WO2019232883A1 - 推送保险产品的方法、装置、计算机设备和存储介质 - Google Patents

推送保险产品的方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2019232883A1
WO2019232883A1 PCT/CN2018/094885 CN2018094885W WO2019232883A1 WO 2019232883 A1 WO2019232883 A1 WO 2019232883A1 CN 2018094885 W CN2018094885 W CN 2018094885W WO 2019232883 A1 WO2019232883 A1 WO 2019232883A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
customer
insurance
profile information
format
Prior art date
Application number
PCT/CN2018/094885
Other languages
English (en)
French (fr)
Inventor
金戈
徐亮
肖京
Original Assignee
平安科技（深圳）有限公司
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 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2019232883A1 publication Critical patent/WO2019232883A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION 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
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/08—Insurance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION 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
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Item recommendations

Definitions

  • the present application relates to the field of computer technology, and in particular, to a method, an apparatus, a computer device, and a storage medium for pushing an insurance product.
  • the main purpose of this application is to provide a method, an apparatus, a computer device, and a storage medium for purposefully pushing an insurance product according to a customer's feedback on insurance information.
  • the present application further provides a computer device including a memory and a processor, where the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, implements the steps of any of the foregoing methods.
  • the present application also provides a computer non-volatile readable storage medium having computer-readable instructions stored thereon, which are executed by a processor to implement the steps of the method according to any one of the foregoing.
  • the method, device, computer equipment and storage medium for pushing an insurance product of this application first send insurance information to a user, and then analyze the user's emotional characteristics of the insurance information according to the user's reply text to determine whether the user needs to correspond to the insurance information. Insurance products, and then purposefully push the insurance products that users need. When the insurance information is pushed, the user's preferred push time period and information format are obtained, and then the insurance information is pushed in the corresponding information format during the user's push time period, so that the user can understand the insurance information well.
  • FIG. 1 is a schematic flowchart of a method for pushing an insurance product according to an embodiment of the present application
  • FIG. 2 is a schematic flowchart of a method for pushing an insurance product according to an embodiment of the present application
  • FIG. 3 is a schematic diagram of a specific process in the foregoing step S1 according to an embodiment of the present application;
  • FIG. 4 is a schematic flowchart of a method for pushing an insurance product according to an embodiment of the present application
  • step S1 is a schematic flowchart of step S1 in a method for pushing an insurance product according to an embodiment of the present application
  • step S4 is a schematic flowchart of step S4 in a method for pushing an insurance product according to an embodiment of the present application
  • FIG. 7 is a detailed flowchart of step S4 in the method for pushing an insurance product according to an embodiment of the present application.
  • FIG. 8 is a schematic block diagram of a structure of an apparatus for pushing an insurance product according to an embodiment of the present application.
  • FIG. 9 is a schematic block diagram of a structure of an apparatus for pushing an insurance product according to an embodiment of the present application.
  • FIG. 10 is a schematic block diagram of a sending module of an apparatus for pushing an insurance product according to an embodiment of the present application
  • FIG. 11 is a schematic block diagram of a device for pushing an insurance product according to an embodiment of the present application.
  • FIG. 12 is a schematic block diagram of a computer device according to an embodiment of the present application.
  • an embodiment of the present application provides a method for pushing an insurance product, including steps:
  • the insurance profile information includes information about insurance properties, such as the information related to property (cars, real estate) and physical (health status) mentioned above.
  • Disease related information refers to insurance profile information.
  • Insurance profile information exists in a variety of formats, including one or more combinations of picture format, text format, video format, audio format, and so on.
  • step S2 after sending an insurance message to the customer, after reading it, the customer makes some feedback in the form of text, and responds to the text message through the keyboard or touch screen. After voice recognition, the voice is converted into text information, and the robot obtains the reply text.
  • the robot inputs the response text into the trained sentiment analysis model, and extracts the emotional characteristics of the customer contained in the customer's response text.
  • the sentiment analysis model is a model based on a large number of trainings based on the LSTM-CNN model. It is used to make a result after a series of calculations on the input response text. The result indicates whether the customer's emotional characteristics are positive or negative.
  • the staff first defines a large amount of text for different business sentiment data, and then inputs a large amount of text and corresponding sentiment features into the LSTM-CNN model.
  • the sentiment features include positive and negative sentiments. Positive emotions indicate interest, while negative emotions indicate disinterest.
  • the LSTM-CNN model form a calculation model for emotional characteristics based on text, that is, an emotion analysis model. After the robot inputs the customer's response text into the sentiment analysis model, it can output the emotional characteristics corresponding to the customer's response text.
  • the insurance product information related to the insurance profile information is output.
  • the insurance product information is stored in the internal memory of the robot in advance, and the insurance product information includes the relevant insurance information tags.
  • the output method may be to display the picture information and / or text information of the insurance product on the display screen, or to play the text information of the insurance product through a speaker in the form of audio.
  • the method includes:
  • the insurance profile information sent in step S1 is determined as the information format preferred by the customer, so that when the insurance product information is pushed to the customer again later, the information is pushed in the customer preferred information format.
  • the foregoing sends the specified insurance profile information to the customer, and one or more of pictures or text or video or audio formats containing the insurance information are sent.
  • the information format of the insurance profile information is obtained, and the information format of the insurance information is determined by obtaining the suffix of the corresponding file name of the insurance profile information. If the suffix of the file name of the insurance profile information is one of doc and txt, the information format of the insurance profile information includes a text format; if the suffix of the file name of the insurance profile information is one of Mp4, rmvb, mkv, and avi If the suffix of the file name of the insurance profile information is one of Mp3, wma, amr, the information format of the insurance profile information includes the audio format; if the insurance profile information The suffix of the file name is one of jpeg, bmp, and gif, and the information format of the insurance profile information includes a picture format. Uploading the above information format to the server is convenient for the server to count the information format of the insurance profile information that the customer likes. When pushing the insurance product information or other information that needs to be displayed later,
  • the step of displaying the specified insurance profile information to the customer includes:
  • the robot calls the customer's customer information when the customer logs in to the personal insurance-related account, automatically obtains the customer's past browsing history, finds the customer preference information format in the historical browsing history, and sends the customer Insurance profile information to customers in preferred information format. If you are a new customer and have n’t browsed before, you can access the server and use the method preset in steps S5 and S6 above to generate a leaderboard based on the number of information formats preferred by all old users. The user likes the information format, and then displays the insurance profile information of this format information to the customer.
  • the method includes:
  • the robot when the insurance profile information is sent to the customer, the robot also records the sending time of the insurance profile information.
  • the customer receives a positive response to this insurance profile information, it indicates that the insurance profile information is approved. It indicates that the customer likes or approves the content of the insurance profile information, and it also indicates that the customer is convenient to receive and read the message during this time period.
  • Obtain the sending time of sending the insurance profile information and then determine the sending time period corresponding to the sending time, and determine the sending time period as the sending time period preferred by the customer.
