An artificial intelligence customer service management method and system based on an enterprise knowledge base

Abstract

The present invention is applicable to the technical field of artificial intelligence customer service, and provides an artificial intelligence customer service management method and system based on an enterprise knowledge base. The present invention updates and obtains an actual live customer service video, performs entry recognition, exit recognition, and time period judgment, and intercepts multiple live standard segment videos; performs product recognition and speech recognition on the multiple live standard segment videos, and updates and constructs an enterprise knowledge base; receives a virtual customer service task and performs sequential virtual customer service live broadcasts; obtains user interaction information and selects a demand standard segment video; inserts the demand standard segment video into the virtual customer service live broadcast. It can intercept multiple live standard segment videos from the actual live customer service video, update and construct an enterprise knowledge base, receive virtual customer service tasks, perform sequential virtual customer service live broadcasts, obtain user interaction information, and select demand standard segment videos for inserting into the virtual customer service live broadcast, so as to perform more immersive and vivid video customer service based on the actual live customer service video.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence customer service, and in particular relates to an artificial intelligence customer service management method and system based on an enterprise knowledge base. Background Art

[0002] Artificial intelligence customer service uses artificial intelligence technology to simulate human customer service representatives to provide users with automated services. Through text chat, voice dialogue, etc., it understands the needs of users and provides services such as information query, question answering, business handling, and product recommendation. Artificial intelligence customer service has demonstrated great value in improving service efficiency, reducing costs, and enhancing user experience, and is widely used in multiple industries such as banking, telecommunications, e-commerce, medical care, and education.

[0003] In the prior art, artificial intelligence customer service usually only enables text communication and voice communication. However, with the rapid development of short videos and live broadcasts, video live broadcast customer service also has broad application prospects. The existing video live broadcast customer service only conducts simple communication through an AI image, which is essentially another manifestation of voice communication customer service and cannot provide a more immersive and vivid video customer service. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide an artificial intelligence customer service management method and system based on an enterprise knowledge base, aiming to solve the technical problems existing in the prior art mentioned in the background art.

[0005] The embodiments of the present invention are implemented as follows:

[0006] An artificial intelligence customer service management method based on an enterprise knowledge base, the method specifically includes the following steps:

[0007] Periodically update and obtain actual live broadcast customer service videos, perform entry recognition, exit recognition, and time period judgment, and intercept multiple live broadcast standard segment videos from the actual live broadcast customer service videos;

[0008] Perform commodity recognition and voice recognition on multiple live broadcast standard segment videos, record multiple corresponding standard segment annotation information, and update and construct an enterprise knowledge base;

[0009] Receive virtual customer service tasks, screen multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and perform sequential virtual customer service live broadcasts on multiple relevant standard segment videos;

[0010] Based on artificial intelligence technology, conduct live broadcast interaction guidance, obtain user interaction information, and based on multiple relevant annotation information, perform demand video matching and select demand standard segment videos;

[0011] Schedule the requirement insertion time, and at the requirement insertion time, insert a virtual customer service live broadcast into the requirement standard segment video.

[0012] As a further limitation of the technical solution of the embodiment of the present invention, the periodic update to obtain the actual live customer service video, perform entry recognition, exit recognition and time period judgment, and intercept multiple live standard segment videos from the actual live customer service video specifically includes the following steps:

[0013] Periodically obtain the actual live customer service video according to a preset live update period;

[0014] Extract the initialization segment video from the actual live customer service video;

[0015] Perform customer face recognition on the initialization segment video to obtain customer face feature data;

[0016] Perform frame-by-frame processing on the initialization segment video to obtain multiple initialization frame images;

[0017] According to the customer face feature data, perform feature matching and image rejection on the multiple initialization frame images to obtain the actual live background image;

[0018] According to the actual live background image, perform entry recognition, exit recognition and time period judgment on the actual live customer service video, and intercept multiple live standard segment videos.

[0019] As a further limitation of the technical solution of the embodiment of the present invention, the step of performing entry recognition, exit recognition and time period judgment on the actual live customer service video according to the actual live background image and intercepting multiple live standard segment videos specifically includes the following steps:

[0020] According to the actual live background image, perform entry recognition and exit recognition on the actual live customer service video, and record multiple customer entry times and multiple customer exit times;

[0021] Determine multiple corresponding customer presence time periods according to the multiple customer entry times and the multiple customer exit times;

[0022] Compare the multiple customer presence time periods with a preset standard presence time period;

[0023] Select multiple qualified presence time periods from the multiple customer presence time periods;

[0024] According to the multiple qualified presence time periods, intercept multiple live standard segment videos from the actual live customer service video.

[0025] As a further limitation of the technical solution of the embodiment of the present invention, the steps of performing product recognition and speech recognition on the multiple live standard segment videos, recording multiple corresponding standard segment annotation information, and updating and constructing an enterprise knowledge base specifically include the following steps:

[0026] Perform product recognition on the multiple live standard segment videos according to a preset product database, and record multiple corresponding product segment annotation information;

[0027] Perform speech recognition on the multiple live standard segment videos, and record multiple corresponding speech segment annotation information;

[0028] Comprehensively organize the multiple product segment annotation information and the multiple speech segment annotation information to generate multiple comprehensive segment annotation information;

[0029] Perform key recognition and information screening on the multiple comprehensive segment annotation information to generate multiple standard segment annotation information;

[0030] Update and construct an enterprise knowledge base according to the multiple live standard segment videos and the corresponding standard segment annotation information.

