WO2021027332A1 - 通信建立方法和系统 - Google Patents

通信建立方法和系统 Download PDF

Info

Publication number
WO2021027332A1
WO2021027332A1 PCT/CN2020/087642 CN2020087642W WO2021027332A1 WO 2021027332 A1 WO2021027332 A1 WO 2021027332A1 CN 2020087642 W CN2020087642 W CN 2020087642W WO 2021027332 A1 WO2021027332 A1 WO 2021027332A1
Authority
WO
WIPO (PCT)
Prior art keywords
customer service
question
current
customer
server
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2020/087642
Other languages
English (en)
French (fr)
Inventor
蔡东艳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
OneConnect Smart Technology Co Ltd
Original Assignee
OneConnect Smart Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by OneConnect Smart Technology Co Ltd filed Critical OneConnect Smart Technology Co Ltd
Publication of WO2021027332A1 publication Critical patent/WO2021027332A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/50Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00Commerce
    • G06Q30/06Buying, selling or leasing transactions
    • G06Q30/0601Electronic shopping [e-shopping]

Definitions

  • This application relates to the field of artificial intelligence technology, in particular to a communication establishment method and system.
  • a communication establishment method includes:
  • the server receives current consultation data sent by the terminal corresponding to the current client, where the current consultation data includes the client identification and current consultation questions;
  • the server stores the current consultation question and the customer identifier in a database correspondingly;
  • the server searches for similar questions corresponding to the current consulting question from a set of questions stored in advance, and when the similar question is successfully found, sends a reply corresponding to the similar question to the terminal;
  • the server After receiving the customer service distribution request, the server parses the customer service distribution request to obtain a customer identification, and queries the corresponding current consulting question from the database according to the customer identification;
  • the server determines the label corresponding to the customer ID according to the current consultation question, matches the determined label with the label corresponding to each customer service ID, determines the target customer service ID according to the matching result, and establishes the corresponding The communication connection between the terminal and the terminal corresponding to the client identifier.
  • a communication establishment system the system includes a server and a terminal corresponding to the current client; wherein:
  • the terminal is configured to, after receiving the reply, visualize the reply on the current interface and obtain feedback information corresponding to the reply through the current interface, and when it is identified that the feedback information is negative feedback, report to the
  • the server sends a customer service distribution request
  • the server is further configured to, after receiving the customer service assignment request, parse the customer service assignment request to obtain a customer ID, and query the corresponding current consultation question from the database according to the customer ID; according to the current consultation question Determine the label corresponding to the customer ID, match the determined label with the label corresponding to each customer service ID, determine the target customer service ID according to the matching result, and establish the terminal corresponding to the target customer service ID and the terminal corresponding to the customer ID Communication connection between.
  • the communication establishment method and system described above can greatly reduce the situation that a communication connection between the terminal corresponding to the customer and the terminal corresponding to the customer service identification is established multiple times because the assigned customer service cannot answer the question, thereby saving computer resources.
  • Figure 1 is an application scenario diagram of a communication establishment method in an embodiment
  • Figure 2 is a schematic flowchart of a communication establishment method in an embodiment
  • FIG. 3 is a schematic flowchart of step S206 in an embodiment
  • Figure 4 is a structural block diagram of a communication establishment system in an embodiment.
  • the communication establishment method provided in this application can be applied to the application environment shown in FIG. 1, including the terminal 102 and the server 104 corresponding to the current client.
  • the server 104 receives the current consultation question sent by the terminal corresponding to the current customer, it searches for similar questions corresponding to the current consultation question from the pre-saved question set, and when it successfully finds the similar question, sends the reply corresponding to the similar question to the current The terminal 102 corresponding to the customer. After receiving the reply, the terminal 102 visualizes the reply and obtains the feedback information for the reply.
  • the feedback information When the feedback information is identified as negative feedback, it sends a customer service distribution request to the server 104, and the server 104 After receiving the customer service assignment request, determine the label corresponding to the current customer according to the current consulting question, match the determined label with the label corresponding to the customer service identity, determine the target customer service identity according to the matching result, and establish the terminal corresponding to the target customer service identity to correspond to the current customer Communication connection between the terminals 102.
  • the terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices.
  • the server 104 may be implemented as an independent server or a server cluster composed of multiple servers.
  • a communication establishment method which includes the following steps:
  • Step S202 The server receives the current consultation data sent by the terminal corresponding to the current customer, and the current consultation data includes the customer identification and the current consultation question.
  • step S206 the server searches for similar questions corresponding to the current consulting question from the pre-stored question set, and when the similar question is successfully found, it sends a reply corresponding to the similar question to the terminal.
  • the current customer refers to the customer who currently consults through the terminal.
  • the current consultation question can be voice or text.
  • the server After receiving the current consultation question sent by the terminal, the server first performs voice recognition on the received voice and converts it into corresponding text.
  • Question set refers to a collection of common consultation questions. Each question in the question set has a corresponding response.
  • the questions in the question set can be the number of consultations counted by the R&D personnel after collecting historical consultation questions. More and more fixed questions. Therefore, after the server receives the current consultation question, it can store the current consultation question and the customer ID in the database, then search for similar questions from the pre-saved question set, and then send the reply corresponding to the similar question to the database in text or voice.
  • the terminal corresponding to the current client.
  • the server may calculate the cosine similarity between the current consultation question and each question in the question set.
  • calculating the cosine similarity between the current consulting question and each question in the question set includes: extracting keywords using the TF-IDF algorithm for the current consulting question and the questions in the question set, and obtaining the current consulting questions and questions based on the extracted keywords
  • the word frequency vector corresponding to each question in the concentration is calculated, and the cosine value of the angle between the two word frequency vectors is calculated.
  • the cosine value is the cosine similarity between the current consultation question and the question in the question concentration. It is understandable that when there are multiple similar problems, the problem with the greatest cosine similarity can be determined as the target similar problem and sent to the terminal.
  • x is the length of the string corresponding to the current consulting question
  • y is the length of the string corresponding to the question in the problem set
  • levenshtein is the edit distance.
  • Step S208 After receiving the reply, the terminal visualizes the reply on the current interface and obtains feedback information corresponding to the reply through the current interface. When it is identified that the feedback information is negative feedback, it sends a customer service distribution request to the server.