  • Upload the sending time period to the server which is convenient for the server to count the time period when the customer likes to receive push insurance profile information.
  • select the sending time that the customer likes to push When pushing insurance profile information or other insurance product information in the future, select the sending time that the customer likes to push.
  • the server When determining the sending time period corresponding to the sending time, the server first divides the time of day into multiple time periods, for example, divides the time of day into the early morning (00: 00-06; 00) and the morning (06: 00-09: 00 ), Morning (09: 00-12: 00), noon (12: 00-14: 00), afternoon (14: 00-18: 00), evening (18: 00-21: 00), late night (21: 00-24: 00) these seven time periods. If the time for sending insurance profile information or other insurance product information to customers is 10:30 and between 9 and 12, the corresponding time period is morning, and the server determines that the customer's preferred sending time period is morning (09:00 -12: 00).
  • the step of displaying the specified insurance profile information to the customer includes:
  • the robot invokes the customer's customer information when the customer logs in to the personal insurance-related account, automatically obtains the personal information of the user, and correspondingly obtains the sending time period of the customer's preference calculated by the server. Then, it is determined whether the current time is within the sending time period of the client, and if it is, it is a time suitable for sending insurance information, and the designated insurance information is sent to the client.
  • the step of outputting insurance product information corresponding to the insurance profile information includes:
  • the emotional feature is a positive emotion
  • the above-mentioned insurance profile information is obtained, and the type of insurance profile information that the above-mentioned insurance profile information belongs to is obtained. If it is property insurance, the property insurance insurance is obtained. Type, further, the insurance product information under the insurance type is displayed for customers to browse and select.
  • one of the specified insurance profile information corresponds to multiple insurance types
  • the above-mentioned step of outputting insurance product information corresponding to the insurance profile information includes:
  • S423 Display the insurance product information corresponding to the insurance type selected by the user on a display screen.
  • the following table is the relationship among the insurance information, insurance type, and insurance product of a specific embodiment.
  • An insurance information corresponds to multiple insurance types, and each insurance type corresponds to multiple insurance products. If all these are displayed on the display screen, the amount of information is large, and the customer cannot quickly and accurately obtain the required product, so For further screening, the insurance type corresponding to the insurance information is displayed on the display screen, and the customer further optimizes and selects the direction. After receiving the insurance type selected by the customer, the robot finally displays only the insurance products of this insurance type on the screen. Convenient for customers to choose the insurance product they want.
  • the step of outputting an insurance product related to the insurance information includes:
  • each insurance product also has at least one product key for indicating the characteristics of the product. Words, such as life, cars, houses, gas, etc. can reflect the subject matter of the insurance products, but also include accidents, medical, financial management, dividends, investment and other words that reflect the main functions of insurance products. If the customer's reply text contains the above keywords, the robot will recommend the insurance product containing the above keywords to the user, and the insurance product will be displayed on the screen for the customer to choose.
  • the method includes:
  • the robot calls its own camera to take a picture.
  • customers interact with the robot, they usually stand or sit at the designated position. After the robot controls the camera to turn on, it looks for the face and then shoots.
  • the closest face information of the robot is the face information of the customer.
  • a facial expression of the facial information is determined. Judging facial expressions on facial information is to input a picture containing facial information into a model trained based on RNN neural network and then input the corresponding facial expression.
  • the output facial expression is positive or negative, if it is positive Expressions, positive expressions include smiles, concerns, and other expressions, it is further confirmed that the customer is interested in the above insurance information.
  • the specific method for determining whether a facial expression is a positive expression is as follows: a robot performs feature extraction on an image containing facial information to obtain a corresponding facial feature vector, and then performs facial expression recognition on the facial feature vector. If the recognition result belongs to a preset positive expression, it indicates that the user is a positive emotion.
  • the step of outputting an insurance product related to the insurance information includes:
  • the robot calls its own camera to take a picture.
  • customers interact with the robot, they usually stand or sit at the designated position. After the robot controls the camera to turn on, it looks for the face and then shoots.
  • the age of the customer corresponding to the facial information is determined. Judging the age and gender of the corresponding customer on the face information is to input the picture containing the facial information into a model trained based on the RNN neural network and then enter the age and gender of the corresponding customer.
  • each insurance product is also preset with a suitable age and gender. Only those who meet the age and gender requirements can purchase the corresponding insurance product.
  • the age range required for some disease insurance is relatively small, while the age range required for dividend insurance is relatively large. Therefore, when displaying insurance products, according to the age and gender of the customer, display the insurance products that the customer is interested in and those that can be purchased, filter out the insurance products that cannot be purchased, and further provide customers with accurate purchase insurance product.
  • the method for pushing an insurance product of this application first sends insurance information to the user, and then analyzes the user's emotional characteristics of the insurance information according to the user's reply text to determine whether the user needs the insurance product corresponding to the insurance information. Then purposefully push the insurance products that users need.
  • the insurance information is pushed, the user's preferred push time period and information format are obtained, and then the insurance information is pushed in the corresponding information format during the user's push time period, so that the user can understand the insurance information well.
  • an embodiment of the present application further provides a device for pushing an insurance product, including:
  • a display module 1 for displaying specified insurance profile information to customers
  • An obtaining module 2 configured to obtain a response text of the customer for the specified insurance profile information
  • Emotional feature module 3 configured to input the reply text into a preset sentiment analysis model trained based on the LSTM-CNN model and perform calculations to obtain the emotional feature of the customer with respect to the insurance information. Including positive emotions and negative emotions; the sentiment analysis model is obtained by training according to training samples of response texts of multiple known sentiment features;
  • An output module 4 is configured to output insurance product information corresponding to the specified insurance profile information if the emotional feature is a positive emotion.
  • the insurance is mainly divided into property insurance and personal insurance.
  • the insurance profile information includes information about insurance properties, such as the information related to property (cars, real estate) and physical (health status, disease) mentioned above. ) Related information refers to insurance profile information.
  • the display module 1 displays the specified insurance profile information to the customer, only one type of insurance profile information is displayed. For example, it only displays the insurance profile information related to property insurance, and does not send personal insurance related insurance profile information, which is convenient for customers to provide intuitive feedback on one of the insurance profile information, so that the system can obtain intuitive feedback of one insurance profile information .
  • Insurance profile information exists in a variety of formats, including one or more combinations of picture format, text format, video format, audio format, and so on.
  • the display module 1 sends an insurance profile to the customer, after reading the information, the customer makes some feedback in the form of text, and responds to the text information through the keyboard or touch screen.
  • the acquisition module 2 obtains the customer's response text.
  • the acquisition module 2 obtains the voice of the customer and converts his voice into text information, and then obtains the reply text.
  • the sentiment feature module 3 inputs the response text into the trained sentiment analysis model, and extracts the sentiment features of the client included in the reply text of the client.
