Intelligent outbound system and method based on customer intention

Abstract

The invention discloses an intelligent outbound call system and method based on customer intention. The system comprises a data acquisition module, a data processing module and an outbound call control module, relates to the technical field of intelligent outbound calling, and solves the technical problems of low customer promotion accuracy and low transaction rate of an intelligent outbound calling system in the prior art. The method comprises the steps of obtaining basic information and call information of a plurality of clients; screening the plurality of clients based on the basic information and the call duration to obtain a target client; calculating an intention transaction coefficient of the target customer based on the keyword sequence and the call feature information; and adjusting an outbound strategy for the target customer based on the intention transaction coefficient. According to the invention, the number of the keywords, the call duration and the number of active questioning times are extracted from the call information and used for calculating the intention transaction coefficient of the target customer, so that the calculated intention transaction coefficient can accurately distinguish the intention level of the target customer, and the success rate of marketing activities and customer satisfaction can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent outbound calling, and in particular to an intelligent outbound calling system and method based on customer intention. Background Art

[0002] From the perspective of the extension of application areas, with the development of the industry and the advancement of technology, call centers have gradually become a key point in the marketing process of enterprises. Many non-service call centers are also constantly emerging, which has greatly extended the application areas of call centers. For example, some companies have expanded the application of call centers, which are widely used in marketing, e-commerce, market research, consulting and other aspects.

[0003] The intelligent outbound call system of the prior art has achieved the efficiency and scale of customer contact to a certain extent through automatic dialing and initial dialogue interaction; however, the distribution of keywords in the outbound call system's sales talk to customers has a certain impact on promoting customer transactions. For example, too dense keywords will make it difficult for customers to fully understand in a short period of time, which will reduce the customer's transaction intention. The prior art is difficult to use the distribution of keywords in the outbound call talk to evaluate the degree of influence on customer intention, resulting in low accuracy and transaction rate of sales promotion by the intelligent outbound call system to customers; in addition, the number of active questions asked by customers can reflect the customer's intention to a certain extent. The prior art is difficult to use the number of active questions asked to measure the customer's transaction intention, resulting in low accuracy in the outbound call system's identification of potential customers.

[0004] The present invention provides an intelligent outbound calling system and method based on customer intention to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an intelligent outbound calling system and method based on customer intention, which is used to solve the technical problem that the intelligent outbound calling system in the prior art is difficult to accurately identify potential intended customers, resulting in low accuracy of sales promotion to customers and low transaction rate.

[0006] To achieve the above-mentioned object, the first aspect of the present invention provides an intelligent outbound call system based on customer intention, comprising: a data processing module, and a data acquisition module and an outbound call control module connected thereto;

[0007] The data collection module is used to obtain basic information and call information of several customers; screen several customers based on the basic information and call duration to obtain target customers; wherein the basic information includes age and gender; the call information includes call duration, call content and call time;

[0008] The data processing module: performs keyword recognition on the target customer's call content based on speech recognition technology to obtain a keyword sequence; obtains the number of active questions based on the target customer's call content, and integrates the call duration and the number of active questions into call feature information; and

[0009] Calculate the target customer's intended transaction coefficient based on keyword sequence and call feature information;

[0010] The outbound call control module adjusts the outbound call strategy for target customers based on the intended transaction coefficient.

[0011] The present invention obtains the basic information and call information of the customer through the data acquisition module, and combines the voice recognition technology and keyword recognition of the call content used in the data processing module, which can not only accurately screen out target customers with potential high value, but also can further accurately evaluate the target customer's intended transaction coefficient by analyzing call feature information such as call duration and the number of active questions. This process effectively improves the accuracy and efficiency of customer screening and reduces the cost of ineffective communication. At the same time, the outbound call control module dynamically adjusts the outbound call strategy according to the calculated intended transaction coefficient, ensuring that the marketing activities can more accurately target customer groups with real needs and purchase intentions, thereby significantly improving the success rate of marketing activities and customer satisfaction, and realizing the effective use of resources and maximization of corporate economic benefits. This method overcomes the blindness and inefficiency problems existing in traditional outbound marketing, and provides strong technical support for the company's precision marketing.

[0012] Preferably, the screening of several customers based on basic information and call duration includes:

[0013] Extract basic information and call duration of several customers;

[0014] Determine whether the age and gender in the basic information of the customer meet the requirements of the product or service; if yes, set the first tag of the corresponding customer to 1; if no, set the first tag of the corresponding customer to 0;

[0015] Determine whether the call duration is greater than the preset effective call duration; if yes, set the second tag of the corresponding customer to 1; if no, set the second tag of the corresponding customer to 0;

[0016] The customer whose sum of the first label and the second label is 2 is marked as the target customer.

[0017] The present invention sets the first label to 1 when the customer's basic information meets the product or service requirements; sets the second label to 1 when the customer's call duration is greater than the effective call duration; and marks the customer whose sum of the first label and the second label is 2 as the target customer, which can effectively eliminate the customers who do not meet the product or service requirements and the customers who have inefficient calls, thereby helping to improve the accuracy of the intelligent outbound call system in promoting the customers.