  • the terminal can convert the reply, including but not limited to displaying the reply to the current user in the form of text, picture or voice. Further, the terminal obtains the current customer’s Feedback information for this reply.
  • the terminal may ask the current customer a question in the form of text or voice, and obtain the answer of the current customer to the question, and obtain feedback information according to the answer. For example, the terminal can ask the current customer: "Are you satisfied with the reply?" or "Has your problem been solved?".
  • the terminal may ask the current customer a question in the form of text or voice, set corresponding answer options, and obtain feedback information according to the answer options selected by the customer.
  • the feedback information is negative feedback, it means that the current customer is not satisfied with the reply sent by the server.
  • the terminal judges that the current customer is not satisfied with the reply sent by the server based on the customer's feedback information, it sends a customer service assignment carrying the current customer ID to the server request.
  • step S210 after receiving the customer service distribution request, the server parses the customer service distribution request to obtain the customer identification, and queries the corresponding current consulting question from the database according to the customer identification.
  • Step S212 The server determines the label corresponding to the customer ID according to the current consultation question, matches the determined label with the label corresponding to each customer service ID, determines the target customer service ID according to the matching result, and establishes the terminal corresponding to the target customer service ID and the terminal corresponding to the customer ID Communication connection between.
  • the tag user corresponding to the current customer represents the feature or category of the question that the current customer is consulting
  • each customer service identification corresponds to at least one tag
  • the tag corresponding to the customer service identification includes at least the type of question that the customer is good at answering.
  • feature labels can be, for example, "pre-sales”, “after-sales”, “complaint”, “suggestion”, “price” and so on. Therefore, after the server determines the label corresponding to the current customer, it can match the label corresponding to the current customer with the label corresponding to each customer service ID, and then determine the target customer service ID according to the matching result, and finally establish the terminal and target corresponding to the customer ID
  • the communication connection between the terminals corresponding to the customer service identifier may be online communication or telephone communication.
  • the server determines the label corresponding to the customer ID according to the current consultation question, matches the determined label with the label corresponding to each customer service ID, and determines the target customer service ID according to the matching result, which specifically includes: The text corresponding to the current consultation question is input into the pre-trained text classifier, and the label corresponding to the current customer is determined according to the output result of the text classifier; the determined label is matched with the classification label corresponding to the customer service ID, and the target customer service is determined according to the matching result Identify the category, and determine the target customer service identifier from the customer service identifier set corresponding to the target customer service identifier category.
  • the text classifier can be obtained by training the text corresponding to the historical consultation through a machine learning algorithm. Before training, the text corresponding to the historical consultation question can be labeled first.
  • the label here is used to represent the category corresponding to the consultation question, for example, It is pre-sale, in-sale, after-sale, etc.
  • all customer service identifications can be classified in advance.
  • the customer service identifications can be classified according to the types of consulting questions that the corresponding customer service is good at, and the labels corresponding to the consulting questions Determine the classification label for this type of customer service identification, for example, classify the customer service identification corresponding to the customer service that is good at handling pre-sales consultation questions into one category, and the classification label of this type of customer service identification is "pre-sale"; will be good at handling in-sale consultation
  • the customer service identification corresponding to the customer service with the problem is divided into one category.
  • the classification label of this type of customer service identification is "On sale”; the customer service identification corresponding to the customer service who is good at handling after-sales consulting problems is divided into one category, and the classification label of this type of customer service identification is "After-sales". Since there are corresponding tags for the customer service ID and the question that the current customer is consulting, in this way, when the server assigns the customer service ID, it can determine the target customer service ID category corresponding to the customer service that is suitable for handling the current customer’s consultation problem according to the label matching, and then The target customer service identifier is determined in the customer service identifier set corresponding to the target customer service identifier category.
  • the accuracy of determining the current customer label can be improved, so that the final target customer service identifier is the most suitable customer service identifier.
  • the server determines the label corresponding to the customer ID according to the current consultation question, matches the determined label with the label corresponding to each customer service ID, and determines the target according to the matching result.
  • step S302 the server determines the first label corresponding to the current customer according to the current consultation question and obtains the label weight of each first label.
  • the current consultation question can be in the form of voice or text.
  • the server first converts the current consultation question in voice form into text form through voice recognition, and further, the server performs word segmentation on the text Process and extract keywords from the segmented text.
  • a database is pre-established on the server. The database stores a plurality of preset tags and keyword sets corresponding to each tag, and sets different weights for each tag. The weight of the tag is used to determine the value of the tag. Therefore, in this application, after the keywords are extracted, the server can compare the extracted keywords with the keyword set corresponding to each tag in the database.
  • the label is determined as the label corresponding to the current consultation question. It can be understood that there can be one or more current consultation questions, and there can also be one or more tags corresponding to the current customer. Among them, the label can be obtained by categorizing historical and current consulting questions through big data analysis. For example, the label can be price, after-sales, complaint, refund, and so on.
  • Step S304 The server obtains the second label of each customer service identifier in the current customer service identifier set.
  • the label of the customer service identification can be determined according to the type of consulting questions that the customer service can service corresponding to the customer service identification. For example, if a customer service can service complaints, the label of the customer service identification corresponding to the customer service is determined as a complaint, which is understandable.
  • One customer service logo can correspond to one or more tags.
  • Step 306 The server matches the second tag of each customer service identification with the first tag, and calculates the matching degree between the tag corresponding to the customer identification and the tag corresponding to each customer service identification according to the weight of the first tag that is successfully matched.
  • customer Zhang Xiao labels include A, B, and C, and the weights are 0.4, 0.2, and 0.1 respectively.
  • the labels of customer service logo 1 are B, C, and F, and the labels of customer service logo 2 are A, D, and E, then
  • Step S308 The server determines the target customer service identifier according to the first matching degree.
  • the server may sort the customer service identifiers according to the first matching degree, and determine the target customer service identifiers according to the sorting result. For example, the server may sort the customer service identifiers in descending order according to the first matching degree, and determine the customer service identifier ranked first as the target customer service identifier.