  • the sentiment analysis model is a model based on a large number of trainings based on the LSTM-CNN model. It is used to make a result after a series of calculations on the input response text. The result indicates whether the customer's emotional characteristics are positive or negative.
  • the staff first defines a large amount of text for different business sentiment data, and then inputs a large amount of text and corresponding sentiment features into the LSTM-CNN model.
  • the sentiment features include positive sentiment and negative sentiment. Positive emotions indicate interest, while negative emotions indicate disinterest.
  • the LSTM-CNN model form a calculation model for emotional characteristics based on text, and obtains an emotion analysis model.
  • the emotion feature module 3 inputs the customer's reply text into the emotion analysis model, it can output the emotion feature corresponding to the customer's reply text. If it is a positive emotion, indicating that the customer is interested in the insurance profile information, the output module 4 outputs insurance product information related to the insurance profile information.
  • the insurance product information is stored in the internal memory of the robot in advance, and the insurance product information includes the relevant insurance information tags.
  • the output mode of the output module 4 may be displaying the picture information and / or text information of the insurance product on the display screen, or may be playing the text information of the insurance product through a speaker in the form of audio.
  • the above device for pushing an insurance product further includes:
  • An information format module 5 configured to obtain an information format of the insurance profile information
  • a first determining module 6 configured to determine that the information format is an information format preferred by the customer
  • the first calling module 7 is configured to call the client's preferred information format for information display when receiving an instruction to display information to the customer.
  • the first determination module 6 determines the information format of the insurance profile information displayed by the display module 1 as the customer's preferred information format, so that when the insurance profile information is pushed to the customer again later, it is pushed in the customer's preferred information format.
  • the foregoing sends the specified insurance profile information to the customer, and one or more of pictures or text or video or audio formats containing the insurance profile information are sent.
  • the information mode module 5 obtains the information format of the insurance profile information, and determines the information format of the insurance profile information by obtaining the suffix of the corresponding file name of the insurance profile information. If the suffix of the file name of the insurance profile information is one of doc and txt, the information format of the insurance profile information includes a text format; if the suffix of the file name of the insurance profile information is one of Mp4, rmvb, mkv, and avi The information mode module 5 determines that the information format of the insurance profile information includes a video format.
  • the information mode module 5 determines that the information format of the insurance profile information includes Audio format; if the suffix of the file name of the insurance profile information is one of jpeg, bmp, and gif, the information mode module 5 determines that the information format of the insurance profile information includes a picture format. Upload the above information format to the server, which is convenient for the server to count the information format of the insurance information that the customer likes. When the insurance product information or other information needs to be displayed later, the first calling module 7 selects the information format that the customer likes to push or display. .
  • the above display module 1 includes:
  • An obtaining information unit 11 is configured to obtain the customer information of the customer, and invoke a historical browsing record of the customer according to the customer information, where the historical browsing record includes an information format of the browsing information;
  • the first fetching unit 12 is configured to, if fetching the historical browsing history of the customer, determine the information format with the most views in the historical browsing history as the preferred information format of the customer, and display the format Insurance profile information corresponding to the preferred information format;
  • the second calling unit 13 is configured to obtain a ranking list of all customer preference information formats if the historical browsing history of the client is not retrieved;
  • the first sending unit 14 is configured to display to the customer the insurance profile information corresponding to the information format ranked first in the information format ranking.
  • the robot invokes the customer's customer information when the customer logs in to the personal insurance-related account
  • the acquisition information unit 11 automatically obtains the customer's previous browsing history, and searches for the information format of the customer preference in the historical browsing history.
  • the first calling unit 12 sends the insurance profile information in the information format preferred by the customer to the customer. If it is a new customer and has not previously viewed the browsing history, the second retrieval unit 13 accesses the information pattern module 5 and the first determination module 6 to obtain the number of other users' favorite information formats to generate a leaderboard, and the highest ranked on the leaderboard
  • the information format is an information format favored by the most users, and then the first sending unit 14 displays the insurance profile information of the format information to the customer.
  • the above device for pushing an insurance product further includes:
  • a second determining module 9 configured to determine that a sending time period at which the sending time is located is a sending time period preferred by the client;
  • the sending module 10 is configured to, when receiving an instruction to display information to the customer, perform information display in a sending time period preferred by the customer.
  • the sending time module 8 when the insurance profile information is sent to the client, the sending time module 8 also records the sending time of sending the insurance profile information.
  • the customer receives a positive response to this insurance profile information, it indicates that the insurance profile information is approved. It indicates that the customer likes or approves the content of the insurance profile information, and it also indicates that the customer is convenient to receive and read the message during this time period.
  • the second determining module 9 determines the sending time period as the sending time period preferred by the customer. Upload the sending time period to the server, which is convenient for the server to count the time period when the customer likes to receive push insurance profile information.
  • the sending module 10 selects a sending time that the customer likes to push.
  • the server first divides the time of day into multiple time periods, for example, divides the time of day into the early morning (00: 00-06; 00) and the morning (06: 00-09: 00 ), Morning (09: 00-12: 00), noon (12: 00-14: 00), afternoon (14: 00-18: 00), evening (18: 00-21: 00), late night (21: 00-24: 00) these seven time periods. If the time for sending insurance profile information or other insurance product information to customers is 10:30 and between 9 and 12, the corresponding time period is morning, and the server determines that the customer's preferred sending time period is morning (09:00 -12: 00).
  • the device for pushing insurance products of this application first sends insurance information to the user, and then analyzes the user's emotional characteristics of the insurance information according to the user's reply text to determine whether the user needs the insurance product corresponding to the insurance information. Then purposefully push the insurance products that users need.
  • the insurance information is pushed, the user's preferred push time period and information format are obtained, and then the insurance information is pushed in the corresponding information format during the user's push time period, so that the user can understand the insurance information well.
  • an embodiment of the present application further provides a computer device.
  • the computer device may be a server, and its internal structure may be as shown in FIG.
  • the computer device includes a processor, a memory, a network interface, and a database connected through a system bus.
  • the computer design processor is used to provide computing and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system, computer-readable instructions, and a database.
  • the memory provides an environment for operating systems and computer-readable instructions in a non-volatile storage medium.
  • the database of the computer equipment is used to store data such as models of pushed insurance products.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the processes of the embodiments of the methods described above are executed.
  • FIG. 12 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied.
  • An embodiment of the present application further provides a computer non-volatile readable storage medium having computer readable instructions stored thereon.
  • the processes of the embodiments of the methods described above are executed.
  • the above is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the description of the application and the drawings, or directly or indirectly used in other related The technical fields are equally included in the patent protection scope of this application.