[0031] As a further limitation of the technical solution of the embodiment of the present invention, the steps of updating and constructing an enterprise knowledge base according to the multiple live standard segment videos and the corresponding standard segment annotation information specifically include the following steps:

[0032] Based on the enterprise knowledge base before the update, perform classification recognition on the multiple standard segment annotation information, and divide it into multiple existing segment annotation information and multiple non-existing segment annotation information;

[0033] Classify and divide the multiple live standard segment videos into multiple existing standard segment videos and multiple non-existing standard segment videos according to the multiple existing segment annotation information and the multiple non-existing segment annotation information;

[0034] Upgrade and update the enterprise knowledge base according to the multiple existing standard segment videos;

[0035] Supplement and update the enterprise knowledge base according to the multiple non-existing standard segment videos.

[0036] As a further limitation of the technical solution of the embodiment of the present invention, the steps of receiving a virtual customer service task, screening multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and performing sequential virtual customer service live broadcasts on the multiple relevant standard segment videos specifically include the following steps:

[0037] Receive a virtual customer service task;

[0038] Identify the virtual customer service task to determine multiple customer service display products and multiple customer service display items;

[0039] Display products according to multiple customer service representatives, and screen multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base;

[0040] Display projects according to multiple customer service representatives, and screen multiple project standard segment videos from multiple relevant standard segment videos;

[0041] Plan the customer service live broadcast order of multiple project standard segment videos;

[0042] Conduct virtual customer service live broadcast for multiple project standard segment videos according to the customer service live broadcast order.

[0043] As a further limitation of the technical solution of the embodiment of the present invention, the live broadcast interaction guidance is performed based on artificial intelligence technology to obtain user interaction information, and the demand video matching is performed based on multiple relevant annotation information, and the selection of the demand standard segment video specifically includes the following steps:

[0044] Perform live broadcast interaction guidance based on artificial intelligence technology;

[0045] Obtain user interaction information in real time;

[0046] Perform key recognition on the user interaction information and extract interaction key information;

[0047] Match target annotation information from multiple relevant annotation information according to the interaction key information;

[0048] Select the corresponding demand standard segment video from multiple relevant standard segment videos according to the target annotation information.

[0049] As a further limitation of the technical solution of the embodiment of the present invention, the planning of the demand insertion time, and the insertion of the demand standard segment video for virtual customer service live broadcast at the demand insertion time specifically includes the following steps:

[0050] Perform sequential positioning based on the customer service live broadcast order to determine the current live broadcast order;

[0051] Plan the demand insertion time according to the current live broadcast order;

[0052] Create a demand insertion task according to the demand insertion time;

[0053] Insert the demand standard segment video for virtual customer service live broadcast according to the demand insertion task.

[0054] An artificial intelligence customer service management system based on an enterprise knowledge base, the system includes an actual live broadcast processing module, a live broadcast video recognition module, a sequential virtual live broadcast module, a demand video matching module, and an insertion virtual live broadcast module, wherein:

[0055] An actual live processing module, which is used to periodically update and obtain the actual live customer service video, perform entry recognition, exit recognition and time period judgment, and intercept multiple live standard segment videos from the actual live customer service video;

[0056] A live video recognition module, which is used to perform commodity recognition and voice recognition on multiple live standard segment videos, record multiple corresponding standard segment annotation information, and update and construct an enterprise knowledge base;

[0057] A sequential virtual live module, which is used to receive virtual customer service tasks, screen multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and perform sequential virtual customer service live broadcast on multiple relevant standard segment videos;

[0058] A demand video matching module, which is used to perform live interaction guidance based on artificial intelligence technology, obtain user interaction information, perform demand video matching based on multiple relevant annotation information, and select demand standard segment videos;

[0059] An inserted virtual live module, which is used to plan the demand insertion time, and perform inserted virtual customer service live broadcast on the demand standard segment video at the demand insertion time.

[0060] As a further limitation of the technical solution of the embodiment of the present invention, the actual live processing module specifically includes:

[0061] A video acquisition unit, which is used to periodically acquire the actual live customer service video according to a preset live update period;

[0062] A segment video extraction unit, which is used to extract an initialization segment video from the actual live customer service video;

[0063] A face recognition unit, which is used to perform customer face recognition on the initialization segment video to obtain customer face feature data;

[0064] A frame-by-frame processing unit, which is used to perform frame-by-frame processing on the initialization segment video to obtain multiple initialization frame images;

[0065] A background acquisition unit, which is used to perform feature matching and image rejection on multiple initialization frame images according to the customer face feature data to obtain an actual live background image;

[0066] A segment video intercepting unit, which is used to perform entry recognition, exit recognition and time period judgment on the actual live customer service video according to the actual live background image, and intercept multiple live standard segment videos.