  • the current customer since the current customer may correspond to multiple tags, a weight is set for each tag, and the weights of the tags that match between the current customer and the customer service identification are respectively accumulated to obtain the tag corresponding to the customer identification and each customer service identification.
  • the matching degree of the label, and finally the target customer service identity is determined according to the matching degree, and the most suitable customer service identity can be determined according to the label matching.
  • determining the target customer service identifier according to the matching degree includes: the server sorts the customer service identifiers in descending order according to the matching degree, determining the first customer service identifier as the initial assigned customer service and obtaining the service status of the initially assigned customer service Score; when the service status score does not exceed the preset threshold, the initially assigned customer service is determined as the target customer service ID; when the service status score exceeds the preset threshold, the next customer service ID is determined as the initial assigned customer service, And repeat the steps of obtaining the service status score of the initially assigned customer service until the target customer service identification is determined.
  • the customer service corresponding to the customer service identifier is online customer service.
  • the service status score is used to characterize the busyness of the customer service corresponding to the customer service identifier. The larger the service status score, the busier the customer service.
  • the service status score can be determined in real time according to the current number of people serving and the maximum number of customers corresponding to the customer service ID. For example, if the current number of people serving for a customer service is 3, and the maximum number of people serving is 5, the service status score can be 60 .
  • the server after the server calculates the matching degree corresponding to each customer service ID, it can sort the customer service IDs in descending order according to the matching degree, and determine the first customer service ID as the initial assigned service ID and obtain the service ID.
  • Service status score determine whether the service status score exceeds the preset threshold, if not, determine the customer service ID as the target customer service ID, if yes, determine the next customer service ID as the initial assigned service ID and repeat The step of obtaining the service status score of the initial assignment of the customer service identifier, until the final target assignment of the customer service is determined.
  • the service status scores of the customer service identifiers are further obtained, and the target customer service identifiers are further determined according to the service status scores, which can reduce the waiting time of customers during consultation and improve user experience.
  • determining the target customer service identifier according to the matching degree includes determining the customer service identifier with the matching degree greater than a preset threshold as the initial assigned customer service, and obtaining the current waiting time corresponding to each initially assigned customer service; and setting the current waiting time to the shortest The initial assigned customer service is determined as the target customer service ID.
  • the customer service corresponding to the customer service identifier is telephone customer service.
  • the server can first filter the customer service IDs according to the matching degree threshold to obtain the initial assigned customer service ID. If there is only one initial assigned customer service ID, then the initial assigned service ID is determined as the target assigned customer service. If there are multiple initial assigned customer service IDs, the server The current waiting time of each initially assigned customer service identification can be further obtained, and the initial assigned customer service identification with the shortest current waiting time is determined as the target customer service identification.
  • q represents the number of incoming calls currently queued for access to the customer service identifier
  • a represents the call abandonment rate
  • t represents the average waiting time
  • s represents the level coefficient of the customer service identifier.
  • the current waiting time of the customer service identification is further obtained, and the target customer service identification is further determined according to the current waiting time, which can reduce the waiting time of the customer during consultation and improve the user experience .
  • a communication establishment system 400 including a server 402 and a terminal 404 corresponding to the current client; wherein:
  • the server 402 is used to receive the current consultation data sent by the terminal 404 corresponding to the current customer.
  • the current consultation data includes the customer identification and the current consultation question; the current consultation question and the customer identification are correspondingly stored in the database; Consult the similar question corresponding to the question, and when the similar question is successfully found, send the reply corresponding to the similar question to the terminal 404;
  • the terminal 404 is used to visualize the reply on the current interface after receiving the reply, and obtain feedback information corresponding to the reply through the current interface, and when the feedback information is identified as negative feedback, send a customer service assignment request to the server 402;
  • the server 402 is also configured to parse the customer service distribution request to obtain the customer ID after receiving the customer service distribution request, query the corresponding current consulting question from the database according to the customer ID; determine the label corresponding to the customer ID according to the current consulting question, and set the determined label Match the tags corresponding to each customer service identifier, determine the target customer service identifier according to the matching result, and establish a communication connection between the terminal 404 corresponding to the target customer service identifier and the terminal 404 corresponding to the customer identifier.
  • the server 402 is further configured to input the text corresponding to the current consultation question into the pre-trained text classifier, and determine the label corresponding to the customer ID according to the output result of the text classifier; and the determined label corresponds to the customer service ID According to the matching result, the target customer service identification category is determined, and the target customer service identification is determined from the customer service identification set corresponding to the target customer service identification category.
  • the server 402 is further configured to determine the first label corresponding to the customer ID according to the current consultation question and obtain the label weight of each first label; obtain the second label of each customer service ID in the current customer service ID set; The second label of each customer service identity is matched with the first label, and the matching degree between the label corresponding to the customer identity and the label corresponding to each customer service identity is calculated according to the weight of the first label that is successfully matched; the target customer service identity is determined according to the matching degree .
  • the server 402 is further configured to sort the customer service identifiers in descending order according to the matching degree, determine the customer service identifier ranked first as the initially assigned customer service identifier and obtain the service status score of the initially assigned customer service identifier; When the service status score does not exceed the preset threshold, the initially assigned customer service ID is determined as the target customer service ID; when the service status score exceeds the preset threshold, the next customer service ID is determined as the initial assigned customer service ID, and repeat The step of obtaining the service status score of the initially assigned customer service identifier until the target customer service identifier is determined.
  • the server 402 is further configured to determine the customer service identifier with a matching degree greater than a preset threshold as the initial assigned customer service identifier, and obtain the current waiting time corresponding to each initially assigned customer service identifier; assign the initial assignment with the shortest current waiting time
  • the customer service identification is determined as the target customer service identification.
  • Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • General Engineering & Computer Science (AREA)
  • Telephonic Communication Services (AREA)