Landscapes

  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Development Economics (AREA)
  • Theoretical Computer Science (AREA)
  • Technology Law (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)

Abstract

一种推送保险产品的方法、装置、计算机设备和存储介质，其中方法为：向客户展示指定保险简介信息（S1）；获取所述客户针对所述指定保险简介信息的回复文本（S2）；将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感，所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到（S3）；若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息（S4）。该方法根据用户的回复文本分析用户对保险简介信息的情感特征来判断是否需要该保险信息对应的保险产品。

Description

推送保险产品的方法、装置、计算机设备和存储介质
本申请要求于2018年6月7日提交中国专利局、申请号为2018105819563，发明名称为“推送保险产品的方法、装置、计算机设备和存储介质”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
本申请涉及到计算机技术领域，特别是涉及到一种推送保险产品的方法、装置、计算机设备和存储介质。
 背景技术
现在市面上保险产品的种类很多，越来越满足众多用户的多样化的需求。但是用户的保险意识还不是特别的多，对保险产品的了解不是特别深入。在选择保险产品时，会显得比较茫然。服务器也会通过手机上的APP向用户推送一些保险产品信息，但是这样也仅仅是单方向的向用户推送保险产品信息，无法获取用户对推送的保险产品信息的反馈。
 技术问题
本申请的主要目的为提供一种根据客户对保险信息的反馈进行有目的的推送保险产品的方法、装置、计算机设备和存储介质。
 技术解决方案
为了实现上述发明目的，本申请提出一种推送保险产品的方法，包括：
向客户展示指定保险简介信息；
获取所述客户针对所述指定保险简介信息的回复文本；
将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感；所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到的；
若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息。
本申请还提供一种计算机设备，包括存储器和处理器，所述存储器存储有计算机可读指令，所述处理器执行所述计算机可读指令时实现上述任一项所述方法的步骤。
本申请还提供一种计算机非易失性可读存储介质，其上存储有计算机可读指令，所述计算机可读指令被处理器执行时实现上述任一项所述的方法的步骤。
 有益效果
本申请的推送保险产品的方法、装置、计算机设备和存储介质，首先发送保险信息给用户，然后根据用户的回复文本分析用户对该保险信息的情感特征来判断用户是否需要对该保险信息对应的保险产品，进而有目的的推送用户需要的保险产品。在推送保险信息时，获取到用户偏好的推送时间段和信息格式，进而在该用户的推送时间段内以对应的信息格式来推送保险信息，使用户能很好的了解保险信息。
 附图说明
图1为本申请一实施例的推送保险产品的方法的流程示意图；
图2为本申请一实施例的推送保险产品的方法中流程示意图；
图3为本申请一实施例的上述步骤S1中的具体流程示意图；
图4为本申请一实施例的推送保险产品的方法的流程示意图；
图5为本申请一实施例的推送保险产品的方法中的步骤S1的具体流程示意图；
图6为本申请一实施例的推送保险产品的方法中的步骤S4的具体流程示意图；
图7为本申请一实施例的推送保险产品的方法中的步骤S4的具体流程示意图；
图8为本申请一实施例的推送保险产品的装置的结构示意框图；
图9为本申请一实施例的推送保险产品的装置的结构示意框图；
图10为本申请一实施例的推送保险产品的装置的发送模块的结构示意框图；
图11为本申请一实施例的推送保险产品的装置的结构示意框图；
图12为本申请一实施例的计算机设备的结构示意框图。
 本发明的最佳实施方式
参照图1，本申请实施例提供一种推送保险产品的方法，包括步骤：
S1、向客户展示指定保险简介信息；
S2、获取所述客户针对所述指定保险简介信息的回复文本；
S3、将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感；所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到的；
S4、若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息。
如上述步骤S1所述，保险主要是分为财产保险与人身保险，保险简介信息是包含有保险属性的信息，比如上述的涉及到财产（汽车、房产）相关的信息和涉及到身体（健康状况、疾病）相关的信息是指保险简介信息。向客户展示指定的保险简介信息时，只展示一种保险简介信息。比如只展示财产保险相关的保险简介信息，而不发送人身保险相关的保险简介信息，便于客户对其中一种保险简介信息进行一个直观的反馈，这样系统就可以获取一种保险简介信息的直观反馈。若一次同时展示多个保险简介信息，若客户反馈的是正面情感，则无法确定是对具体哪个保险简介信息是正面情感，无法准确获取到客户直面正面情感的保险类型。保险简介信息是以多种格式存在，包括图片格式、文本格式、视频格式、音频格式等的一种或多种结合。
如上述步骤S2所述，向客户发送一种保险信息后，客户阅读后，对该信息以文字的形式做出一些反馈，通过键盘或者是触摸屏回复文本信息，也可以是机器人获取到客户说话的声音后将其语音识别后转换成文本信息，机器人获取到该回复文本。
如上述步骤S3所述，机器人将回复文本输入到训练好的情感分析模型中，提取出客户的回复文本中包含的客户的情感特征。情感分析模型是一个基于LSTM-CNN模型经过大量训练后得到的模型，用于将输入的回复文本经一系列计算后得出一个结果，该结果是表示客户的情感特征是正面情感还是负面情感。在训练该情感分析模型时，工作人员先将大量的文本定义不同的业务情绪数据，然后将大量的文本以及对应的情感特征分别输入到LSTM-CNN模型中，情感特征包括正面情感和负面情感，正面情感表示感兴趣，负面情感是表示不感兴趣。使得该LSTM-CNN模型形成了一个根据文本来进行情绪特征的计算模型，即情感分析模型。之后机器人将客户的回复文本输入到该情感分析模型中后，即可以输出该客户的回复文本对应的情感特征。
如上述步骤S4所述，如果是正面情感，表示客户是对该保险信息感兴趣的，则输出与该保险简介信息相关的保险产品信息。保险产品信息是预先存储在机器人内部的存储器内，而且保险产品信息都包含了相关的保险信息的标签。输出的方式可以是将保险产品的图片信息和/或文字信息显示在显示屏上，也可以是将保险产品的文字信息以音频的形式通过扬声器播放出来。
参照图2，进一步地，上述若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息的步骤之后，包括：
S5、获取所述保险简介信息的信息格式；
S6、判定该信息格式为所述客户偏好的信息格式；
S7、当接收到向所述客户展示信息的指令时，调用所述客户偏好的信息格式进行信息展示。
在本实施例中，当上述客户对保险简介信息是正面情感，则表示对保险简介信息认可。说明客户对保险简介信息的内容表示喜欢或认可，也说明客户对保险简介信息的信息格式是易于接受与认可的。因此，要将步骤S1发送的保险简介信息的信息格式判定为客户偏好的信息格式，便于之后再次向客户推送保险产品信息时，以客户偏好的信息格式来推送。上述向客户发送指定的保险简介信息，发送包含有保险信息的图片或文本或视频或音频格式的一种或多种。当接收到客户对此保险简介信息回复的是正面情感，获取上述保险简介信息的信息格式，通过获取保险简介信息的对应的文件名称的后缀来判断保险信息的信息格式。如果保险简介信息的文件名称的后缀是doc、txt中的一种，说明保险简介信息的信息格式包含了文本格式；如果保险简介信息的文件名称的后缀是Mp4、rmvb、mkv、avi中的一种，说明保险简介信息的信息格式包含了视频格式；如果保险简介信息的文件名称的后缀是Mp3、wma、amr中的一种，说明保险简介信息的信息格式包含了音频格式；如果保险简介信息的文件名称的后缀是jpeg、bmp、gif中的一种，说明保险简介信息的信息格式包含了图片格式。将上述信息格式上传到服务器，便于服务器统计客户喜欢的保险简介信息的信息格式，在之后的推送保险产品信息或者其他需要展示信息的时候，选择客户喜欢的信息格式进行推送或展示。
参照图3，进一步地，上述向客户展示指定保险简介信息的步骤包括：
