remote agent. In another example, the user may initiate a communication with the remote agent by using text messaging services, chatbots, email using other user devices 202. In one example, the remote agent may use the remote agent systems 204 to communicate with the user and with the system 102 to resolve the issue with the user device 202.

[0042] In one example, the remote agent systems 204 may include agent devices 224-1 and 224-2, hereinafter collectively referred to as agent devices 224 and individually as agent device 224. Examples of the agent device 224 may include, but are not limited to, a telephone, a mobile phone, and a computing device. The remote agent may use the agent device 224 to communicate with the user device 202 over the communication network 206 using analog or digital telephony systems, such as Voice over Internet Protocol (VoIP) telephone calls, PSTN, GSM, and CDMA. The remote agent may further use the agent device 224 for making notes of the remote agent’s communication with the user and for communicating with the system 102.

[0043] In one example, upon receiving the troubleshoot query from the user of the user device 202 in the communication, the remote agent may initiate the communication with the user and ask for a device ID of the user device 202 and further details of the issue. The remote agent may further suggest some preliminary actions to the user to overcome the issue. Further, the remote agent may interact with the query engine 104 using the agent device 224 to create a call log corresponding to the user device 202 for the troubleshoot query. The query engine 104, on receiving a new troubleshoot query notification from the agent device 224, may generate the call log from the communication between the user of the user device 202 and the remote agent. The call log may capture the issue, the preliminary actions suggested by the remote agent to the user, and response from the user for the preliminary actions provided by the remote agent. In one example, the query engine 104 may associate the call log the device ID of the user device 202. Further, the query engine 104 may update the call log in real time based on the communication between the remote agent and the user to obtain an updated call log. In one example implementation, the remote agent may create the call log on the agent device 224. The remote agent may capture the device ID in the call log and share the device ID with the query engine 104.

[0044] The query engine 104 may then obtain the device data corresponding to the user device 202 using the device ID. In one example implementation, the query engine 104 may obtain the device data from at least one of the central server 226, a remote server, a group of servers or any cloud based platform. In one example, the device data may include at least one of telemetry data, audio data, image data, text data from call log, device status, and recent service and maintenance logs corresponding to the user device 202. In one example, the query engine 104 may obtain device data for a predetermined past number of days for the user device 202.

[0045] As previously described, the telemetry data may indicate error codes that may be generated when a part of the user device 202 fails or encounters any error in functioning. For example, when a display unit of the user device 202 starts to flicker, an error code indicting flickering of the display unit may be generated. In one example, the user devices 202 may regularly or periodically share the error codes with the central server 226. The central server 226 may save the error codes of different user devices 202 and tag the error codes of each user device 202 with the device ID corresponding to the user device 202.

[0046] The audio data, as previously described, may include audio samples or audio files received from the user device 202. The sample audio may provide decibel (dB) information that may represent information about disturbance in the audio due to noise created by different parts of the user device 202. The audio file may indicate whether a part is functioning normally and may, therefore, indicate health of the plurality of parts of the user device 202, such as a speaker, microphone, print nozzle, and roller, of the user device 202.

[0047] The call log may include notes that may describe issues faced by the user of the user device 202, preliminary actions suggested by the remote agent to the user to overcome the issue, and response from the user for the preliminary actions provided by the remote agent. In one example, the call log may be a text document. In another example, the call log may be audio recordings of the communication between the remote agent and the user. In another example, the call log may include text notes and the audio recordings of the communication between the remote agent and the user. The call log may be created and updated by the remote agent in real time while communicating with the user. The call log may be stored on the agent device 224 or the device data 218 of the system 102. In one example, the remote agent may also save the call log at the central server 226.

[0048] Further, the image data may include images received from the user by the remote agent. Example of the image include, but are not limited to, a scanned image of a printed page from a print device, an image of a part of the user device 202 or the user device 202, an image of an error code displayed on a screen of the user device 202, a copy of a page scanned or photocopied by a scanner device. In one example, pixel data of a scanned image of a printed document may represent the health of the scanning unit of an imaging device, such as the imaging device 202-1. Further, the scanned image of the printed document may also indicate that there may be an issue with the printing unit or the cartridge of the user device 202, such as a print device.