Abstract

本申请涉及人工智能领域,提供了一种通信建立方法和系统。方法包括:服务器接收当前客户对应的终端发送的当前咨询数据,将当前咨询数据中的当前咨询问题与客户标识对应存储至数据库中,从预先保存的问题集中查找与当前咨询问题对应的类似问题,将类似问题对应的回复发送至终端;终端将回复可视化并获取与回复对应的反馈信息,当反馈信息为负面反馈时,向服务器发送客服分配请求;服务器解析客服分配请求以获取客户标识,并查询对应的当前咨询问题,根据当前咨询问题确定客户标识对应的标签,将确定的标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立目标客服标识对应的终端与客户标识对应的终端之间的通信连接。

Description

通信建立方法和系统
本申请要求于2019年8月13日提交中国专利局、申请号为201910746105.4,发明名称为“通信建立方法和系统”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及人工智能技术领域,特别是涉及一种通信建立方法和系统。
背景技术
随着计算机技术的飞速发展,越来越多的人选择网上购买物品或服务,通过官方网站或其他应用app等,在网上购买物品或服务时,经常需要对购买的物品或服务进行咨询。
发明人意识到,在传统技术中,此类网站一般都会提供人工客服服务,但是,通常情况下,后台在分配客服时,都是随机分配客服,导致分配的客服有时候无法满足用户的咨询需求,这种情况下,通常需要由分配的客服再次重新帮忙接通满足客户需求的客服,这样一来后台经常需要多次建立客户与人工客服之间的通信链路,导致了计算机资源的浪费。
发明内容
基于此,有必要针对上述技术问题,提供一种能够节省计算机资源的通信建立方法和系统。
一种通信建立方法,所述方法包括:
服务器接收当前客户对应的终端发送的当前咨询数据,所述当前咨询数据包括客户标识及当前咨询问题;
所述服务器将所述当前咨询问题与所述客户标识对应存储至数据库中;
所述服务器从预先保存的问题集中查找与所述当前咨询问题对应的类似问题,当成功查找到类似问题时,将所述类似问题对应的回复发送至所述终端;
所述终端接收到所述回复后,在当前界面将所述回复可视化并通过所述当 前界面获取与所述回复对应的反馈信息,当识别出所述反馈信息为负面反馈时,向所述服务器发送客服分配请求;
所述服务器在接收到所述客服分配请求后,解析所述客服分配请求以获取客户标识,根据所述客户标识从所述数据库中查询对应的当前咨询问题;
所述服务器根据所述当前咨询问题确定所述客户标识对应的标签,将确定的所述标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立所述目标客服标识对应的终端与所述客户标识对应的终端之间的通信连接。
一种通信建立系统,所述系统包括服务器以及当前客户对应的终端;其中:
所述服务器用于接收当前客户对应的终端发送的当前咨询数据,所述当前咨询数据包括客户标识及当前咨询问题;将所述当前咨询问题与所述客户标识对应存储至数据库中;从预先保存的问题集中查找与所述当前咨询问题对应的类似问题,当成功查找到类似问题时,将所述类似问题对应的回复发送至所述终端;
所述终端用于接收到所述回复后,在当前界面将所述回复可视化并通过所述当前界面获取与所述回复对应的反馈信息,当识别出所述反馈信息为负面反馈时,向所述服务器发送客服分配请求;
所述服务器还用于在接收到所述客服分配请求后,解析所述客服分配请求以获取客户标识,根据所述客户标识从所述数据库中查询对应的当前咨询问题;根据所述当前咨询问题确定所述客户标识对应的标签,将确定的所述标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立所述目标客服标识对应的终端与所述客户标识对应的终端之间的通信连接。
上述通信建立方法和系统,可以大大减少由于分配的客服无法解答问题而多次建立客户对应的终端与客服标识对应的终端之间的通信连接的情况,从而可以节省计算机资源。
附图说明
图1为一个实施例中通信建立方法的应用场景图;
图2为一个实施例中通信建立方法的流程示意图;
图3为一个实施例中步骤S206的流程示意图;
图4为一个实施例中通信建立系统的结构框图。
具体实施方式
本申请提供的通信建立方法,可以应用于如图1所示的应用环境中,包括当前客户对应的终端102、服务器104。服务器104在接收到当前客户对应的终端发送的当前咨询问题时,从预先保存的问题集中查找与当前咨询问题对应的类似问题,当成功查找到类似问题时,将类似问题对应的回复发送至当前客户对应的终端102,该终端102在接收到该回复后,将该回复可视化并获取对该回复的反馈信息,当识别出反馈信息为负面反馈时,向服务器104发送客服分配请求,服务器104在接收到客服分配请求后,根据当前咨询问题确定当前客户对应的标签,将确定的标签与客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立目标客服标识对应的终端与当前客户对应的终端102之间的通信连接。
其中,终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备,服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。
在一个实施例中,如图2所示,提供了一种通信建立方法,包括以下步骤:
步骤S202,服务器接收当前客户对应的终端发送的当前咨询数据,当前咨询数据包括客户标识及当前咨询问题。
步骤S204,服务器将当前咨询问题与客户标识对应存储至数据库中。
步骤S206,服务器从预先保存的问题集中查找与当前咨询问题对应的类似问题,当成功查找到类似问题时,将类似问题对应的回复发送至终端。
其中,当前客户指的是当前通过终端进行问题咨询的客户。当前咨询问题可以是语音或者文字,当当前咨询问题为语音时,服务器在接收到终端发送的当前咨询问题后,先对接收到的语音进行语音识别并转化为对应的文本。问题 集指的是一些常见的咨询问题所组成的集合,问题集中的每一个问题都保存了对应的回复,问题集中的问题可以是由研发人员对历史咨询问题进行收集后,统计出的咨询次数比较多且回复比较固定的问题。因此,服务器接收到当前咨询问题后,可以将当前咨询问题与客户标识对应存储至数据库中,然后从预先保存的问题集中查找类似问题,然后将类似问题对应的回复以文字或语音的方式发送至当前客户对应的终端。
在一个实施例中,服务器可计算当前咨询问题与问题集中每一个问题之间的余弦相似度,当存在余弦相似度大于预设阈值的问题时,则表示成功查找到类似问题。其中,计算当前咨询问题与问题集中每一个问题之间的余弦相似度包括:分别对当前咨询问题与问题集中的问题采用TF-IDF算法提取关键词,根据提取的关键词得到当前咨询问题与问题集中的问题各自对应的词频向量,计算两个词频向量之间的夹角的余弦值,该余弦值即为当前咨询问题与问题集中的问题之间的余弦相似度。可以理解的是,当存在多个类似问题时,可以将余弦相似度最大的问题确定为目标类似问题并发送至终端。