S11、获取所述客户的客户信息，并根据所述客户信息调取所述客户的历史浏览记录，所述历史浏览记录中包括浏览信息的信息格式；
S12、若调取到所述客户的历史浏览记录，则确定所述历史浏览记录中浏览次数最多的信息格式为所述客户的偏好信息格式，并向所述客户展示偏好的信息格式对应的保险简介信息；
S13、若未调取到所述客户的历史浏览记录，获取所有客户偏好信息格式的排行榜；
S14、向所述客户展示所述信息格式排行榜上第一名的信息格式对应的保险简介信息。
在这一实施例中，机器人在客户登录个人的保险相关的账户时就调用了客户的客户信息，自动获取了客户的以往的浏览记录，查找历史浏览记录中的客户偏好的信息格式，发送客户偏好的信息格式的保险简介信息给客户。如果是新客户，之前没有过浏览记录，就访问服务器上通过上述S5、S6步骤预设的方法根据所有老用户喜欢的信息格式的数量生成排行榜，排行榜上排名最高的信息格式是受最多用户喜欢的信息格式，然后将这个格式信息的保险简介信息展示给客户。
参照图4，进一步地，上述若所述情感特征为正面情感，则输出所述保险简介信息对应的保险产品信息的步骤之后，包括：
S8、获取发送所述保险简介信息的发送时刻；
S9、判定所述发送时刻所在的发送时间段为所述客户偏好的发送时间段；
S10、当接收到向所述客户展示信息的指令时，在所述客户偏好的发送时间段进行信息展示。
本实施例中，上述向客户发送保险简介信息时，机器人还记录了发送保险简介信息的发送时刻。当接收到客户对此保险简介信息回复的是正面情感，则表示对保险简介信息认可。说明客户对保险简介信息的内容表示喜欢或认可，也说明客户在这个时间段内是方便接收阅读消息的。获取发送上述保险简介信息的发送时刻，然后判断该发送时刻所对应的发送时间段，将该发送时间段判定为客户偏好的发送时间段。将该发送时间段上传到服务器，便于服务器统计客户喜欢接收推送保险简介信息的时间段。在之后的推送保险简介信息或其他保险产品信息的时候，选择客户喜欢的发送时刻进行推送。服务器在判定发送时刻对应的发送时间段时，先将一天的时间分成多个时间段，例如将一天的时间分为凌晨(00:00-06;00)、早晨（06:00-09:00）、上午(09:00-12:00)、中午(12:00-14:00)、下午(14:00-18:00)、晚上(18:00-21:00)、深夜(21:00-24:00)这七个时间段。若向客户发送保险简介信息或其他保险产品信息的时刻是10点半，在9点与12点之间，则对应的时间段是上午，服务器判定客户偏好的发送时间段是上午(09:00-12:00)。
参照图5，进一步地，上述向客户展示指定保险简介信息的步骤包括：
S15、获取所述客户的客户信息，判断所述客户信息中是否存在所述客户偏好的发送时间段；
S16、若存在，判断当前时刻是否是属于所述客户偏好的发送时间段；
S17、若是，向客户展示指定保险简介信息。
在这一实施例中，机器人在客户登录个人的保险相关的账户时就调用了客户的客户信息，自动获取了用户的个人信息，对应的获取到服务器计算得到的该客户偏好的发送时间段。然后判断当前的时候是不是在该客户的发送时间段内，如果是，说明是适合发送保险信息的时刻，就向客户发送指定的保险信息。
参照图6，进一步地，上述若所述情感特征为正面情感，则输出所述保险简介信息对应的保险产品信息的步骤，包括：
S41、若所述情感特征为正面情感，获取所述保险简介信息对应的保险产品类型；
S42、输出所述保险产品类型对应的保险产品信息。
本实施例中，如果情感特征为正面情感，则获取发送的上述保险简介信息，判断上述保险简介信息是属于哪个类别的保险类型的保险简介信息，如果是属于财产保险，则获取财产保险的保险类型，进一步地，将该保险类型下的保险产品信息都显示出来，供客户浏览选择。
参照图7，进一步地，一个所述指定的保险简介信息对应多个保险类型，上述输出所述保险简介信息对应的保险产品信息的步骤，包括：
S421、将多个所述保险类型显示在显示屏上；
S422、接收用户选择的保险类型；
S423、将所述用户选择的保险类型对应的保险产品信息显示在显示屏上。
本实施例中，下表是一具体实施例的保险信息、保险类型、保险产品三者的关系。
Figure dest_path_image001
一个保险信息对应有多个保险类型，而每个保险类型也对应有多个保险产品，如果将这些全部显示在显示屏上，信息量较大，客户无法迅速准确获取到所需要的产品，因此，做进一步筛选，将保险信息对应的保险类型显示在显示屏上，客户对其进行进一步优化选择方向，机器人接收到客户选择的保险类型后，最终只在屏幕上显示这个保险类型的保险产品。方便客户准确的选择到心仪的保险产品。
进一步地，上述若所述情感特征为正面情感，则输出与所述保险信息相关的保险产品的步骤，包括：
S43、若所述情感特征为正面情感，查找所述回复文本中的关键词；
S44、显示与所述关键词相关的保险产品。
本实施例中，如果情感特征为正面情感，那么再次调用客户的回复文本，读取客户文本中的关键词，对应的，每个保险产品也都有至少一个用于表示产品的特色的产品关键词，比如人寿、汽车、房屋、煤气等能体现保险产品的所承保的标的物的词语，也包括意外、医疗、理财、分红、投资等体现保险产品主要功能的词语。如果客户的回复文本中含有上述关键词，则机器人会推荐含有上述关键词的保险产品给用户，将该保险产品显示在屏幕上，方便客户选择。
进一步地，上述将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征的步骤之后，包括：
S301、获取客户的脸部表情；
S302、判断所述脸部表情是否是积极表情；
S302、若是，判定情感特征为正面情感。
本实施例中，获取客户的脸部表情，是机器人调用自身的摄像头来进行拍摄。一般客户在与机器人互动时，都是站着或坐在指定的位置，机器人控制摄像头开启后，寻找脸部，然后进行拍摄。一般而言，机器人最近的脸部信息是客户的脸部信息。根据获取的脸部信息，判断出该脸部信息的脸部表情。判断脸部信息上的脸部表情，是将包含有脸部信息的图片输入一个基于RNN神经网络训练好的模型后输入对应的脸部表情，输出的脸部表情是积极或消极，如果是积极表情，积极表情包括微笑、关注等表情，则进一步确认客户是对上述保险信息是感兴趣的。上述判断脸部表情是否是积极表情的具体方法是：机器人对拍摄的含有脸部信息的图像进行特征提取，得到对应的人脸特征向量，然后对人脸特征向量进行表情识别。若识别的结果是属于预设的积极表情，则表示用户是正面情感。
进一步地，上述若所述情感特征为正面情感，则输出所述保险信息相关的保险产品的步骤，包括：
S45、若所述情感特征为正面情感，获取客户的脸部信息；
S46、根据所述脸部信息，判断客户的年龄及性别；
S47、根据所述年龄以及性别，显示所述与保险信息相关的保险产品。
本实施例中，获取客户的脸部信息，是机器人调用自身的摄像头来进行拍摄。一般客户在与机器人互动时，都是站着或坐在指定的位置，机器人控制摄像头开启后，寻找脸部，然后进行拍摄。根据获取的脸部信息，判断出该脸部信息对应的客户的年龄。判断脸部信息上对应的客户的年龄和性别，是将包含有脸部信息的图片输入一个基于RNN神经网络训练好的模型后输入对应的客户的年龄和性别。对应的，每个保险产品也都预设有适合的年龄与性别，符合该年龄要求与性别要求的人才能购买对应的保险产品。比如一些疾病保险要求的年龄范围比较小，而分红保险要求的年龄范围比较大。因此，在显示保险产品时，根据客户的年龄以及性别，将客户感兴趣的保险产品而且是能购买的保险产品显示出来，过滤掉无法购买的保险产品，更进一步的给客户提供准确的购买保险产品。
综上所述，本申请的推送保险产品的方法，首先发送保险信息给用户，然后根据用户的回复文本分析用户对该保险信息的情感特征来判断用户是否需要对该保险信息对应的保险产品，进而有目的的推送用户需要的保险产品。在推送保险信息时，获取到用户偏好的推送时间段和信息格式，进而在该用户的推送时间段内以对应的信息格式来推送保险信息，使用户能很好的了解保险信息。
参照图8，本申请实施例中还提供一种推送保险产品的装置，包括：
展示模块1，用于向客户展示指定的保险简介信息；
获取模块2，用于获取所述客户针对所述指定保险简介信息的回复文本；
情感特征模块3，用于将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感；所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到的；
输出模块4，用于若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息。