[0049] The device status may represent information about the status of the user device 202. For example, the device status may indicate a disconnected to network state for the user device 202. This may indicate that there may be an issue with the communication unit of the user device 202. Further, the recent service and maintenance logs may provide information about services done in past, and part replaced and repaired for the user device 202. For example, the parts of the user device 202 may wear and tear with regular use and may require to be serviced or repaired after a certain period of time. The maintenance log may contain information about past services and list of parts of the user device 202 replaced in past. In case, the user device 202 is not been serviced according to the prescribed time period, it may affect the health of the parts, and may thus be used to compute health of a part of the user device 202.

[0050] The query engine 104 may save the device data in the user device data 218 for further processing and analysis by the machine learning engine 106. Upon receiving the device data, the machine learning engine 106 may preprocess the device data. In one example, the machine learning engine 106 may perform at least one of stemming, lemmatization, speech tagging, parsing, named-entity recognition, coreference resolution, and removal of word clutters from the call log to preprocess the call log for analysis. The machine learning engine 106 may also perform band-pass filtering, locating region of interests, and Fourier transformation to process the call log if the call log includes the audio recordings of the communication between the remote agent and the user. On receiving the telemetry data, the machine learning engine 106 may generate a vector representation of the error codes. In one example implementation, the machine learning engine 106 may generate an audio vector representation from the audio data. In one example, the machine learning engine 106 may perform band-pass filtering, locating region of interests, and Fourier transformation to process the audio data. Upon receiving the image data, the machine learning engine 106 may generate an image matrix for each image of the image data. In one example, the image matrix for an image may be generated based on pixel data of the image using techniques of image processing.

[0051] The machine learning engine 106 may subsequently compute a PHI for a plurality of parts of the user device 202 and a DHI for the user device 202 based on the device data and the PHI mapping table. In one example, the PHI for a part of the user device 202 may indicate a health index of the part. The DHI for the user device 202 may indicate health index of the user device 202. The health of the user device 202, as provided by the DHI, and the health of the parts of the user device 202, as provided by the PHI, may indicate whether the user device 202 and the parts of the user device 202, respectively, are functioning normally. In one example, the PHI of each part of the user device 202 may be computed based on the device data and the PHI mapping table. For example, if the telemetry data indicates that a particular error code was generated by the user device 202 in last five to ten days and the mapping table indicates, based on the historical device data, that in past a particular part was replaced on generation of the particular error code, the PHI for that particular part may be low. In the present example, the mapping table may indicate that in past, replacing the particular part was provided as a solution for the issue as indicated by the particular error code. The mapping table may also indicate a health index value of below 30% to indicate that the particular part may have to be replaced.

[0052] In another example, a scanner unit of an imaging device may be capable of performing 10000 scans in its complete lifecycle and the device status and recent service and maintenance logs indicate that the imaging device has performed 2000 scans till now. In such a case, the PHI of the scanner unit maybe good or high.

[0053] Further, the DHI may be computed as an aggregation or summarization of the PHI values of the plurality pf parts of the user device 202. In one example, the DHI may be computed as a mean, median or average of the PHI of the plurality of parts of the user device 202. In one example the DHI may indicate the overall health of the user device 202 as a mean of the PHI of the plurality of parts. Further, the PHI and the DHI maybe indicated in terms of percentage values. For, instance, in the above example, the PHI may be calculated as a value equal to ((10000-2000)/10000) * 100. The PHI of for the scanning unit may thus be 80%.

[0054] Further, the machine learning engine 106 may analyze the DHI and the PHI for the plurality of parts of the user device 202 to identify a probable cause for the issue. For example, by analyzing the DHI and the PHI for the plurality of parts of the user device, the machine learning engine 106 may determine a part of the user device 202 with low PHI value and may determine that the part may probably be damaged and may be a probable cause for the issue. For example, the DHI greater than 90%, may indicate that user device 202 is functioning fine and there may not be a failure and the issue may be easily resolved. The DHI of less than 40%, may indicate that multiple parts of the user device 202 may have to be further analyzed and probably repaired or replaced. The DHI in the range of 40%- 90%, may indicate that one or two parts may be further analyzed. Similarly, for a part, the PHI of less than 30%, may indicate that that part may be replaced. The PHI of higher than 30%, may indicate that further data, such as the previous maintenance history, part usage frequency, part age, average life of the part, may have to be analyzed to decide if that has to be replaced. The machine learning engine 106 may further identify a probable cause, such as a part failure, a particular error in a part, a device malfunctioning, and a part replacement due to full utilization of the part, like empty ink cartridge, for the issue based on the DHI and the PHI.