在另一个实施例中,服务器可计算当前咨询问题与问题集中每一个问题之间的字符串相似度,当存在字符串相似度大于预设阈值的问题时,则表示成功查找到类似问题。其中,计算当前咨询问题与问题集中每一个问题之间的字符串相似度,包括:首先计算当前咨询问题与问题集中问题之间的编辑距离,其中,编辑距离指的是从一个字符串修改到另一个字符串时,其中编辑单个字符(比如修改、插入、删除)所需要的最少次数。然后根据编辑距离计算当前咨询问题与问题集中问题之间的字符串相似度,公式为:
similarity=[max(x,y)-levenshtein]/max(x,y)
其中,x为当前咨询问题对应的字符串长度,y为问题集中的问题所对应的字符串长度,levenshtein为编辑距离。
在一个实施例中,当服务器没有在问题集中查找到类似问题时,直接根据当前咨询问题确定当前客户对应的标签,将确定的标签与客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,将当前客户对应的终端与目标客 服标识对应的终端之间建立通信连接。
步骤S208,终端接收到回复后,在当前界面将回复可视化并通过当前界面获取与回复对应的反馈信息,当识别出反馈信息为负面反馈时,向服务器发送客服分配请求。
具体地,终端接收到服务器发送的对当前咨询问题的回复后,可以将该回复可是化,包括但不限于以文字、图片或语音的形式向当前用户展示该回复,进一步,终端获取当前客户针对该回复的反馈信息。在一个实施例中,终端可以以文字或语音的形式向当前客户进行提问,并获取当前客户针对该提问的回答,根据该回答得到反馈信息。例如,终端可以询问当前客户:“您对该回复是否满意?”或者“您的问题是否得到了解决?”。在另一个实施例中,终端可以以文字或语音的形式向当前客户进行提问,并设置对应的回答选项,根据客户选择的回答选项得到反馈信息。
进一步,反馈信息为负面反馈时,表征当前客户对服务器发送的回复不满意,当终端根据客户的反馈信息判断出当前客户对服务器发送的回复不满意时,向服务器发送携带当前客户标识的客服分配请求。
可以理解的是,当终端根据客户的反馈信息判断出当前客户对服务器发送的回复满意或者终端未获取到当前客户的反馈信息时,终端将不会向服务器发送客服分配请求,此时,服务器不需要确定目标客服标识。
步骤S210,服务器在接收到客服分配请求后,解析客服分配请求以获取客户标识,根据客户标识从数据库中查询对应的当前咨询问题。
步骤S212,服务器根据当前咨询问题确定客户标识对应的标签,将确定的标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立目标客服标识对应的终端与客户标识对应的终端之间的通信连接。
本实施例中,当前客户对应的标签用户表征当前客户所咨询的问题的特征或类别,每一个客服标识对应至少一个标签,客服标识对应的标签中至少包括可以表征该客服所擅长回答的问题类型或特征的标签,这种标签例如可以是“售前”、“售后”、“投诉”、“建议”、“价格”等等。因此,服务器在确定了当前客户对应的标签后,可将当前客户对应的标签与每一个客服标识对应的标签进行匹 配,然后根据匹配结果来确定目标客服标识,最后建立客户标识对应的终端与目标客服标识对应的终端之间的通信连接。可以理解的是,本实施例中的通信连接可以是在线通信,也可以是电话通信。
上述通信建立方法中,服务器中预先保存了问题集,在接收到当前客户对应的终端发送的咨询问题后,服务器首先查找是否存在类似问题,若成功查找到类似问题,则直接将该类似问题对应的回复发送至终端,这样,如果该类似问题的回复可以解决客户的问题,则服务器不再需要进行客服分配,从而可以从一定程度上节省计算机资源,进一步,若该类似问题的回复无法使当前客户满意时,当前客户对应的终端还可以向服务器发送客服分配请求,服务器通过将当前客户对应的标签与客服对应的标签进行匹配,可以确定一个适合当前客户的人工客服对当前客户所咨询的问题进行解答,这样就可以大大减少由于分配的客服无法解答问题而多次建立客户标识对应的终端与客服标识对应的终端之间的通信连接的情况,从而可以节省计算机资源。
在一个实施例中,上述步骤S212中,服务器根据当前咨询问题确定客户标识对应的标签,将确定的标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,具体包括:服务器将当前咨询问题对应的文本输入至预先训练的文本分类器中,根据文本分类器的输出结果确定当前客户对应的标签;将确定的标签与客服标识对应的分类标签进行匹配,根据匹配结果确定目标客服标识类别,从目标客服标识类别对应的客服标识集合中确定目标客服标识。
其中,文本分类器可以通过机器学习算法对历史咨询对应的文本进行训练得到,在训练之前,可首先对历史咨询问题对应的文本打标签,这里的标签用于表征咨询问题对应的类别,例如可以是售前、售中、售后等。本实施例中,可事先对所有的客服标识进行分类,对客服标识进行分类时,可将客服标识按照其对应的客服所擅长的咨询问题的类别进行分类,并将该类咨询问题对应的标签确定为该类客服标识的分类标签,例如,将擅长处理售前类咨询问题的客服对应的客服标识分为一类,该类客服标识的分类标签为“售前”;将擅长处理售中咨询问题的客服对应的客服标识分为一类,该类客服标识的分类标签为“售中”; 将擅长处理售后咨询问题的客服对应的客服标识分为一类,该类客服标识的分类标签为“售后”。由于客服标识和当前客户所咨询的问题都存在对应的标签,这样,服务器在进行客服标识分配时,可以根据标签匹配来确定适合处理当前客户的咨询问题的客服对应的目标客服标识类别,然后从该目标客服标识类别对应的客服标识集合中确定目标客服标识。
上述实施例中,由于当前客户对应的标签由文本分类器的输出结果进行确定,可以提高对当前客户标签进行确定的准确性,从而使得最终确定的目标客服标识为最合适的客服标识。
在一个实施例中,如图3所示,上述步骤S212中,服务器根据当前咨询问题确定客户标识对应的标签,将确定的标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标,具体包括:
步骤S302,服务器根据当前咨询问题确定当前客户对应的第一标签并获取每一个第一标签的标签权重。
其中,当前咨询问题可以是语音的形式也可以是文本的形式,若当前咨询问题为语音形式时,服务器首先通过语音识别将语音形式的当前咨询问题转换为文本形式,进一步,服务器对文本进行分词处理,并对分词后的文本提取关键词。本实施例中,服务器上预先建立了一个数据库,该数据库中保存了预先设置的多个标签及各个标签对应的关键词集合,并对各个标签设置不同的权重,标签的权重用于确定标签的重要性,因此,本申请中,在提取到关键词后,服务器可以将提取的关键词与数据库中各个标签对应的关键词集合进行比对,当提取的关键词与任意一个标签对应的关键词集合中的关键词匹配上时,将该标签确定为当前咨询问题对应的标签。可以理解,当前咨询问题可以是一个或多个,得到的当前客户对应的标签也可以是一个或多个。其中,标签可以通过大数据分析对历史当前咨询问题进行分类得到,例如,标签可以是价格、售后、投诉、退款等等。
步骤S304,服务器获取当前客服标识集合中每一个客服标识的第二标签。
其中,客服标识的标签可以根据该客服标识对应的客服可服务的咨询问题类型进行确定,例如,某个客服可服务投诉类问题,则将客服对应的客服标识 的标签确定为投诉,可以理解,一个客服标识可以对应一个或多个标签。