[0067] Compared with the prior art, the beneficial effects of the present invention are:

[0068] In the embodiment of the present invention, the actual live customer service video is obtained and updated, and entry recognition, exit recognition, and time period judgment are performed to intercept multiple live standard segment videos; commodity recognition and speech recognition are performed on the multiple live standard segment videos, and the enterprise knowledge base is updated and constructed; a virtual customer service task is received, and sequential virtual customer service live broadcast is performed; user interaction information is obtained, and a demand standard segment video is selected; and the demand standard segment video is inserted into the virtual customer service live broadcast. Multiple live standard segment videos can be intercepted from the actual live customer service video, the enterprise knowledge base can be updated and constructed, a virtual customer service task can be received, sequential virtual customer service live broadcast can be performed, and user interaction information can be obtained, and a demand standard segment video can be selected for insertion into the virtual customer service live broadcast, so as to perform more immersive and vivid video customer service based on the actual live customer service video. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 FIG. shows a flowchart of an artificial intelligence customer service management method based on an enterprise knowledge base provided by an embodiment of the present invention;

[0070] Figure 2 FIG. shows a flowchart of processing an actual live customer service video in the method provided by an embodiment of the present invention;

[0071] Figure 3 FIG. shows a flowchart of intercepting a live standard segment video in the method provided by an embodiment of the present invention;

[0072] Figure 4 FIG. shows a flowchart of identifying a live standard segment video in the method provided by an embodiment of the present invention;

[0073] Figure 5 FIG. shows a flowchart of updating and constructing an enterprise knowledge base in the method provided by an embodiment of the present invention;

[0074] Figure 6 FIG. shows a flowchart of sequential virtual customer service live broadcast in the method provided by an embodiment of the present invention;

[0075] Figure 7 FIG. shows a flowchart of performing demand video matching in the method provided by an embodiment of the present invention;

[0076] Figure 8 FIG. shows a flowchart of inserting a virtual customer service live broadcast in the method provided by an embodiment of the present invention;

[0077] Figure 9 FIG. shows an application architecture diagram of an artificial intelligence customer service management system based on an enterprise knowledge base provided by an embodiment of the present invention;

[0078] Figure 10 FIG. shows an application architecture diagram of an actual live processing module in the system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0079] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0080] It can be understood that artificial intelligence customer service has a wide range of application prospects. In the prior art, artificial intelligence customer service can usually only achieve text communication and voice communication. However, with the rapid development of short videos and live broadcasts, customer service for video live broadcasts also has broad application prospects. Existing video live broadcast customer service only conducts simple communication through an AI image, which is essentially another manifestation of voice communication customer service and cannot provide a more immersive and vivid video customer service.

[0081] To solve the above problems, an artificial intelligence customer service management method and system based on an enterprise knowledge base disclosed in an embodiment of the present invention periodically updates and obtains actual live customer service videos, performs entry recognition, exit recognition and time period judgment, and intercepts multiple live standard segment videos from the actual live customer service videos; performs product recognition and voice recognition on the multiple live standard segment videos, records multiple corresponding standard segment annotation information, and updates and constructs an enterprise knowledge base; receives virtual customer service tasks, screens multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and performs sequential virtual customer service live broadcasts on the multiple relevant standard segment videos; based on artificial intelligence technology, conducts live interaction guidance to obtain user interaction information, and based on the multiple relevant annotation information, performs demand video matching and selects demand standard segment videos; plans the demand insertion time, and at the demand insertion time, inserts the demand standard segment videos into the virtual customer service live broadcast. It can intercept multiple live standard segment videos from the actual live customer service videos, update and construct an enterprise knowledge base, receive virtual customer service tasks, perform sequential virtual customer service live broadcasts, obtain user interaction information, and select demand standard segment videos for insertion into the virtual customer service live broadcast, so as to perform a more immersive and vivid video customer service based on the actual live customer service videos.

[0082] Specifically, Figure 1 The flowchart of the artificial intelligence customer service management method based on an enterprise knowledge base provided by an embodiment of the present invention is shown.

[0083] In a preferred embodiment provided by the present invention, an artificial intelligence customer service management method based on an enterprise knowledge base, the method specifically includes the following steps:

[0084] Step S101, periodically update and obtain actual live customer service videos, perform entry recognition, exit recognition and time period judgment, and intercept multiple live standard segment videos from the actual live customer service videos.

[0085] In the embodiments of the present invention, during the actual live broadcast, live recording is performed. According to a preset live update period, the actual live customer service video recorded is periodically obtained. Then, according to a preset initialization period, the initialization segment video is extracted from the actual live customer service video. By dynamically identifying the initialization segment video, the live customer service for the actual live broadcast is determined, and model models, poster models, etc. in the initialization segment video are removed. Then, face recognition of the live customer service for the actual live broadcast is performed to obtain customer service facial feature data. Furthermore, the initialization segment video is processed frame by frame to obtain a plurality of initialization frame images. According to the plurality of customer service facial feature data, the initialization frame images with the live customer service are identified and removed from the plurality of initialization frame images. Then, from the remaining plurality of initialization frame images, the clearest initialization frame image is selected and marked as the actual live background image. With the actual live background image as a reference, entry recognition and exit recognition are performed on the actual live customer service video, and a plurality of customer service entry times and a plurality of customer service exit times of the live customer service in the actual live customer service video are recorded. According to the plurality of customer service entry times and the corresponding plurality of customer service exit times, a plurality of corresponding customer service presence periods are determined. Then, the plurality of customer service presence periods are compared with a preset standard presence period, and the customer service presence periods shorter than the standard presence period are marked as qualified presence periods, so as to screen and obtain a plurality of qualified presence periods. Furthermore, according to the plurality of qualified presence periods, a plurality of corresponding live standard segment videos are intercepted from the actual live customer service video.

[0086] It can be understood that during the initialization period of the actual live broadcast, live tests need to be carried out, and the initialization period for each live test can be set to be the same; the initialization period is the time length of the live test, which is determined by the test start time and the test end time.