[0055] The machine learning engine 106 may further determine a probability of remotely resolving the issue by the remote agent based on the device data, the DHI, the PHI for the plurality of parts of the user device 202, and the resolution mapping table. The machine learning engine 106 may compare the probability with a threshold to determine if the issue may be remotely resolved by the remote agent. In one example, the threshold may be defined by a manufacturer of the user device 202, a service provider for the user device 202, or by the machine learning engine 106 during a training phase based on the historical device data and service and maintenance logs.

[0056] If it is determined that the probability is less than the threshold, the machine learning engine 106 may ascertain that a field visit by a field agent may be recommended for resolving the issue. Further, the machine learning engine 106 may determine parts to be serviced based at least on the PHI mapping table, the part replacement mapping table, and the device data. In one example, the machine learning engine 106 may also analyze the probable cause to determine the parts to be replaced. If it is determined that the probability is greater than the threshold, the machine learning engine 106 may ascertain that the remote agent may be suggested to continue interacting with the user for remotely resolving the issue. [0057] The machine learning engine 106 may subsequently recommend probable resolution steps for resolving the issue. The probable solution steps may suggest whether the issue could be remotely resolved by the remote agent or the field visit by the field agent would be suggested. In one example, the recommended probable solution steps may further suggest probable actions that the remote agent may suggest to the user for resolving the issue or the parts that may be replaced during a field visit. For example, if it is suggested that the remote agent could resolve the issue remotely, the remote agent may be provided with probable resolution steps, including the probable actions, for resolving the issue. If a field visit by a field agent is suggested for resolving the issue, the machine learning engine 106 may provide a recommendation for parts to be examined, parts to be probably repaired, and parts to be carried for probable replacement for resolving the issue.

[0058] In one example, the machine learning engine 106 may also provide rationale for the recommendations provided to the remote agent. In one example, the rationale for the recommendation may provide reasonings and basis of determining the probable cause and recommending the probable resolution. For example, if the machine learning engine 106 identified a non-functional display as the probable cause and recommended the display to be replaced, based on presence of terms, such as “flickering display” or “screen not working” in the call log, the machine learning engine 106 may provide the rationale for recommendation as presence of terms “flickering display” in call log.

[0059] In one example implementation, the query engine 104 may render on the agent device 224 of the remote agent, at least one of the DHI, the PHI for a plurality of parts of the user device, the probable cause of the issue, the probability of remotely resolving the issue by the remote agent, recommendations for parts to be examined and probably repaired during a field visit for possible resolutions, recommendations for parts to be carried for probable replacement during the field visit for resolving the issue, recommendations for resolution steps to be provided to the user by the remote agent on call, and the rationale for the probable cause and recommendations of the probable resolution steps.

[0060] In one example, the DHI and the PHI for the plurality of parts of the user device 202 may be rendered as a bar graph, pie chart, line graph, cartesian graph, histogram, and the like. Further, the probable cause may be rendered on the agent device in form of text or a graphical representation of the issue. The query engine 104 may also render the possible resolutions, for example, in form of text, images or video, that the field agent may follow in order to resolve the issue. In one example, the query engine 104 may render at least one of a graphical representation of the part, name of the part, and a part ID on the agent device 224. In one example, the query engine 104 may render a probable list of parts that may be replaced or serviced on the agent device 224 of the remote agent. The remote agent may then analyze the lists of part and the probable cause to determine the parts that may be carried by the field agent. In another example, the machine learning engine 106 may recommend the parts that may be examined or carried by the field agent for replacement. The machine learning engine 106 may further generate a report to be used by the field agent. The report may include the call log, the list of recommended parts to the field agent for being carried, and recommendations for further diagnosis and examinations to be performed by the field agent for probable replacement during the field visit for resolving the issue to the field agent.