步骤306,服务器将每一个客服标识的第二标签分别与第一标签进行匹配,根据匹配成功的第一标签的权重计算客户标识对应的标签与每一个客服标识对应的标签的匹配度。
举个例子,客户张晓的标签包括A、B、C,权重分别为0.4、0.2、0.1,客服标识1的标签为B、C、F,客服标识2的标签为A、D、E,则该客户张晓与客服标识1的匹配度为0.2+0.1=0.3,该客户与客服标识2的匹配度为0.4。
步骤S308,服务器根据第一匹配度确定目标客服标识。
具体的,服务器可以根据第一匹配度对客服标识进行排序,根据排序结果确定目标客服标识。例如,服务器可以根据第一匹配度对客服标识进行降序排列,将排在第一位的客服标识确定为目标客服标识。
上述实施例中,由于当前客户可能对应多个标签,对每个标签设置权重,分别累加当前客户与客服标识之间相匹配的标签的权重,得到客户标识对应的标签与每一个客服标识对应的标签的匹配度,最后根据匹配度来确定目标客服标识,根据标签匹配可以确定出最适合的客服标识,
在一个实施例中,根据匹配度确定目标客服标识,包括:服务器根据匹配度对各个客服标识进行降序排序,将排在第一位的客服标识确定为初始分配客服并获取初始分配客服的服务状态分值;当服务状态分值不超过预设阈值时,将初始分配客服确定为目标客服标识;当服务状态分值超过预设阈值时,将排在下一位的客服标识确定为初始分配客服,并重复获取初始分配客服的服务状态分值的步骤,直至确定目标客服标识。
本实施例中,客服标识对应的客服为在线客服。服务状态分值用于表征客服标识对应的客服的繁忙程度,服务状态分值越大,表示该客服越繁忙。服务状态分值可以根据客服标识对应的客服的当前服务人数及最大服务人数实时确定,例如,某个客服的当前服务人数为3,最大服务人数为5,则可得到其服务状态分值为60。
本实施例中,服务器在计算了各个客服标识对应的匹配度后,可根据匹配度对客服标识进行降序排列,将排在第一位的客服标识确定为初始分配客服标 识并获取该客服标识的服务状态分值,判断该服务状态分值是否超过预设阈值,若否,则将该客服标识确定为目标客服标识,若是,则将排在下一位的客服标识确定为初始分配客服标识并重复获取初始分配客服标识的服务状态分值的步骤,直至最终确定目标分配客服。
上述实施例中,在根据匹配度对客服标识进行排序后,进一步获取客服标识的服务状态分值,根据服务状态分值来进一步确定目标客服标识,可以减少客户在进行咨询时的等待时间,提高用户体验。
在一个实施例中,根据匹配度确定目标客服标识,包括将匹配度大于预设阈值的客服标识确定为初始分配客服,并获取每一个初始分配客服对应的当前等待时间;将当前等待时间最短的初始分配客服确定为目标客服标识。
本实施例中,客服标识对应的客服为电话客服。服务器可以首先根据匹配度阈值对客服标识进行筛选,得到初始分配客服标识,若初始分配客服标识只有一个,则将该初始分配客服标识确定为目标分配客服,若初始分配客服标识有多个,服务器可以进一步获取每一个初始分配客服标识的当前等待时间,将当前等待时间最短的初始分配客服标识确定为目标客服标识。
其中,初始分配客服标识的当前等待时间可以参照以下公式进行计算:score=[(q+1)×(1-a)×t]/s
其中,q表示当前接入该客服标识的来电排队个数,a表示来电放弃接入率,t表示平均等待时间,s表示该客服标识的级别系数。
上述实施例中,在根据匹配度对客服标识进行排序后,进一步获取客服标识的当前等待时间,根据当前等待时间来进一步确定目标客服标识,可以减少客户在进行咨询时的等待时间,提高用户体验。
应该理解的是,虽然图2-3的流程图中的各个步骤按照箭头的指示依次显示,但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明,这些步骤的执行并没有严格的顺序限制,这些步骤可以以其它的顺序执行。而且,图2-3中的至少一部分步骤可以包括多个子步骤或者多个阶段,这些子步骤或者阶段并不必然是在同一时刻执行完成,而是可以在不同的时刻执行,这些子步骤或者阶段的执行顺序也不必然是依次进行,而是可以与其它步骤或 者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
在一个实施例中,如图4所示,提供了一种通信建立系统400,包括服务器402以及当前客户对应的终端404;其中:
服务器402用于接收当前客户对应的终端404发送的当前咨询数据,当前咨询数据包括客户标识及当前咨询问题;将当前咨询问题与客户标识对应存储至数据库中;从预先保存的问题集中查找与当前咨询问题对应的类似问题,当成功查找到类似问题时,将类似问题对应的回复发送至终端404;
终端404用于接收到回复后,在当前界面将回复可视化并通过当前界面获取与回复对应的反馈信息,当识别出反馈信息为负面反馈时,向服务器402发送客服分配请求;
服务器402还用于在接收到客服分配请求后,解析客服分配请求以获取客户标识,根据客户标识从数据库中查询对应的当前咨询问题;根据当前咨询问题确定客户标识对应的标签,将确定的标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立目标客服标识对应的终端404与客户标识对应的终端404之间的通信连接。
在一个实施例中,服务器402还用于将当前咨询问题对应的文本输入至预先训练的文本分类器中,根据文本分类器的输出结果确定客户标识对应的标签;将确定的标签与客服标识对应的分类标签进行匹配,根据匹配结果确定目标客服标识类别,从目标客服标识类别对应的客服标识集合中确定目标客服标识。
在一个实施例中,服务器402还用于根据当前咨询问题确定客户标识对应的第一标签并获取每一个第一标签的标签权重;获取当前客服标识集合中每一个客服标识的第二标签;将每一个客服标识的第二标签分别与第一标签进行匹配,根据匹配成功的第一标签的权重计算客户标识对应的标签与每一个客服标识对应的标签的匹配度;根据匹配度确定目标客服标识。
在一个实施例中,服务器402还用于根据匹配度对各个客服标识进行降序排序,将排在第一位的客服标识确定为初始分配客服标识并获取初始分配客服 标识的服务状态分值;当服务状态分值不超过预设阈值时,将初始分配客服标识确定为目标客服标识;当服务状态分值超过预设阈值时,将排在下一位的客服标识确定为初始分配客服标识,并重复获取初始分配客服标识的服务状态分值的步骤,直至确定目标客服标识。
在一个实施例中,服务器402还用于将匹配度大于预设阈值的客服标识确定为初始分配客服标识,并获取每一个初始分配客服标识对应的当前等待时间;将当前等待时间最短的初始分配客服标识确定为目标客服标识。
关于通信建立系统的具体限定可以参见上文中对于通信建立方法的限定,在此不再赘述。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机程序来指令相关的硬件来完成,所述的计算机程序可存储于一非易失性计算机可读取存储介质中,该计算机程序在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM以多种形式可得,诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。