[0087] It can be understood that the entry recognition and the exit recognition are the recognition of the pictures in the actual live customer service video that only have the actual live background image, so as to determine a plurality of customer service entry times and a plurality of customer service exit times. Specifically: when the picture in the actual live customer service video that only has the actual live background image is first recognized, the last time when only the actual live background image exists is recorded and determined as the customer service entry time; when the picture in the actual live customer service video that only has the actual live background image is recognized for the second time, the start time when only the actual live background image exists is recorded and determined as the customer service exit time; when the picture in the actual live customer service video that only has the actual live background image is recognized for the third time, the last time when only the actual live background image exists is recorded and determined as the customer service entry time;...; and so on alternately to determine a plurality of customer service entry times and a plurality of customer service exit times.

[0088] Specifically, Figure 2 The flowchart of the actual live customer service video processing in the method provided by the embodiments of the present invention is shown.

[0089] Among them, in the preferred embodiment provided by the present invention, the periodic update to obtain the actual live customer service video, perform entry recognition, exit recognition and time period judgment, and intercept multiple live standard segment videos from the actual live customer service video specifically includes the following steps:

[0090] Step S1011: Periodically obtain the actual live customer service video according to a preset live update period.

[0091] Step S1012: Extract the initialization segment video from the actual live customer service video.

[0092] Step S1013: Perform customer face recognition on the initialization segment video to obtain customer face feature data.

[0093] Step S1014: Perform frame-by-frame processing on the initialization segment video to obtain multiple initialization frame images.

[0094] Step S1015: According to the customer face feature data, perform feature matching and image elimination on the multiple initialization frame images to obtain the actual live background image.

[0095] Step S1016: According to the actual live background image, perform entry recognition, exit recognition and time period judgment on the actual live customer service video, and intercept multiple live standard segment videos.

[0096] Specifically, Figure 3 shows the flowchart of intercepting the live standard segment video in the method provided by the embodiment of the present invention.

[0097] Among them, in the preferred embodiment provided by the present invention, the step of performing entry recognition, exit recognition and time period judgment on the actual live customer service video according to the actual live background image, and intercepting multiple live standard segment videos specifically includes the following steps:

[0098] Step S10161: According to the actual live background image, perform entry recognition and exit recognition on the actual live customer service video, and record multiple customer entry times and multiple customer exit times.

[0099] Step S10162: Determine multiple corresponding customer presence time periods according to the multiple customer entry times and the multiple customer exit times.

[0100] Step S10163: Compare the multiple customer presence time periods with a preset standard presence time period.

[0101] Step S10164: Screen multiple qualified presence time periods from the multiple customer presence time periods.

[0102] Step S10165: According to the multiple qualified presence time periods, a plurality of live standard segment videos are captured from the actual live customer service video.

[0103] Furthermore, the artificial intelligence customer service management method based on the enterprise knowledge base also includes the following steps:

[0104] Step S102: perform product recognition and voice recognition on the plurality of live standard segment videos, record a plurality of corresponding standard segment annotation information, and update and construct an enterprise knowledge base.

[0105] In an embodiment of the present invention, according to a preset commodity database, commodity identification is performed on multiple live standard segment videos, and multiple corresponding commodity segment annotation information is recorded. Then, voice recognition of live customer service is performed on multiple live standard segment videos, and multiple corresponding voice segment annotation information is recorded. According to a preset annotation information sorting format, multiple commodity segment annotation information and multiple corresponding voice segment annotation information are comprehensively sorted to generate multiple comprehensive segment annotation information. Then, key identification is performed on the multiple comprehensive segment annotation information, and non-key information irrelevant to the commodity is eliminated. Multiple standard segment annotation information is screened from the multiple comprehensive segment annotation information to generate multiple standard segment annotation information. Based on the enterprise knowledge base before updating, relevant matching is performed on the multiple standard segment annotation information, and according to the relevant matching results, multiple standard The segment annotation information is classified and marked into multiple existing segment annotation information and multiple non-segment annotation information, and the live standard segment videos corresponding to the multiple existing segment annotation information are marked as existing standard segment videos, and the live standard segment videos corresponding to the multiple non-segment annotation information are marked as non-standard segment videos, thereby classifying the multiple live standard segment videos into multiple existing standard segment videos and multiple non-standard segment videos, and then according to the multiple existing standard segment videos and the multiple existing segment annotation information, the corresponding segment videos and annotation information in the enterprise knowledge base are replaced to realize the upgrade and update of the enterprise knowledge base. At the same time, according to the multiple non-standard segment videos and the multiple non-segment annotation information, the segment videos and annotation information are supplemented to the enterprise knowledge base to realize the supplement and update of the enterprise knowledge base.

[0106] It is understandable that the product segment annotation information includes product name, product style, product number and product brand, etc.

[0107] Specifically, Figure 4 A flow chart of live standard segment video recognition in the method provided by an embodiment of the present invention is shown.

[0108] Among them, in the preferred embodiment provided by the present invention, the performing of commodity recognition and voice recognition on a plurality of live standard segment videos, recording a plurality of corresponding standard segment annotation information, and updating and constructing an enterprise knowledge base specifically comprises the following steps:

[0109] Step S1021: Identify products for multiple of the live standard segment videos according to a preset product database, and record multiple corresponding product segment annotation information.

[0110] Step S1022: Perform speech recognition on multiple of the live standard segment videos, and record multiple corresponding speech segment annotation information.

[0111] Step S1023: Comprehensively organize multiple of the product segment annotation information and multiple of the speech segment annotation information to generate multiple comprehensive segment annotation information.

[0112] Step S1024: Perform key identification and information screening on multiple of the comprehensive segment annotation information to generate multiple standard segment annotation information.

[0113] Step S1025: Update and construct an enterprise knowledge base according to multiple of the live standard segment videos and the corresponding standard segment annotation information.