本实施例中，保险主要是分为财产保险与人身保险，保险简介信息是包含有保险属性的信息，比如上述的涉及到财产（汽车、房产）相关的信息和涉及到身体（健康状况、疾病）相关的信息是指保险简介信息。展示模块1向客户展示指定的保险简介信息时，只展示一种保险简介信息。比如只展示财产保险相关的保险简介信息，而不发送人身保险相关的保险简介信息，便于客户对其中一种保险简介信息进行一个直观的反馈，这样系统就可以获取一种保险简介信息的直观反馈。若一次同时展示多个保险简介信息，若客户反馈的是正面情感，则无法确定是对具体哪个保险简介信息是正面情感，无法准确获取到客户直面正面情感的保险类型。保险简介信息是以多种格式存在，包括图片格式、文本格式、视频格式、音频格式等的一种或多种结合。展示模块1向客户发送一种保险简介信息后，客户阅读后，对该信息以文字的形式做出一些反馈，通过键盘或者是触摸屏回复文本信息，获取模块2获取到客户的回复文本。当客户是通过声音进行回复，获取模块2获取到客户说话的声音后将其语音识别后转换成文本信息，进而获取到回复文本。情感特征模块3将回复文本输入到训练好的情感分析模型中，提取出客户的回复文本中包含的客户的情感特征。情感分析模型是一个基于LSTM-CNN模型经过大量训练后得到的模型，用于将输入的回复文本经一系列计算后得出一个结果，该结果是表示客户的情感特征是正面情感还是负面情感。在训练该情感分析模型时，工作人员先将大量的文本定义不同的业务情绪数据，然后将大量的文本以及对应的情感特征分别输入到LSTM-CNN模型中，情感特征包括正面情感和负面情感，正面情感表示感兴趣，负面情感是表示不感兴趣。使得该LSTM-CNN模型形成了一个根据文本来进行情绪特征的计算模型，得到情感分析模型。之后情感特征模块3将客户的回复文本输入到该情感分析模型中后，即可以输出该客户的回复文本对应的情感特征。如果是正面情感，表示客户是对该保险简介信息感兴趣的，则输出模块4输出与该保险简介信息相关的保险产品信息。保险产品信息是预先存储在机器人内部的存储器内，而且保险产品信息都包含了相关的保险信息的标签。输出模块4输出的方式可以是将保险产品的图片信息和/或文字信息显示在显示屏上，也可以是将保险产品的文字信息以音频的形式通过扬声器播放出来。
参照图9，进一步地，上述推送保险产品的装置还包括：
信息格式模块5，用于获取所述保险简介信息的信息格式；
第一判定模块6，用于判定该信息格式为所述客户偏好的信息格式；
第一调用模块7，用于当接收到向所述客户展示信息的指令时，调用所述客户偏好的信息格式进行信息展示。
本实施例中，当上述客户对保险信息是正面情感，则表示对保险简介信息认可。说明客户对保险简介信息的内容表示喜欢或认可，也说明客户对保险简介信息的信息格式是易于接受与认可的。因此，第一判定模块6要将展示模块1展示的保险简介信息的信息格式判定为客户偏好的信息格式，便于之后再次向客户推送保险简介信息时，以客户偏好的信息格式来推送。上述向客户发送指定的保险简介信息，发送包含有保险简介信息的图片或文本或视频或音频格式的一种或多种。当接收到客户对此保险简介信息回复的是正面情感，信息模式模块5获取上述保险简介信息的信息格式，通过获取保险简介信息的对应的文件名称的后缀来判断保险简介信息的信息格式。如果保险简介信息的文件名称的后缀是doc、txt中的一种，说明保险简介信息的信息格式包含了文本格式；如果保险简介信息的文件名称的后缀是Mp4、rmvb、mkv、avi中的一种，信息模式模块5判定保险简介信息的信息格式包含了视频格式；如果保险简介信息的文件名称的后缀是Mp3、wma、amr中的一种，信息模式模块5判定保险简介信息的信息格式包含了音频格式；如果保险简介信息的文件名称的后缀是jpeg、bmp、gif中的一种，信息模式模块5判定保险简介信息的信息格式包含了图片格式。将上述信息格式上传到服务器，便于服务器统计客户喜欢的保险信息的信息格式，在之后的推送保险产品信息或者其他需要展示信息的时候，第一调用模块7选择客户喜欢的信息格式进行推送或展示。
参照图10，进一步地，上述展示模块1包括：
获取信息单元11，用于获取所述客户的客户信息，并根据所述客户信息调用所述客户的历史浏览记录，所述历史浏览记录中包含浏览信息的信息格式；
第一调取单元12，用于若调取到所述客户的历史浏览记录，则确定所述历史浏览记录中浏览次数最多的信息格式为所述客户的偏好信息格式，并向所述客户展示偏好的信息格式对应的保险简介信息；
第二调取单元13，用于若未调取到所述客户的历史浏览记录，获取所有客户偏好信息格式的排行榜；
第一发送单元14，用于向所述客户展示所述信息格式排行榜上第一名的信息格式对应的保险简介信息。
本实施例中，机器人在客户登录个人的保险相关的账户时就调用了客户的客户信息，获取信息单元11自动获取了客户的以往的浏览记录，查找历史浏览记录中的客户偏好的信息格式，第一调取单元12发送客户偏好的信息格式的保险简介信息给客户。如果是新客户，之前没有过浏览记录，第二调取单元13就访问信息模式模块5与第一判定模块6获取的其他老用户喜欢的信息格式的数量生成排行榜，排行榜上排名最高的信息格式是受最多用户喜欢的信息格式，然后第一发送单元14将这个格式信息的保险简介信息展示给客户。
参照图11，进一步地，上述推送保险产品的装置还包括：
发送时刻模块8，用于获取发送所述保险简介信息的发送时刻；
第二判定模块9，用于判定所述发送时刻所在的发送时间段为所述客户偏好的发送时间段；
发送模块10，用于当接收到向所述客户展示信息的指令时，在所述客户偏好的发送时间段进行信息展示。
本实施例中，上述向客户发送保险简介信息时，发送时刻模块8还记录了发送保险简介信息的发送时刻。当接收到客户对此保险简介信息回复的是正面情感，则表示对保险简介信息认可。说明客户对保险简介信息的内容表示喜欢或认可，也说明客户在这个时间段内是方便接收阅读消息的。获取发送上述保险简介信息的发送时刻，然后判断该发送时刻所对应的发送时间段，第二判定模块9将该发送时间段判定为客户偏好的发送时间段。将该发送时间段上传到服务器，便于服务器统计客户喜欢接收推送保险简介信息的时间段。发送模块10在之后的发送保险简介信息或其他保险产品信息的时候，选择客户喜欢的发送时刻进行推送。服务器在判定发送时刻对应的发送时间段时，先将一天的时间分成多个时间段，例如将一天的时间分为凌晨(00:00-06;00)、早晨（06:00-09:00）、上午(09:00-12:00)、中午(12:00-14:00)、下午(14:00-18:00)、晚上(18:00-21:00)、深夜(21:00-24:00)这七个时间段。若向客户发送保险简介信息或其他保险产品信息的时刻是10点半，在9点与12点之间，则对应的时间段是上午，服务器判定客户偏好的发送时间段是上午(09:00-12:00)。
综上所述，本申请的推送保险产品的装置，首先发送保险信息给用户，然后根据用户的回复文本分析用户对该保险信息的情感特征来判断用户是否需要对该保险信息对应的保险产品，进而有目的的推送用户需要的保险产品。在推送保险信息时，获取到用户偏好的推送时间段和信息格式，进而在该用户的推送时间段内以对应的信息格式来推送保险信息，使用户能很好的了解保险信息。
参照图12，本申请实施例中还提供一种计算机设备，该计算机设备可以是服务器，其内部结构可以如图12所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中，该计算机设计的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机可读指令和数据库。该内存器为非易失性存储介质中的操作系统和计算机可读指令的运行提供环境。该计算机设备的数据库用于存储推送保险产品的模型等数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机可读指令在执行时，执行如上述各方法的实施例的流程。本领域技术人员可以理解，图12中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定。
本申请一实施例还提供一种计算机非易失性可读存储介质，其上存储有计算机可读指令，该计算机可读指令在执行时，执行如上述各方法的实施例的流程。以上所述仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种推送保险产品的方法，其特征在于，包括：
    向客户展示指定保险简介信息；
    获取所述客户针对所述指定保险简介信息的回复文本；
    将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感；所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到的；
    若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息。
  2. 如权利要求1所述的推送保险产品的方法，其特征在于，所述输出所述指定保险简介信息对应的保险产品信息的步骤之后，包括：
    获取所述保险简介信息的信息格式；
    判定该信息格式为所述客户偏好的信息格式；
    当接收到向所述客户展示信息的指令时，调用所述客户偏好的信息格式进行信息展示。
  3. 如权利要求1所述的推送保险产品的方法，其特征在于，所述向客户展示指定的保险简介信息的步骤，包括：
    获取所述客户的客户信息，并根据所述客户信息调取所述客户的历史浏览记录，所述历史浏览记录中包括浏览信息的信息格式；