Claims (20)

  1. 一种通信建立方法,其中,所述方法包括:
    服务器接收当前客户对应的终端发送的当前咨询数据,所述当前咨询数据包括客户标识及当前咨询问题;
    所述服务器将所述当前咨询问题与所述客户标识对应存储至数据库中;
    所述服务器从预先保存的问题集中查找与所述当前咨询问题对应的类似问题,当成功查找到类似问题时,将所述类似问题对应的回复发送至所述终端;
    所述终端接收到所述回复后,在当前界面将所述回复可视化并通过所述当前界面获取与所述回复对应的反馈信息,当识别出所述反馈信息为负面反馈时,向所述服务器发送客服分配请求;
    所述服务器在接收到所述客服分配请求后,解析所述客服分配请求以获取客户标识,根据所述客户标识从所述数据库中查询对应的当前咨询问题;
    所述服务器根据所述当前咨询问题确定所述客户标识对应的标签,将确定的所述标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立所述目标客服标识对应的终端与所述客户标识对应的终端之间的通信连接。
  2. 根据权利要求1所述的方法,其中,所述当前咨询问题包括语音和文字,当所述当前咨询问题为语音时,所述服务器在接收到终端发送的当前咨询问题后,先对接收到的语音进行语音识别并转化为对应的文本。
  3. 根据权利要求1所述的方法,其中,所述服务器从预先保存的问题集中查找与所述当前咨询问题对应的类似问题包括:
    所述服务器计算当前咨询问题与问题集中每一个问题之间的余弦相似度,若余弦相似度大于预设阈值,则表示成功查找到类似问题。
  4. 根据权利要求3所述的方法,其中,分别对当前咨询问题与问题集中的问题采用TF-IDF算法提取关键词,根据提取的关键词得到当前咨询问题与问题集中的问题各自对应的词频向量,计算两个词频向量之间的夹角的余弦值,所述余弦值即为当前咨询问题与问题集中的问题之间的余弦相似度。
  5. 根据权利要求1所述的方法,其中,所述服务器从预先保存的问题集中查找与所述当前咨询问题对应的类似问题包括:
    所述服务器计算当前咨询问题与问题集中每一个问题之间的字符串相似度,若字符串相似度大于预设阈值,则表示成功查找到类似问题。
  6. 根据权利要求1所述的方法,其中,若所述服务器没有在问题集中查找到类似问题,直接根据当前咨询问题确定当前客户对应的标签,将确定的标签与客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,将当前客户对应的终端与目标客服标识对应的终端之间建立通信连接。
  7. 根据权利要求1所述的方法,其中,所述服务器根据所述当前咨询问题确定所述客户标识对应的标签,将确定的所述标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,包括:
    所述服务器将所述当前咨询问题对应的文本输入至预先训练的文本分类器中,根据所述文本分类器的输出结果确定所述客户标识对应的标签;
    所述服务器将确定的所述标签与客服标识对应的分类标签进行匹配,根据匹配结果确定目标客服标识类别,从所述目标客服标识类别对应的客服标识集合中确定目标客服标识。
  8. 根据权利要求1所述的方法,其中,所述服务器根据所述当前咨询问题确定所述客户标识对应的标签,将确定的所述标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,包括:
    所述服务器根据所述当前咨询问题确定所述客户标识对应的第一标签并获取每一个所述第一标签的标签权重;
    所述服务器获取当前客服标识集合中每一个客服标识的第二标签;
    所述服务器将每一个客服标识的第二标签分别与所述第一标签进行匹配,根据匹配成功的第一标签的权重计算所述客户标识对应的标签与每一个客服标识对应的标签的匹配度;
    所述服务器根据所述匹配度确定目标客服标识。
  9. 根据权利要求8所述的方法,其中,所述根据所述匹配度确定目标客服标识,包括:
    所述服务器根据所述匹配度对各个客服标识进行降序排序,将排在第一位的客服标识确定为初始分配客服标识并获取所述初始分配客服标识的服务状态 分值;
    当所述服务状态分值不超过预设阈值时,所述服务器将所述初始分配客服标识确定为目标客服标识;
    当所述服务状态分值超过预设阈值时,所述服务器将排在下一位的客服标识确定为初始分配客服标识,并重复获取所述初始分配客服标识的服务状态分值的步骤,直至确定所述目标客服标识。
  10. 根据权利要求8所述的方法,其中,所述根据所述匹配度确定目标客服标识,包括:
    所述服务器将匹配度大于预设阈值的客服标识确定为初始分配客服标识,并获取每一个初始分配客服标识对应的当前等待时间;
    所述服务器将当前等待时间最短的初始分配客服标识确定为所述目标客服标识。
  11. 一种通信建立系统,其中,所述系统包括服务器以及当前客户对应的终端;其中:
    所述服务器用于接收当前客户对应的终端发送的当前咨询数据,所述当前咨询数据包括客户标识及当前咨询问题;将所述当前咨询问题与所述客户标识对应存储至数据库中;从预先保存的问题集中查找与所述当前咨询问题对应的类似问题,当成功查找到类似问题时,将所述类似问题对应的回复发送至所述终端;
    所述终端用于接收到所述回复后,在当前界面将所述回复可视化并通过所述当前界面获取与所述回复对应的反馈信息,当识别出所述反馈信息为负面反馈时,向所述服务器发送客服分配请求;
    所述服务器还用于在接收到所述客服分配请求后,解析所述客服分配请求以获取客户标识,根据所述客户标识从所述数据库中查询对应的当前咨询问题;根据所述当前咨询问题确定所述客户标识对应的标签,将确定的所述标签与各个客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,建立所述目标客服标识对应的终端与所述客户标识对应的终端之间的通信连接。
  12. 根据权利要求11所述的系统,其中,所述当前咨询问题包括语音和文 字,当所述当前咨询问题为语音时,所述服务器在接收到终端发送的当前咨询问题后,先对接收到的语音进行语音识别并转化为对应的文本。
  13. 根据权利要求11所述的系统,其中,所述从预先保存的问题集中查找与所述当前咨询问题对应的类似问题包括:
    所述服务器计算当前咨询问题与问题集中每一个问题之间的余弦相似度,若余弦相似度大于预设阈值,则表示成功查找到类似问题。
  14. 根据权利要求13所述的系统,其中,分别对当前咨询问题与问题集中的问题采用TF-IDF算法提取关键词,根据提取的关键词得到当前咨询问题与问题集中的问题各自对应的词频向量,计算两个词频向量之间的夹角的余弦值,所述余弦值即为当前咨询问题与问题集中的问题之间的余弦相似度。
  15. 根据权利要求11所述的系统,其中,所述从预先保存的问题集中查找与所述当前咨询问题对应的类似问题包括:
    所述服务器计算当前咨询问题与问题集中每一个问题之间的字符串相似度,若字符串相似度大于预设阈值,则表示成功查找到类似问题。
  16. 根据权利要求11所述的系统,其中,若所述服务器没有在问题集中查找到类似问题,直接根据当前咨询问题确定当前客户对应的标签,将确定的标签与客服标识对应的标签进行匹配,根据匹配结果确定目标客服标识,将当前客户对应的终端与目标客服标识对应的终端之间建立通信连接。
  17. 根据权利要求11所述的系统,其中,所述服务器还用于将所述当前咨询问题对应的文本输入至预先训练的文本分类器中,根据所述文本分类器的输出结果确定所述客户标识对应的标签;将确定的所述标签与客服标识对应的分类标签进行匹配,根据匹配结果确定目标客服标识类别,从所述目标客服标识类别对应的客服标识集合中确定目标客服标识。
  18. 根据权利要求11所述的系统,其中,所述服务器还用于根据所述当前咨询问题确定所述客户标识对应的第一标签并获取每一个所述第一标签的标签权重;获取当前客服标识集合中每一个客服标识的第二标签;将每一个客服标识的第二标签分别与所述第一标签进行匹配,根据匹配成功的第一标签的权重计算所述客户标识对应的标签与每一个客服标识对应的标签的匹配度;根据所 述匹配度确定目标客服标识。
  19. 根据权利要求18所述的系统,其中,所述服务器还用于根据所述匹配度对各个客服标识进行降序排序,将排在第一位的客服标识确定为初始分配客服标识并获取所述初始分配客服标识的服务状态分值;当所述服务状态分值不超过预设阈值时,将所述初始分配客服标识确定为目标客服标识;当所述服务状态分值超过预设阈值时,将排在下一位的客服标识确定为初始分配客服标识,并重复获取所述初始分配客服标识的服务状态分值的步骤,直至确定所述目标客服标识。
  20. 根据权利要求18所述的系统,其中,所述服务器还用于将匹配度大于预设阈值的客服标识确定为初始分配客服标识,并获取每一个初始分配客服标识对应的当前等待时间;将当前等待时间最短的初始分配客服标识确定为所述目标客服标识。
PCT/CN2020/087642 2019-08-13 2020-04-29 通信建立方法和系统 Ceased WO2021027332A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910746105.4A CN110597624A (zh) 2019-08-13 2019-08-13 通信建立方法和系统
CN201910746105.4 2019-08-13

Publications (1)

Publication Number Publication Date
WO2021027332A1 true WO2021027332A1 (zh) 2021-02-18

Family

ID=68853972

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/087642 Ceased WO2021027332A1 (zh) 2019-08-13 2020-04-29 通信建立方法和系统

Country Status (2)

Country Link
CN (1) CN110597624A (zh)
WO (1) WO2021027332A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN120181864A (zh) * 2025-03-03 2025-06-20 山东八浚通信科技有限公司 基于人工智能的物联网行业智能客服监管调控系统