[0061] If it is determined that the probability is greater than the threshold, the machine learning engine 106 may recommend the remote agent to continue interacting with the user for remotely resolving the issue. In one example implementation, the machine learning engine 106 may recommended probable solution steps that may suggest probable actions that the remote agent may suggest to the user for resolving the issue. In one example, the query engine 104 may render, on the agent device 224, the DHI, the PHI for the plurality of parts of the user device 202 and recommendations for probable actions to be suggested to the user by the remote agent on call. For example, the query engine 104 may render text, image or video representation to the remote agent on the agent device 224. The remote agent may then provide the resolution steps to the user, for example, by guiding the user to follow the recommended steps by providing instructions verbally, in text, by images, videos or link to a webpage containing the resolution steps for solving the issue and the probable cause to suggest the probable actions to the user. The remote agent may further update the call log in real time by including the resolution steps recommended by the remote agent.

[0062] The query engine 104 may thus update the call log in real time based on the communication between the remote agent and the user to obtain the updated call log. The updated call log may capture new actions suggested by the remote agent to overcome the issue based on the probable cause and the user’s response after performing the actions. Further, the machine learning engine 106 may provide updated recommendations to the remote agent for remotely resolving the issue based on the updated call logs created by the remote agent and the resolution mapping table.

[0063] Figures 3, 4, and 5 illustrate example methods 300, 400, and 500, respectively, to provide support to a remote agent for resolving issues with user devices 202. The order in which the methods are described is not intended to be construed as a limitation, and any number of the described method blocks may be combined in any order to implement the methods, or an alternative method. Furthermore, methods 300, 400, and 500 may be implemented by processing resource(s) or computing device(s) through any suitable hardware, non-transitory machine-readable instructions, or a combination thereof.

[0064] It may also be understood that methods 300, 400, and 500 may be performed by programmed computing devices, such as the system 102, as depicted in Figures 1-2. Furthermore, the methods 300, 400, and 500 may be executed based on instructions stored in a non-transitory computer readable medium, as will be readily understood. The non-transitory computer readable medium may include, for example, digital memories, magnetic storage media, such as one or more magnetic disks and magnetic tapes, hard drives, or optically readable digital data storage media. While the methods 300, 400, and 500 are described below with reference to the user device 202 and the system 102 as described above. Other suitable systems for the execution of these methods may also be utilized. Additionally, implementation of these methods is not limited to such examples.

[0065] Figure 3 illustrates the method 300 for providing remote agent support, according to an example implementation of the present subject matter. At block 302, a device ID of a user device may be received from a communication between a user of the user device and a remote agent. In one example, the remote agent may receive a troubleshoot query from the user of the user device in the communication, the troubleshoot query indicating the device ID and the issue with the user device. The remote agent may then create a call log for the troubleshoot query to capture the issue, preliminary actions suggested by the remote agent to the user to overcome the issue, and response from the user for the preliminary actions provided by the remote agent. The remote agent may further provide the device ID to a query engine, such as the query engine 104 of the system 102 for getting support for resolving the issue.

[0066] At block 304, device data corresponding to the user device is obtained using the device ID. The device data may include at least one of telemetry data, audio data, image data, the call log, device status, and recent service and maintenance logs corresponding to the user device. In one example, a machine learning engine may obtain the device data from at least one of a central server and an agent device of the remote agent.

[0067] At block 306, probable resolution steps may be recommended to the remote agent for resolving the issue. In one example, the resolution steps may be determined and recommended based on the device data and mapping tables. The mapping tables include mappings between the issue, historical device data, and past solutions provided for solving the issue. In one example, the mapping tables may include a PHI mapping table, a resolution mapping table, and a part replacement mapping table. To provide the probable resolutions, a probability of remotely resolving the issue by the remote agent may be determined based at least on the device data and the mapping tables. For the probability being less than threshold, a field visit by a field agent for resolving the issue may be suggested. For the probability being greater than the threshold, the remote agent may be suggested to continue interacting with the user for remotely resolving the issue. The probable resolution steps may further include suggestions for actions to be suggested to the user by the remote agent to resolve the issue, if the remote agent is suggested to continue to provide remote resolution. The probable resolution steps may further include suggestions for parts to be replaced or repaired by the field agent, if the field visit is suggested.

[0068] Figure 4 illustrates the method 400 for training a machine learning model for providing remote agent support, according to an example implementation of the present subject matter. At block 402, historical device data is obtained for a plurality of user devices. In one example, the historical data may be obtained from a central server and a server of a customer support center. The historical device data may include at least one of historic telemetry data, historic audio data, historic image data, and historic call logs.