[0114] Specifically, Figure 5 The flowchart of updating and constructing an enterprise knowledge base in the method provided by the embodiment of the present invention is shown.

[0115] Among them, in the preferred embodiment provided by the present invention, the updating and constructing of the enterprise knowledge base according to multiple of the live standard segment videos and the corresponding standard segment annotation information specifically includes the following steps:

[0116] Step S10251: Based on the enterprise knowledge base before updating, perform classification identification on multiple of the standard segment annotation information, and divide multiple existing segment annotation information and multiple non-existing segment annotation information.

[0117] Step S10252: Classify and divide multiple of the live standard segment videos into multiple existing standard segment videos and multiple non-existing standard segment videos according to multiple of the existing segment annotation information and multiple of the non-existing segment annotation information.

[0118] Step S10253: Upgrade and update the enterprise knowledge base according to multiple of the existing standard segment videos.

[0119] Step S10254: Supplement and update the enterprise knowledge base according to multiple of the non-existing standard segment videos.

[0120] Furthermore, the artificial intelligence customer service management method based on the enterprise knowledge base further includes the following steps:

[0121] Step S103: Receive virtual customer service tasks, screen multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and perform sequential virtual customer service live broadcasts on multiple relevant standard segment videos.

[0122] In an embodiment of the present invention, when a virtual customer service application for an artificial intelligence customer service is required, a virtual customer service task is received, and then the virtual customer service task is identified to determine multiple customer service display products and multiple customer service display items that need to be displayed in the customer service. According to the multiple customer service display products, multiple relevant standard segment videos and corresponding relevant annotation information are screened from the enterprise knowledge base. Then, according to the multiple customer service display items, multiple item standard segment videos are screened from the multiple relevant standard segment videos, and the customer service live broadcast order of the multiple item standard segment videos is planned. Furthermore, according to the customer service live broadcast order, virtual customer service live broadcast is performed on the multiple item standard segment videos.

[0123] It can be understood that the customer service display items may include basic product introductions, clothing product dressing displays, product after-sales introductions, food product tasting displays, etc.

[0124] Specifically, Figure 6 FIG. shows a flowchart of sequential virtual customer service live broadcast in the method provided by the embodiment of the present invention.

[0125] Among them, in a preferred embodiment provided by the present invention, the steps of receiving the virtual customer service task, screening multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and performing sequential virtual customer service live broadcast on the multiple relevant standard segment videos specifically include the following steps:

[0126] Step S1031: Receive a virtual customer service task.

[0127] Step S1032: Identify the virtual customer service task to determine multiple customer service display products and multiple customer service display items.

[0128] Step S1033: Screen multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base according to the multiple customer service display products.

[0129] Step S1034: Screen multiple item standard segment videos from the multiple relevant standard segment videos according to the multiple customer service display items.

[0130] Step S1035: Plan the customer service live broadcast order of the multiple item standard segment videos.

[0131] Step S1036: Perform virtual customer service live broadcast on the multiple item standard segment videos according to the customer service live broadcast order.

[0132] Furthermore, the artificial intelligence customer service management method based on the enterprise knowledge base further includes the following steps:

[0133] Step S104: Based on artificial intelligence technology, conduct live interaction guidance to obtain user interaction information, and based on multiple pieces of the relevant annotation information, perform demand video matching to select the demand standard segment video.

[0134] In the embodiment of the present invention, during the process of virtual customer service live broadcast, based on artificial intelligence technology, conduct live interaction guidance for the users in the virtual live broadcast room, obtain user interaction information in real time, through key identification of the user interaction information, extract the interaction key information, and then according to the interaction key information, match the corresponding target annotation information from multiple relevant annotation information, and further according to the target annotation information, select the corresponding demand standard segment video from multiple relevant standard segment videos.

[0135] Specifically, Figure 7 The flowchart of performing demand video matching in the method provided by the embodiment of the present invention is shown.

[0136] Among them, in the preferred embodiment provided by the present invention, the step of based on artificial intelligence technology, conducting live interaction guidance, obtaining user interaction information, performing demand video matching based on multiple pieces of the relevant annotation information, and selecting the demand standard segment video specifically includes the following steps:

[0137] Step S1041: Based on artificial intelligence technology, conduct live interaction guidance.

[0138] Step S1042: Obtain user interaction information in real time.

[0139] Step S1043: Conduct key identification on the user interaction information to extract the interaction key information.

[0140] Step S1044: According to the interaction key information, match the target annotation information from multiple relevant annotation information.

[0141] Step S1045: According to the target annotation information, select the corresponding demand standard segment video from multiple relevant standard segment videos.

[0142] Furthermore, the artificial intelligence customer service management method based on the enterprise knowledge base further includes the following steps:

[0143] Step S105: Plan the demand insertion time, and at the demand insertion time, insert the demand standard segment video into the virtual customer service live broadcast.

[0144] In the embodiments of the present invention, based on the customer service live broadcast sequence, the virtual customer service live broadcast is sequenced to determine the current live broadcast sequence of the virtual customer service live broadcast. Then, according to the current live broadcast sequence, the demand insertion time is planned. According to the demand insertion time, a demand insertion task is created. Furthermore, according to the demand insertion task, the demand standard segment video is inserted for the virtual customer service live broadcast, so as to realize the change of the live broadcast according to the needs of the users in the virtual live broadcast room during the virtual customer service live broadcast, making the virtual live broadcast more user-friendly. For example, when the virtual customer service live broadcast is displaying the dressing of clothing products, if a user needs to know about delivery, exchange, etc., the demand standard segment video corresponding to the after-sales introduction of the corresponding product can be selected and inserted for display.