Families Citing this family (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110597624A (zh) * 2019-08-13 2019-12-20 深圳壹账通智能科技有限公司 通信建立方法和系统
CN111131453A (zh) * 2019-12-24 2020-05-08 中国平安财产保险股份有限公司 一种服务请求响应方法、装置、计算机设备及存储介质
CN111125336A (zh) * 2019-12-28 2020-05-08 上海智臻智能网络科技股份有限公司 信息提示方法、装置、设备及介质
CN111340354A (zh) * 2020-02-24 2020-06-26 上海东普信息科技有限公司 工单管控方法、系统、计算机设备和存储介质
CN111324786B (zh) * 2020-03-03 2023-11-07 北京京东振世信息技术有限公司 咨询问题信息的处理方法和装置
JP7268149B2 (ja) 2020-03-26 2023-05-02 ベイジン バイドゥ ネットコム サイエンス テクノロジー カンパニー リミテッド 問い合わせ情報の処理方法及び装置、電子機器、記憶媒体並びにコンピュータプログラム
CN111274490B (zh) * 2020-03-26 2024-01-19 北京百度网讯科技有限公司 咨询信息的处理方法和装置
CN114385898A (zh) * 2020-10-20 2022-04-22 腾讯科技(深圳)有限公司 服务人员推荐方法、终端设备、以及服务器
CN112287087A (zh) * 2020-11-16 2021-01-29 重庆满集网络科技有限公司 一种客服消息的显示方法及存储介质
CN112686528B (zh) * 2020-12-28 2024-04-05 京东科技控股股份有限公司 用于分配客服资源的方法、装置、服务器和介质
CN113177694B (zh) * 2021-04-06 2024-03-29 北京水滴科技集团有限公司 客户分配方法、装置、存储介质及计算机设备
CN113840006B (zh) * 2021-09-27 2023-07-11 深圳平安智慧医健科技有限公司 问诊接口的管理方法、装置、电子设备以及存储介质
CN114006987A (zh) * 2021-10-29 2022-02-01 联想(北京)有限公司 可视化交互方法及装置
CN114238596B (zh) * 2021-12-03 2024-10-18 北京声智科技有限公司 一种在线客服应答方法、装置、设备及计算机存储介质
CN114257524A (zh) * 2021-12-17 2022-03-29 马上消费金融股份有限公司 一种通讯建立方法、装置及设备
CN117041633A (zh) * 2022-02-14 2023-11-10 杭州脸脸会网络技术有限公司 虚拟智能客服方法、装置、电子装置和存储介质
CN114757430A (zh) * 2022-04-27 2022-07-15 杭州凡声优行科技有限公司 服务资源的分配方法及系统
CN114971652A (zh) * 2022-05-16 2022-08-30 山东浪潮爱购云链信息科技有限公司 针对企业采购平台的线上客服服务方法、设备及介质
CN115866065A (zh) * 2022-11-16 2023-03-28 竹间智能科技(上海)有限公司 一种服务资源调度方法、装置、计算机设备以及存储介质
CN116385016A (zh) * 2023-03-28 2023-07-04 陕西厚凯智能科技有限公司 一种企业管理在线咨询服务系统

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102624675A (zh) * 2011-01-27 2012-08-01 腾讯科技(深圳)有限公司 自助式客服系统、方法
US8391464B1 (en) * 2010-06-24 2013-03-05 Nuance Communications, Inc. Customer service system, method, and software program product for responding to queries using natural language understanding
CN105207890A (zh) * 2015-08-24 2015-12-30 北京智齿博创科技有限公司 一种在线客服的方法
CN105808694A (zh) * 2016-03-04 2016-07-27 上海携程商务有限公司 在线客服应答系统及方法
CN109389425A (zh) * 2018-09-24 2019-02-26 上海力自高实业有限公司 一种组合人工与ai的智能客服系统
CN109670666A (zh) * 2018-09-26 2019-04-23 深圳壹账通智能科技有限公司 理赔案件的自动分配方法、装置、终端及存储介质
CN110597624A (zh) * 2019-08-13 2019-12-20 深圳壹账通智能科技有限公司 通信建立方法和系统

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106572273B (zh) * 2016-10-19 2019-04-09 上海携程商务有限公司 用户来电接入客服线路的匹配系统和方法
CN107301229B (zh) * 2017-06-22 2019-10-01 美味不用等(上海)信息科技股份有限公司 基于语义分析的反馈指派方法及系统
CN110110049A (zh) * 2017-12-29 2019-08-09 深圳市优必选科技有限公司 服务咨询方法、装置、系统、服务机器人及存储介质
CN108170835A (zh) * 2018-01-12 2018-06-15 深圳市富途网络科技有限公司 一种组合人工与ai的智能客服系统
CN109947909B (zh) * 2018-06-19 2024-03-12 平安科技(深圳)有限公司 智能客服应答方法、设备、存储介质及装置
CN109857850A (zh) * 2019-01-18 2019-06-07 深圳壹账通智能科技有限公司 咨询请求处理方法、装置、计算机设备及存储介质
CN110059178A (zh) * 2019-02-12 2019-07-26 阿里巴巴集团控股有限公司 问题派发方法及装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8391464B1 (en) * 2010-06-24 2013-03-05 Nuance Communications, Inc. Customer service system, method, and software program product for responding to queries using natural language understanding
CN102624675A (zh) * 2011-01-27 2012-08-01 腾讯科技(深圳)有限公司 自助式客服系统、方法
CN105207890A (zh) * 2015-08-24 2015-12-30 北京智齿博创科技有限公司 一种在线客服的方法
CN105808694A (zh) * 2016-03-04 2016-07-27 上海携程商务有限公司 在线客服应答系统及方法
CN109389425A (zh) * 2018-09-24 2019-02-26 上海力自高实业有限公司 一种组合人工与ai的智能客服系统
CN109670666A (zh) * 2018-09-26 2019-04-23 深圳壹账通智能科技有限公司 理赔案件的自动分配方法、装置、终端及存储介质
CN110597624A (zh) * 2019-08-13 2019-12-20 深圳壹账通智能科技有限公司 通信建立方法和系统

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN120181864A (zh) * 2025-03-03 2025-06-20 山东八浚通信科技有限公司 基于人工智能的物联网行业智能客服监管调控系统

Also Published As

Publication number Publication date
CN110597624A (zh) 2019-12-20

Similar Documents

Publication Publication Date Title
WO2021027332A1 (zh) 通信建立方法和系统
CN111400607B (zh) 搜索内容输出方法、装置、计算机设备及可读存储介质
US10997258B2 (en) Bot networks
CN109272396B (zh) 客户风险预警方法、装置、计算机设备和介质
US10404629B2 (en) Automatic reply method, device, apparatus, and storage medium
KR102369876B1 (ko) 인공지능 기반의 비즈니스 중개 서비스 제공장치 및 방법
CN108388558B (zh) 问题匹配方法、装置、客服机器人和存储介质
WO2020143844A1 (zh) 意图分析方法、装置、显示终端及计算机可读存储介质
CN113223532A (zh) 客服通话的质检方法、装置、计算机设备及存储介质
CN108509424B (zh) 制度信息处理方法、装置、计算机设备和存储介质
WO2020077896A1 (zh) 提问数据生成方法、装置、计算机设备和存储介质
CN110059923A (zh) 岗位画像和简历信息的匹配方法、装置、设备及存储介质
CN114399396A (zh) 保险产品推荐方法、装置、计算机设备及存储介质
CN114610859B (zh) 基于内容与协同过滤的产品推荐方法、装置及设备
CN107784051A (zh) 在线客服应答系统及方法
US20140052445A1 (en) Voice search and response based on relevancy
CN107609023A (zh) 一种基于大数据的文字商标可注册性判断方法及装置
CN107633022A (zh) 人员画像分析方法、装置及存储介质
CN110516057A (zh) 一种信访问题答复方法及装置
CN116204624A (zh) 应答方法、装置、电子设备及存储介质
CN116720692A (zh) 客服派工方法、装置、计算机设备及存储介质
CN118261613B (zh) 一种基于ai的智能营销和身份认证方法及装置
CN111046151B (zh) 一种消息处理方法及装置
CN111382254B (zh) 电子名片推荐方法、装置、设备及计算机可读存储介质
CN117459637A (zh) 业务数据处理方法、装置、计算机设备和存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20852760

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20852760

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 03.08.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 20852760

Country of ref document: EP

Kind code of ref document: A1