[0069] At block 404, a machine learning model may be trained using historical device data. In one example, the machine learning model may be trained to be used by a machine learning engine to determine cause for issues with a plurality of user devices and provide recommendations for resolving issues with the devices in a prediction phase. The machine learning model may use deep learning to analyze the historical device data and determine mappings and linkages between the past issues associated with multiple user device models, past solutions provided for solving the issues, remote solution probability generated in past for the issues, list of parts replaced in past for solving the issue.

[0070] At block 406, a resolution mapping table, a part health index (PHI) mapping table, and a part replacement mapping table may be generated based on the historical device data and the training. The PHI mapping table may indicate, for each part of the device, a mapping between the issue, historical device data, past solutions provided for solving the issue, and health index values. The resolution mapping table may indicate a mapping between the issue, the historical device data, the past solutions provided for solving the issue, and a remote resolution probability score. The part replacement mapping table may indicate a mapping between the issue, historical device data, and list of parts replaced in past for solving the issue. In one example, the machine learning model may analyze the issues and the past solutions, including whether parts were replaced or not to determine the mapping between the issue and the health index of parts to generate the PHI mapping table.

[0071] The machine learning model may further analyze the issues and the past solutions, including whether a part was replaced, to determine the mapping between the issue and the remote resolution probability score and generate the resolution mapping table. The machine learning model may further analyze the issues and the past solutions, including which parts were replaced or repaired, to determine the mapping between the issue and the list of parts replaced in past and generate the part replacement mapping table.

[0072] Figure 5 illustrates a method 500 for providing remote agent support, according to an example implementation of the present subject matter. At block 502, a device ID of a user device may be received from a communication between a user of the user device and a remote agent. In one example, the remote agent may receive a troubleshoot query from the user of the user device in the communication. The troubleshoot query may indicate the device ID and an issue with the user device. A call log for the troubleshoot query may then be created to capture the issue, preliminary actions suggested by the remote agent to the user to overcome the issue, and response from the user for the preliminary actions provided by the remote agent. The remote agent may further provide the device ID to a query engine, such as the query engine 104 of the system 102 for getting support for resolving the issue.

[0073] At block 504, device data corresponding to the user device is obtained using the device ID. In one example, the device data may include at least one of telemetry data, audio data, image data, the call log, device status, and recent service and maintenance logs corresponding to the user device. In one example, a machine learning engine may obtain the device data from at least one of a central server and an agent device of the remote agent. In one example implementation, the machine learning model may obtain entire or a combination of device data available for the user device. For instance, in case any type of data, such as the audio data or the image data is not available, the machine learning model may obtain and use the available device data, such as the call log or the telemetry data.

[0074] At block 506, part health index (PHI) for a plurality of parts of the user device and a device health index (DHI) for the user device may be determined based on the device data and a PHI mapping table. In one example, the PHI mapping table may indicate, for each part of the user device, a mapping between the issue, historical device data, and past solutions provided for solving the issue, and may thus be used to compute health for each part of the user device. The DHI may indicate overall health of the user device that may be computed as a mean, median or average of the PHI of the plurality of parts of the user device.

[0075] At block 508, the DHI of the user device and the PHI for the plurality of parts of the user device may be analyzed to identify a probable cause for the issue. In one example, by analyzing the DHI of the user device and the PHI for the plurality of parts of the user device, a machine learning engine, such as the machine learning engine 106 of system 102, may determine a part of the user device with low PHI value and may determine that the part may probably be damaged and may be a probable cause for the issue. The machine learning engine may further analyze the device data to identify the probable cause for a device with low PHI. For example, for a device with low PHI, the machine learning engine may analyze the telemetry data to ascertain if any error codes have been generated in past few days for the part. The machine learning engine may also analyze the service and maintenance logs to determine the time lapse and the usage of the part since the last time the device was replaced or repaired.

[0076] In one example, to determine the PHI for a plurality of parts of the user device and the DHI for the user device, the device data may be initially pre- processed. For example, to pre-process the call log, at least one of stemming, lemmatization, speech tagging, parsing, named-entity recognition, coreference resolution, and removal of word clutters from the call log may be performed. Further, to analyze the telemetry data, a telemetry vector representation may be generated to indicate for various error codes and part related information generated by the user device. Further, the image data may be processed using image processing techniques for generating an image matrix for each image of the image data. In one example, the image matrix for an image may be generated based on pixel data of the image. Further, for the audio data, an audio vector may be generated for each audio clip of the audio data. The recent service and maintenance logs corresponding to the user device may be analyzed to ascertain performance statistics of a plurality of parts of the user device.