[0145] Specifically, Figure 8 FIG. shows the flowchart of inserting the virtual customer service live broadcast in the method provided by the embodiments of the present invention.

[0146] Among them, in the preferred embodiment provided by the present invention, the planning of the demand insertion time and the insertion of the demand standard segment video into the virtual customer service live broadcast at the demand insertion time specifically include the following steps:

[0147] Step S1051: Based on the customer service live broadcast sequence, perform sequence positioning to determine the current live broadcast sequence.

[0148] Step S1052: According to the current live broadcast sequence, plan the demand insertion time.

[0149] Step S1053: According to the demand insertion time, create a demand insertion task.

[0150] Step S1054: According to the demand insertion task, insert the demand standard segment video for the virtual customer service live broadcast.

[0151] Furthermore, Figure 9 FIG. shows the application architecture diagram of the artificial intelligence customer service management system based on the enterprise knowledge base provided by the embodiments of the present invention.

[0152] Among them, in another preferred embodiment provided by the present invention, an artificial intelligence customer service management system based on an enterprise knowledge base includes:

[0153] The actual live broadcast processing module 101 is used to periodically update and obtain the actual live broadcast customer service video, perform entry recognition, exit recognition and time period judgment, and intercept multiple live broadcast standard segment videos from the actual live broadcast customer service video.

[0154] In an embodiment of the present invention, live recording is performed during an actual live broadcast. The actual live broadcast processing module 101 periodically obtains the recorded actual live customer service video according to a preset live update period, and then extracts the initialization segment video from the actual live customer service video according to a preset initialization time period. By dynamically identifying the initialization segment video, the live customer service for the actual live broadcast is determined, and model models, poster models, etc. in the initialization segment video are removed. Then, face recognition of the live customer service for the actual live broadcast is performed to obtain customer face feature data. Furthermore, the initialization segment video is frame-by-frame processed to obtain multiple initialization frame images. According to the multiple customer face feature data, the initialization frame images with the live customer service are identified and removed from the multiple initialization frame images. Then, from the remaining multiple initialization frame images, the clearest initialization frame image is selected and marked as the actual live background image. Taking the actual live background image as a reference, entry recognition and exit recognition are performed on the actual live customer service video, and multiple customer entry times and multiple customer exit times of the live customer service in the actual live customer service video are recorded. According to the multiple customer entry times and the corresponding multiple customer exit times, multiple corresponding customer presence periods are determined. Then, the multiple customer presence periods are compared with a preset standard presence period, and the customer presence periods shorter than the standard presence period are marked as qualified presence periods, so as to screen and obtain multiple qualified presence periods. Furthermore, according to the multiple qualified presence periods, multiple corresponding live standard segment videos are intercepted from the actual live customer service video.

[0155] Specifically, Figure 10 Fig. shows the application architecture diagram of the actual live broadcast processing module 101 in the system provided by the embodiment of the present invention.

[0156] Among them, in the preferred embodiment provided by the present invention, the actual live broadcast processing module 101 specifically includes:

[0157] The video acquisition unit 1011 is used to periodically acquire the actual live customer service video according to a preset live update period.

[0158] The segment video extraction unit 1012 is used to extract the initialization segment video from the actual live customer service video.

[0159] The face recognition unit 1013 is used to perform face recognition on the live customer service of the initialization segment video to obtain customer face feature data.

[0160] The frame-by-frame processing unit 1014 is used to perform frame-by-frame processing on the initialization segment video to obtain multiple initialization frame images.

[0161] The background acquisition unit 1015 is used to perform feature matching and image removal on the multiple initialization frame images according to the customer face feature data to obtain the actual live background image.

[0162] A video clip extraction unit 1016 is configured to perform entry recognition, exit recognition, and time period judgment on the actual live customer service video according to the actual live background image, and extract multiple standard live video clips.

[0163] Furthermore, the artificial intelligence customer service management system based on the enterprise knowledge base further includes:

[0164] A live video recognition module 102 is configured to perform product recognition and speech recognition on multiple standard live video clips, record multiple corresponding standard clip annotation information, and update and construct the enterprise knowledge base.

[0165] In an embodiment of the present invention, the live video recognition module 102 performs product recognition on multiple standard live video clips according to a preset product database, records multiple corresponding product clip annotation information, then performs speech recognition on the live customer service of multiple standard live video clips, records multiple corresponding speech clip annotation information, comprehensively organizes multiple product clip annotation information and multiple corresponding speech clip annotation information according to a preset annotation information organization format, generates multiple comprehensive clip annotation information, then performs key recognition on multiple comprehensive clip annotation information, eliminates non-key information irrelevant to the product, screens and generates multiple standard clip annotation information from multiple comprehensive clip annotation information, based on the enterprise knowledge base before update, performs relevant matching on multiple standard clip annotation information, and classifies and marks multiple standard clip annotation information as multiple existing clip annotation information and multiple non-existing clip annotation information according to the result of the relevant matching, marks the standard live video clips corresponding to multiple existing clip annotation information as existing standard video clips, and marks the standard live video clips corresponding to multiple non-existing clip annotation information as non-existing standard video clips, so as to classify and divide multiple standard live video clips into multiple existing standard video clips and multiple non-existing standard video clips, and then replace the corresponding video clips and annotation information in the enterprise knowledge base according to multiple existing standard video clips and multiple existing clip annotation information, so as to realize the upgrade and update of the enterprise knowledge base. At the same time, according to multiple non-existing standard video clips and multiple non-existing clip annotation information, supplement the video clips and annotation information of the enterprise knowledge base, so as to realize the supplementary update of the enterprise knowledge base.