    若调取到所述客户的历史浏览记录，则确定所述历史浏览记录中浏览次数最多的信息格式为所述客户的偏好信息格式，并向所述客户展示所述偏好的信息格式对应的保险简介信息；
    若未调取到所述客户的历史浏览记录，获取所有客户偏好信息格式的排行榜；
    向所述客户发送所述信息格式排行榜上第一名的信息格式对应的保险简介信息。
  4. 如权利要求1所述的推送保险产品的方法，其特征在于，所述输出所述指定保险简介信息对应的保险产品信息的步骤之后，包括：
    获取发送所述保险简介信息的发送时刻；
    判定所述发送时刻所在的发送时间段为所述客户偏好的发送时间段；
    当接收到向所述客户展示信息的指令时，在所述客户偏好的发送时间段进行信息展示。
  5. 如权利要求1所述的推送保险产品的方法，其特征在于，所述向客户展示指定保险简介信息的步骤，包括：
    获取所述客户的客户信息，判断所述客户信息中是否存在所述客户偏好的发送时间段；
    若存在，判断当前时刻是否是属于所述客户偏好的发送时间段；
    若是，向客户展示指定保险简介信息。
  6. 如权利要求1所述的推送保险产品的方法，其特征在于，所述若所述情感特征为正面情感，则输出所述保险简介信息对应的保险产品信息的步骤，包括：
    若所述情感特征为正面情感，获取所述保险简介信息对应的保险产品类型；
    输出所述保险产品类型对应的保险产品信息。
  7. 一种推送保险产品的装置，其特征在于，包括：
    展示模块，用于向客户展示指定保险简介信息；
    获取模块，用于获取所述客户针对所述指定保险简介信息的回复文本；
    情感特征模块，用于将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感；所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到的；
    输出模块，用于若所述情感特征为正面情感，则输出所述保险简介信息对应的保险产品信息。
  8. 如权利要求7所述的保险产品的装置，其特征在于，还包括：
    信息格式模块，用于获取所述保险简介信息的信息格式；
    第一判定模块，用于判定该信息格式为所述客户偏好的信息格式；
    第一调用模块，用于当接收到向所述客户展示信息的指令时，调用所述客户偏好的信息格式进行信息展示。
  9. 如权利要求7所述的保险产品的装置，其特征在于，所述展示模块包括：
    获取信息单元，用于获取所述客户的客户信息，并根据所述客户信息调用所述客户的历史浏览记录，所述历史浏览记录中包含浏览信息的信息格式；
    第一调取单元，用于若调取到所述客户的历史浏览记录，则确定所述历史浏览记录中浏览次数最多的信息格式为所述客户的偏好信息格式，并向所述客户展示偏好的信息格式对应的保险简介信息；
    第二调取单元，用于若未调取到所述客户的历史浏览记录，获取所有客户偏好信息格式的排行榜；
    第一发送单元，用于向所述客户展示所述信息格式排行榜上第一名的信息格式对应的保险简介信息。
  10. 如权利要求7所述的推送保险产品的装置，其特征在于，还包括：
    发送时刻模块，用于获取发送所述保险简介信息的发送时刻；
    第二判定模块，用于判定所述发送时刻所在的发送时间段为所述客户偏好的发送时间段；
    发送模块，用于当接收到向所述客户展示信息的指令时，在所述客户偏好的发送时间段进行信息展示。
  11. 如权利要求7所述的推送保险产品的装置，其特征在于，所述展示模块包括：
    第二判断单元，用于获取所述客户的客户信息，判断所述客户信息中是否存在所述客户偏好的发送时间段；
    第三判断单元，用于若所述客户有偏好的发送时间段，判断当前时刻是否是属于所述客户偏好的发送时间段；
    第二发送单元，用于若当前时刻是属于客户偏好的发送时间段，向客户展示指定保险简介信息。
  12. 如权利要求7所述的推送保险产品的装置，其特征在于，还包括：
    发送时刻模块，用于获取发送所述保险简介信息的发送时刻；
    第二判定模块，用于判定所述发送时刻所在的发送时间段为所述客户偏好的发送时间段；
    发送模块，用于当接收到向所述客户展示信息的指令时，在所述客户偏好的发送时间段进行信息展示。
  13. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机可读指令，其特征在于，所述处理器执行所述计算机可读指令时实现推送保险产品的方法，该推送保险产品的方法，包括：
    向客户展示指定保险简介信息；
    获取所述客户针对所述指定保险简介信息的回复文本；
    将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感；所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到的；
    若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息。
  14. 如权利要求13所述的计算机设备，其特征在于，所述输出所述指定保险简介信息对应的保险产品信息的步骤之后，包括：
    获取所述保险简介信息的信息格式；
    判定该信息格式为所述客户偏好的信息格式；
    当接收到向所述客户展示信息的指令时，调用所述客户偏好的信息格式进行信息展示。
  15. 如权利要求13所述的计算机设备，其特征在于，所述向客户展示指定的保险简介信息的步骤，包括：
    获取所述客户的客户信息，并根据所述客户信息调取所述客户的历史浏览记录，所述历史浏览记录中包括浏览信息的信息格式；
    若调取到所述客户的历史浏览记录，则确定所述历史浏览记录中浏览次数最多的信息格式为所述客户的偏好信息格式，并向所述客户展示所述偏好的信息格式对应的保险简介信息；
    若未调取到所述客户的历史浏览记录，获取所有客户偏好信息格式的排行榜；
    向所述客户发送所述信息格式排行榜上第一名的信息格式对应的保险简介信息。
  16. 如权利要求13所述的计算机设备，其特征在于，所述输出所述指定保险简介信息对应的保险产品信息的步骤之后，包括：
    获取发送所述保险简介信息的发送时刻；判定所述发送时刻所在的发送时间段为所述客户偏好的发送时间段；当接收到向所述客户展示信息的指令时，在所述客户偏好的发送时间段进行信息展示。
  17. 一种计算机非易失性可读存储介质，其上存储有计算机可读指令，其特征在于，所述计算机可读指令被处理器执行时实现推送保险产品的方法，该推送保险产品的方法，包括：
    向客户展示指定保险简介信息；获取所述客户针对所述指定保险简介信息的回复文本；将所述回复文本输入到预设的基于LSTM-CNN模型训练得到的情感分析模型中进行计算，以得到所述客户针对所述保险信息的情感特征，所述情感特征包括正面情感和负面情感；所述情感分析模型为根据多个已知情感特征的回复文本的训练样本训练得到的；
    若所述情感特征为正面情感，则输出所述指定保险简介信息对应的保险产品信息。
  18. 如权利要求17所述的计算机非易失性可读存储介质，其特征在于，所述输出所述指定保险简介信息对应的保险产品信息的步骤之后，包括：
    获取所述保险简介信息的信息格式；判定该信息格式为所述客户偏好的信息格式；当接收到向所述客户展示信息的指令时，调用所述客户偏好的信息格式进行信息展示。
  19. 如权利要求17所述的计算机非易失性可读存储介质，其特征在于，所述向客户展示指定的保险简介信息的步骤，包括：
    获取所述客户的客户信息，并根据所述客户信息调取所述客户的历史浏览记录，所述历史浏览记录中包括浏览信息的信息格式；若调取到所述客户的历史浏览记录，则确定所述历史浏览记录中浏览次数最多的信息格式为所述客户的偏好信息格式，并向所述客户展示所述偏好的信息格式对应的保险简介信息；若未调取到所述客户的历史浏览记录，获取所有客户偏好信息格式的排行榜；
    向所述客户发送所述信息格式排行榜上第一名的信息格式对应的保险简介信息。
  20. 如权利要求17所述的计算机非易失性可读存储介质，其特征在于，所述输出所述指定保险简介信息对应的保险产品信息的步骤之后，包括：
    获取发送所述保险简介信息的发送时刻；判定所述发送时刻所在的发送时间段为所述客户偏好的发送时间段；当接收到向所述客户展示信息的指令时，在所述客户偏好的发送时间段进行信息展示。
PCT/CN2018/094885 2018-06-07 2018-07-06 推送保险产品的方法、装置、计算机设备和存储介质 WO2019232883A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810581956.3A CN108961072A (zh) 2018-06-07 2018-06-07 推送保险产品的方法、装置、计算机设备和存储介质
CN201810581956.3 2018-06-07