[0077] At block 510, the DHI, the PHI for the plurality parts of the user device, and the probable cause may be rendered on an agent device of the remote agent. In one example, the DHI and the PHI for the plurality of parts of the user device may be rendered as, but not limited to, a bar graph, pie chart, line graph, cartesian graph, histogram, and the like. Further, the probable cause may be rendered on the agent device in form of text or a graphical representation of the probable cause and the issue.

[0078] At block 512, a probability of remotely resolving the issue by the remote agent may be determined based at least on the device data and mapping tables. In one example, the mapping tables may include the PHI mapping table, and a resolution mapping table. The resolution mapping table may indicate a mapping between the issue, the historical device data, the past solutions provided for solving the issue, and a remote resolution probability score. In one example implementation, to determine the probability of remotely resolving the issue, the machine learning engine, such as the machine learning engine 106 analyze the DHI and the PHI for a plurality of parts of the user device to identify a probable cause for the issue and render the DHI, the PHI for a plurality of parts, and the probable cause to an agent device of the remote agent.

[0079] At block 514, it is determined if the probability of remotely resolving the issue by the remote agent less than a threshold. In one example, the threshold be defined by a manufacturer of the user device, a service provider for the user device, or by the machine learning engine 106 during a training phase based on the historical device data and service and maintenance logs. If it is determined that the probability of remotely resolving the issue by the remote agent is less than the threshold (‘YES’ path from block 514), a field visit by the field agent may be suggested at block 516 for resolving the issue.

[0080] Further, at block 518, parts to be serviced may be determined based on the PHI mapping table, a part replacement mapping table and the device data. In one example implementation, the part replacement mapping table may indicate a mapping between the issue, historical device data, and list of parts replaced in past for solving the issue.

[0081] At block 520, a recommendation may be provided for possible resolutions and parts to be serviced and carried by the field agent. In one example, the recommendation may provide possible resolutions and parts to be serviced and carried by the field agent for probable replacement for resolving the issue based on the part replacement mapping table. In one example, the machine learning engine may render a probable list of parts that may be replaced or serviced on an agent device of the remote agent. The remote agent may then analyze the lists of parts and the probable cause to determine the parts that may be carried by the filed agent. In another example, the machine learning engine may recommend the parts that may be carried by the field agent for replacement.

[0082] If, at block 514, it is determined that the probability of remotely resolving the issue by the remote agent is greater than the threshold (‘NO’ path from block 514), the remote agent may be suggested to continue interacting with the user for remotely resolving the issue at block 522.

[0083] At block 524, updated recommendations may be provided to the remote agent on an agent device for resolving the issue. In one example, the DHI, the PHI for the plurality of parts of the user device, and the probable cause may be rendered on the agent device of the remote agent to assist the remote agent in further analyzing and understanding the issue. Further, an updated call log may be obtained and analyzed to further ascertain the issue. The updated call log may indicate updated actions suggested by the remote agent to the user to overcome the issue and response from the user for the updated actions provided by the remote agent. Further, updated recommendations may be dynamically provided for resolving the issue based on the resolutions mapping table and updated call logs.

[0084] Figure 6 illustrates an example network environment 600 using a non-transitory computer readable medium 602 for providing troubleshooting solution, according to an example implementation of the present subject matter. The network environment 600 may be public networking environment or a private networking environment. In one example, the network environment 600 may include processing resource 604 communicatively coupled to the non-transitory computer readable medium 602 through a communication link 606.

[0085] In an example, the processing resource 604 may be a processor of an electronic device, such as the system 102. The non-transitory computer readable medium 602 may be, for example an internal memory device or an external memory device. In one example, the communication link 606 may be a direct communication link, such as one formed through a memory read/ write interface. In another example, the communication link 606 may be an indirect communication link, such as one formed through a network interface. In such a case, the processing resource 604 may access the non-transitory computer readable medium 602 through a network 608. The network 608 may be a single network or a combination of multiple networks and may use a variety of communication protocols.