[0166] A sequential virtual live module 103 is configured to receive virtual customer service tasks, screen multiple relevant standard video clips and corresponding relevant annotation information from the enterprise knowledge base, and perform sequential virtual customer service live broadcast on multiple relevant standard video clips.

[0167] In an embodiment of the present invention, when a virtual customer service application for an artificial intelligence customer service is required, the sequential virtual live broadcast module 103 receives a virtual customer service task, then identifies the virtual customer service task to determine multiple customer service display products and multiple customer service display items that need to be displayed by the customer service in the virtual customer service task. According to the multiple customer service display products, multiple relevant standard segment videos and corresponding relevant annotation information are screened from the enterprise knowledge base. Then, according to the multiple customer service display items, multiple item standard segment videos are screened from the multiple relevant standard segment videos, and the customer service live broadcast order of the multiple item standard segment videos is planned. Furthermore, according to the customer service live broadcast order, virtual customer service live broadcast is performed on the multiple item standard segment videos.

[0168] The demand video matching module 104 is used to perform live broadcast interaction guidance based on artificial intelligence technology, obtain user interaction information, and perform demand video matching based on the multiple relevant annotation information to select a demand standard segment video.

[0169] In an embodiment of the present invention, during the virtual customer service live broadcast, the demand video matching module 104 performs live broadcast interaction guidance on the users in the virtual live broadcast room based on artificial intelligence technology, obtains user interaction information in real time, extracts interaction key information by performing key identification on the user interaction information, and then matches the corresponding target annotation information from the multiple relevant annotation information according to the interaction key information. Furthermore, according to the target annotation information, the corresponding demand standard segment video is selected from the multiple relevant standard segment videos.

[0170] The inserted virtual live broadcast module 105 is used to plan the demand insertion time, and perform inserted virtual customer service live broadcast on the demand standard segment video at the demand insertion time.

[0171] In an embodiment of the present invention, the inserted virtual live broadcast module 105 performs sequential positioning on the virtual customer service live broadcast based on the customer service live broadcast order, determines the current live broadcast order of the virtual customer service live broadcast, then plans the demand insertion time according to the current live broadcast order, creates a demand insertion task according to the demand insertion time, and further inserts the demand standard segment video for virtual customer service live broadcast according to the demand insertion task, realizing the change of the live broadcast according to the needs of the users in the virtual live broadcast room during the virtual customer service live broadcast, so that the virtual live broadcast is more user-friendly. For example, when the virtual customer service live broadcast is performing clothing product dressing display, if a user needs to understand the situation of delivery, exchange, etc., the demand standard segment video corresponding to the after-sales introduction of the corresponding product can be selected for insertion and display.

[0172] It should be understood that although the steps in the flowcharts of the embodiments of the present invention are shown sequentially in the direction of the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in each embodiment may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0173] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above-described embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-described methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0174] The above-described embodiments merely represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims

1. An artificial intelligence customer service management method based on an enterprise knowledge base, characterized in that, The method specifically includes the following steps: Periodically update and obtain the actual live customer service video, perform entry recognition, exit recognition, and time period judgment, and intercept multiple live standard segment videos from the actual live customer service video; Perform product recognition and speech recognition on multiple live standard segment videos, record multiple corresponding standard segment annotation information, and update and construct the enterprise knowledge base; Receive virtual customer service tasks, screen multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and perform sequential virtual customer service live broadcasts on multiple relevant standard segment videos; Based on artificial intelligence technology, conduct live interaction guidance, obtain user interaction information, and perform demand video matching based on multiple relevant annotation information to select demand standard segment videos; Plan the demand insertion time, and at the demand insertion time, perform an inserted virtual customer service live broadcast on the demand standard segment video; The step of periodically updating and obtaining the actual live customer service video, performing entry recognition, exit recognition, and time period judgment, and intercepting multiple live standard segment videos from the actual live customer service video specifically includes the following steps: Periodically obtain the actual live customer service video according to a preset live update cycle; Extract the initialization segment video from the actual live customer service video; Perform customer service face recognition on the initialization segment video to obtain customer service face feature data; Perform frame-by-frame processing on the initialization segment video to obtain multiple initialization frame images; According to the customer service face feature data, perform feature matching and image elimination on multiple initialization frame images to obtain the actual live background image; According to the actual live background image, perform entry recognition, exit recognition, and time period judgment on the actual live customer service video, and intercept multiple live standard segment videos.

2. The artificial intelligence customer service management method based on an enterprise knowledge base according to claim 1, characterized in that The step of performing entry recognition, exit recognition, and time period judgment on the actual live customer service video according to the actual live background image, and intercepting multiple live standard segment videos specifically includes the following steps: According to the actual live background image, perform entry recognition and exit recognition on the actual live customer service video, and record multiple customer service entry times and multiple customer service exit times; Determine multiple corresponding customer service presence time periods according to multiple customer service entry times and multiple customer service exit times; Compare multiple customer service presence time periods with a preset standard presence time period; Screen multiple qualified presence time periods from multiple customer service presence time periods; According to multiple qualified presence time periods, intercept multiple live standard segment videos from the actual live customer service video.