Publications (1)

Publication Number Publication Date
WO2019232883A1 true WO2019232883A1 (zh) 2019-12-12

Family

ID=64494121

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/094885 WO2019232883A1 (zh) 2018-06-07 2018-07-06 推送保险产品的方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN108961072A (zh)
WO (1) WO2019232883A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023121609A1 (en) * 2021-12-23 2023-06-29 Turkiye Garanti Bankasi Anonim Sirketi An insurance claim system

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109754312A (zh) * 2018-12-18 2019-05-14 深圳壹账通智能科技有限公司 产品推送方法、装置、计算机设备和存储介质
CN109767290A (zh) * 2018-12-18 2019-05-17 深圳壹账通智能科技有限公司 产品推送方法、装置、计算机设备和存储介质
CN110335596A (zh) * 2019-06-19 2019-10-15 深圳壹账通智能科技有限公司 基于语音识别的产品推荐方法、装置、设备和存储介质
CN110890088B (zh) * 2019-10-12 2022-07-15 中国平安财产保险股份有限公司 语音信息反馈方法、装置、计算机设备和存储介质
CN112565471B (zh) * 2021-02-26 2021-07-27 深圳追一科技有限公司 信息推送方法、推送机器人、计算机设备和存储介质
CN113010784B (zh) * 2021-03-17 2024-02-06 北京十一贝科技有限公司 用于生成预测信息的方法、装置、电子设备和介质
CN113283934A (zh) * 2021-05-28 2021-08-20 北京量冠科技有限公司 一种信息确定方法、装置、电子设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106599933A (zh) * 2016-12-26 2017-04-26 哈尔滨工业大学 一种基于联合深度学习模型的文本情感分类方法
CN106782602A (zh) * 2016-12-01 2017-05-31 南京邮电大学 基于长短时间记忆网络和卷积神经网络的语音情感识别方法
CN107092596A (zh) * 2017-04-24 2017-08-25 重庆邮电大学 基于attention CNNs和CCR的文本情感分析方法
CN107562784A (zh) * 2017-07-25 2018-01-09 同济大学 基于ResLCNN模型的短文本分类方法
CN107679234A (zh) * 2017-10-24 2018-02-09 上海携程国际旅行社有限公司 客服信息提供方法、装置、电子设备、存储介质

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104462468A (zh) * 2014-12-17 2015-03-25 百度在线网络技术（北京）有限公司 信息提供方法及装置
CN107730389A (zh) * 2017-09-30 2018-02-23 平安科技（深圳）有限公司 电子装置、保险产品推荐方法及计算机可读存储介质
CN107895284B (zh) * 2017-11-08 2020-09-08 平安科技（深圳）有限公司 交流页面管控方法、装置、终端和计算机可读存储介质
CN107918649A (zh) * 2017-11-13 2018-04-17 百度在线网络技术（北京）有限公司 信息推荐和搜索方法、装置、设备及计算机可读介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106782602A (zh) * 2016-12-01 2017-05-31 南京邮电大学 基于长短时间记忆网络和卷积神经网络的语音情感识别方法
CN106599933A (zh) * 2016-12-26 2017-04-26 哈尔滨工业大学 一种基于联合深度学习模型的文本情感分类方法
CN107092596A (zh) * 2017-04-24 2017-08-25 重庆邮电大学 基于attention CNNs和CCR的文本情感分析方法
CN107562784A (zh) * 2017-07-25 2018-01-09 同济大学 基于ResLCNN模型的短文本分类方法
CN107679234A (zh) * 2017-10-24 2018-02-09 上海携程国际旅行社有限公司 客服信息提供方法、装置、电子设备、存储介质

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023121609A1 (en) * 2021-12-23 2023-06-29 Turkiye Garanti Bankasi Anonim Sirketi An insurance claim system

Also Published As

Publication number Publication date
CN108961072A (zh) 2018-12-07

Similar Documents

Publication Publication Date Title
WO2019232883A1 (zh) 推送保险产品的方法、装置、计算机设备和存储介质
US10938832B2 (en) Systems and methods for providing an interactive media presentation
Brown et al. A snapshot of the life as lived: Wearable cameras in social and personality psychological science
US20200073904A1 (en) Lifestream Annotation Method and System
US10394881B2 (en) Automatic suggestions to share images
US20130289991A1 (en) Application of Voice Tags in a Social Media Context
KR101686830B1 (ko) 온라인 소셜 네트워크 상의 이미지를 위한 태그 제안
US10839136B2 (en) Generation of collateral object representations in collaboration environments
US9805163B1 (en) Apparatus and method for improving compliance with a therapeutic regimen
US20200152304A1 (en) Systems And Methods For Intelligent Voice-Based Journaling And Therapies
WO2021114936A1 (zh) 信息推荐方法、装置、电子设备及计算机可读存储介质
US20220019689A1 (en) Privacy Preserving Server-Side Personalized Content Selection
CN112445970A (zh) 一种信息推荐方法、装置、电子设备及存储介质
Mulvenna et al. HILDA-A health interaction log data analysis workflow to aid understanding of usage patterns and behaviours
TW200933384A (en) Social service system for digital photos
JP6752330B1 (ja) 機械学習ベースのマッチング装置およびマッチング方法
KR20220060054A (ko) 스마트 헬스케어 서비스 제공 방법 및 그를 위한 장치
US10554768B2 (en) Contextual user experience
WO2019000051A1 (en) DATA ANALYSIS METHOD AND LEARNING SYSTEM
CN114265989A (zh) 好友推荐方法、电子设备及计算机可读存储介质
CN113901337A (zh) 应用推荐方法、装置、设备和计算机可读存储介质
TWM617933U (zh) 新聞及公眾意見分析系統
JP2020149366A (ja) コミュニケーションシステム、コミュニケーション方法及びプログラム
JP7180921B1 (ja) プログラム、情報処理装置および情報処理方法
US11445273B1 (en) System and method for creating a video summary based on video relevancy

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: 18921302

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: 18921302

Country of ref document: EP

Kind code of ref document: A1