[0086] The processing resource 604 and the non-transitory computer readable medium 602 may also be communicatively coupled to data sources 610 over the network 608. The data sources 610 may include, for example, databases and computing devices. The data sources 610 may be used by the database administrators and other users to communicate with the processing resource 604. [0087] In one example, the non-transitory computer readable medium 602 may include a set of computer readable instructions, such as a query engine 612, and a machine learning engine 614. As would be understood, the query engine 612 implements the functionality of the query engine 104, and the machine learning engine 614 implements the functionality of the machine learning engine 106. The set of computer readable instructions, referred to as instructions hereinafter, can be accessed by the processing resource 604 through the communication link 606 and subsequently executed to perform acts for facilitating facsimile communication.

[0088] For discussion purposes, the execution of the instructions by the processing resource 604 has been described with reference to various components introduced earlier with reference to the description of Figs. 1-5. On execution by the processing resource 604, the query engine 612 may receive device data corresponding to a user device. In one example, a user of the user device may be in communication with a remote agent for obtaining solution for an issue with the user device. The issue may indicate a problem that the user may encounter while operating the electronic device. For example, a user may experience problems, such as boot failure, user error, hardware failure, software confliction, and programming error, while operating a computing device. Similarly, the user may experience problems, such as flickering, loose wire connection, no sound output, or screen blackout, while operating audio/video devices.

[0089] The device data includes at least one of telemetry data, audio data, image data, call log, device status, and recent service and maintenance logs corresponding to the user device, wherein the call log includes an issue indicated by a user of the user device and notes capturing conversation between a remote agent and the user.

[0090] The machine learning engine 614 may subsequently determine a probability of remotely resolving the issue by the remote agent based on the device data, a resolution mapping table, and a part health index (PHI) mapping table. In one example, the PHI mapping table may indicate, for each part of the user device, a mapping between the issue, historical device data, past solutions provided for solving the issue, and health index values. The resolution mapping table may indicate a mapping between the issue, the historical device data, the past solutions provided for solving the issue, and remote resolution probability score. [0091] In one example implementation, to determine the probability of remotely resolving the issue, the machine learning engine 614 may initially determine a PHI for a plurality of parts of the user device and a device health index (DHI) for the user device based on the device data and the PHI mapping table. The machine learning engine 614 may further analyze the DHI and the PHI for a plurality of parts of the user device to identify a probable cause for the issue.

[0092] In one example, the machine learning engine 614 may also provide a rationale for the recommendations provided to the remote agent. In one example, the rationale for the recommendation may provide reasonings and basis of determining the probable cause and recommending the probable resolution. For example, if the machine learning engine 614 identified a non-functional display as the probable cause and recommended the display to be replaced, based on presence of terms, such as “flickering display” or “screen not working” in the call log, the machine learning engine 614 may provide the rationale for recommendation as presence of terms “flickering display” in call log.

[0093] The machine learning engine 614 may further compare the probability of remotely resolving the issue by the remote agent with a threshold and recommend one of a field visit by a field agent and remotely resolving the issue by the remote agent as a probable resolution step for resolving the issue based on the comparison. In one example, for the probability of remotely resolving the issue by the remote agent being less than threshold, the machine learning engine 614 may suggest a field visit by a field agent for resolving the issue. The machine learning engine 614 may further provide a recommendation for parts to be examined and probably repaired for possible resolutions and parts to be carried for probable replacement for resolving the issue based on a part replacement mapping table. The part replacement mapping table may indicate a mapping between the issue, historical device data, and list of parts replaced in past for solving the issue.

[0094] Further, for the probability of remotely resolving the issue by the remote agent being greater than the threshold, the machine learning engine 614 may suggest the remote agent to continue interacting with the user for remotely resolving the issue. The machine learning engine 614 may further provide updated recommendations on the agent device for resolving the issue based on the resolution mapping table, and an updated call log.

[0095] In one example implementation, the query engine 612 may further render on an agent device of the remote agent, at least one of the DHI, the PHI for a plurality of parts of the user device, the probable cause of the issue, the probability of remotely resolving the issue by the remote agent, recommendations for parts to be examined and probably repaired during a field visit for possible resolutions, recommendations for parts to be carried for probable replacement during the field visit for resolving the issue, recommendations for resolution steps to be provided to the user by the remote agent on call, and the rationale for the probable cause and recommendations of the probable resolution steps.

[0096] Although examples for the present subject matter have been described in language specific to structural features and/or methods, it should be understood that the appended claims are not limited to the specific features or methods described. Rather, the specific features and methods are disclosed and explained as examples of the present subject matter.