3. The artificial intelligence customer service management method based on an enterprise knowledge base according to claim 1, wherein The step of performing product recognition and speech recognition on multiple live standard segment videos, recording multiple corresponding standard segment annotation information, and updating and constructing the enterprise knowledge base specifically includes the following steps: Perform product recognition on multiple live standard segment videos according to a preset product database, and record multiple corresponding product segment annotation information; Perform speech recognition on multiple live standard segment videos, and record multiple corresponding speech segment annotation information; Comprehensively organize multiple product segment annotation information and multiple speech segment annotation information to generate multiple comprehensive segment annotation information; Perform key identification and information screening on multiple pieces of the comprehensive segment annotation information to generate multiple pieces of standard segment annotation information; Update and construct an enterprise knowledge base according to multiple pieces of the live standard segment videos and the corresponding standard segment annotation information.

4. The artificial intelligence customer service management method based on an enterprise knowledge base according to claim 3, wherein The step of updating and constructing an enterprise knowledge base according to multiple pieces of the live standard segment videos and the corresponding standard segment annotation information specifically includes the following steps: Based on the enterprise knowledge base before the update, perform classification identification on multiple pieces of the standard segment annotation information to divide them into multiple existing segment annotation information and multiple non-existing segment annotation information; According to multiple pieces of the existing segment annotation information and multiple pieces of the non-existing segment annotation information, classify and divide multiple pieces of the live standard segment videos into multiple existing standard segment videos and multiple non-existing standard segment videos; Upgrade and update the enterprise knowledge base according to multiple pieces of the existing standard segment videos; Supplement and update the enterprise knowledge base according to multiple pieces of the non-existing standard segment videos.

5. The artificial intelligence customer service management method based on an enterprise knowledge base according to claim 1, characterized in that The step of receiving a virtual customer service task, screening multiple relevant standard segment videos and the corresponding relevant annotation information from the enterprise knowledge base, and performing sequential virtual customer service live broadcast on multiple relevant standard segment videos specifically includes the following steps: Receive a virtual customer service task; Identify the virtual customer service task to determine multiple customer service displayed products and multiple customer service displayed items; Screen multiple relevant standard segment videos and the corresponding relevant annotation information from the enterprise knowledge base according to multiple pieces of the customer service displayed products; Screen multiple item standard segment videos from multiple pieces of the relevant standard segment videos according to multiple pieces of the customer service displayed items; Plan the customer service live broadcast sequence of multiple pieces of the item standard segment videos; Perform virtual customer service live broadcast on multiple pieces of the item standard segment videos according to the customer service live broadcast sequence.

6. The artificial intelligence customer service management method based on an enterprise knowledge base according to claim 1, wherein The step of performing live broadcast interaction guidance based on artificial intelligence technology, obtaining user interaction information, performing demand video matching based on multiple pieces of the relevant annotation information, and selecting a demand standard segment video specifically includes the following steps: Perform live broadcast interaction guidance based on artificial intelligence technology; Obtain user interaction information in real time; Perform key identification on the user interaction information to extract interaction key information; Match target annotation information from multiple pieces of the relevant annotation information according to the interaction key information; Select the corresponding demand standard segment video from multiple pieces of the relevant standard segment videos according to the target annotation information.

7. The artificial intelligence customer service management method based on an enterprise knowledge base according to claim 5, wherein The step of planning a demand insertion time and performing inserted virtual customer service live broadcast on the demand standard segment video at the demand insertion time specifically includes the following steps: Perform sequential positioning based on the customer service live broadcast sequence to determine the current live broadcast sequence; Plan the demand insertion time according to the current live broadcast sequence; Create a demand insertion task according to the demand insertion time; Perform virtual customer service live broadcast by inserting the demand standard segment video according to the demand insertion task.

8. An artificial intelligence customer service management system based on an enterprise knowledge base, characterized in that, The system includes an actual live broadcast processing module, a live broadcast video identification module, a sequential virtual live broadcast module, a demand video matching module, and an inserted virtual live broadcast module, where: An actual live processing module, which is used to periodically update and obtain actual live customer service videos, perform entry recognition, exit recognition and time period judgment, and intercept multiple live standard segment videos from the actual live customer service videos; A live video recognition module, which is used to perform commodity recognition and speech recognition on multiple live standard segment videos, record multiple corresponding standard segment annotation information, and update and construct an enterprise knowledge base; A sequential virtual live module, which is used to receive virtual customer service tasks, screen multiple relevant standard segment videos and corresponding relevant annotation information from the enterprise knowledge base, and perform sequential virtual customer service live broadcasts on multiple relevant standard segment videos; A demand video matching module, which is used to perform live interaction guidance based on artificial intelligence technology, obtain user interaction information, perform demand video matching based on multiple relevant annotation information, and select demand standard segment videos; An inserted virtual live module, which is used to plan the demand insertion time, and perform an inserted virtual customer service live broadcast on the demand standard segment video at the demand insertion time; The actual live processing module specifically includes: A video acquisition unit, which is used to periodically acquire actual live customer service videos according to a preset live update cycle; A segment video extraction unit, which is used to extract an initialization segment video from the actual live customer service video; A face recognition unit, which is used to perform customer service face recognition on the initialization segment video to obtain customer service face feature data; A frame-by-frame processing unit, which is used to perform frame-by-frame processing on the initialization segment video to obtain multiple initialization frame images; A background acquisition unit, which is used to perform feature matching and image rejection on multiple initialization frame images according to the customer service face feature data to obtain an actual live background image; A segment video interception unit, which is used to perform entry recognition, exit recognition and time period judgment on the actual live customer service video according to the actual live background image, and intercept multiple live standard segment videos.

Images