[0018] Preferably, the keyword recognition of the target customer's call content based on the speech recognition technology includes:

[0019] A1: Extract the conversation contents of several customers;

[0020] A2: Convert the call content into text messages;

[0021] A3: Process the text information based on the word segmentation software to obtain several word segments;

[0022] A4: Input several segmented words into the preset keyword database for matching;

[0023] A5: Mark the successfully matched segmented words as keywords, and integrate all keywords into a keyword sequence.

[0024] Preferably, obtaining the number of active questions based on the call content of the target customer includes:

[0025] B1: Extract the voice information from the call content of each target customer and obtain the voiceprint features of the outbound call system;

[0026] B2: Based on voiceprint recognition technology, the sentences with voiceprint features belonging to the outbound call system in the voice information are removed to obtain the customer's voice;

[0027] B3: Recognize some question words from the customer's voice based on voice recognition technology;

[0028] B4: Mark the total number of customer's question words as the number of active questions.

[0029] Preferably, the calculation of the target customer's intended transaction coefficient based on the keyword sequence and call feature information includes:

[0030] Extract keyword sequences and call feature information of several target customers; count the number of keywords in the keyword sequence of each target customer; calculate the intended transaction coefficient YCXi of the corresponding target customer by the formula YCXi=a×GJSi+b×ln(1+TSCi)+c×ZWSi; wherein GJSi is the number of keywords in the keyword sequence of target customer i, TSCi is the call duration in the call feature information of target customer i, and ZWSi is the number of active questions in the call feature information of target customer i; a, b, c are proportional coefficients greater than 0; ln() is a logarithmic function with a natural constant as the base; i=1, 2, 3, ..., n, and n is the total number of target customers.

[0031] Preferably, the method for obtaining the proportional coefficient a includes:

[0032] Obtaining the position of each keyword in the keyword sequence of each target customer in the call content, and obtaining the keyword dispersion degree of the corresponding target customer based on the position of each keyword in the call content;

[0033] The proportionality coefficient a is calculated by the formula a=0.4×si; wherein si is the keyword dispersion degree of the target customer i.

[0034] The present invention calculates the degree of keyword dispersion according to the position distribution of the keywords of each target customer in the call content, and determines the proportional coefficient a based on the keyword dispersion degree, so that the obtained proportional coefficient a and the number of keywords are substituted into the formula to calculate the intended transaction coefficient to be more accurate, which is beneficial to improving the subsequent intelligent outbound call system to adjust the outbound call strategy for the customer according to the intended transaction coefficient, and further helps to increase the probability of customer transaction.

[0035] Preferably, the obtaining the keyword dispersion degree of the corresponding target customer based on the position of each keyword in the call content includes:

[0036] Get the distance between adjacent keywords in the call content of each target customer; divide the total number of words in the call content of each target customer by the total number of words in the corresponding keywords in the call content to get the average distance between keywords corresponding to each target customer; use the standard deviation formula Calculate the keyword dispersion degree of the corresponding target customer si; where GJi,k is the number of words between the kth keyword and the k+1th keyword; M is the average keyword spacing of the target customer i; k=1, 2, 3, ..., m, m is the total number of keywords in the keyword sequence of the target customer i.

[0037] Preferably, the method for obtaining the proportional coefficient c includes:

[0038] Obtain the questions actively asked by each target customer, and classify the questions into price inquiry, parameter inquiry, and after-sales inquiry. Calculate the proportional coefficient c using the formula c=0.5×JGi+0.2×CSi+0.3×SHi; wherein JGi is the number of price inquiry questions asked by target customer i, CSi is the number of parameter inquiry questions asked by target customer i, and SHi is the number of after-sales inquiry questions asked by target customer i.

[0039] The present invention classifies the questions actively asked by target customers, sets corresponding weight coefficients for different categories of questions, and multiplies the number of questions in each category by the corresponding weight coefficient to calculate the proportional coefficient c, so that the number of active questions asked by customers uses an appropriate proportional coefficient when calculating the intended transaction coefficient, which is conducive to more accurately reflecting the customer's transaction intention.

[0040] Preferably, the outbound calling strategy for target customers is adjusted based on the intended transaction coefficient, including:

[0041] Extract the intended transaction coefficient of each target customer; determine whether the intended transaction coefficient is greater than a preset intention threshold; if yes, mark the intention level of the corresponding target customer as level one; if no, mark the intention level of the corresponding target customer as level two;

[0042] Set the outbound call strategy for target customers with a first-level intention level to arrange a human agent to call back the target customer:

[0043] The outbound calling strategy for target customers with the second level of intent is set to arrange an outbound calling robot to call back the target customers.

[0044] A second aspect of the present invention provides an intelligent outbound calling method based on customer intention, comprising:

[0045] S1: Obtain basic information and call information of several customers;

[0046] S2: Screen several customers based on basic information and call duration to obtain target customers;

[0047] S3: Perform keyword recognition on the target customer's call content based on speech recognition technology to obtain a keyword sequence;

[0048] S4: Obtain the number of active questions asked based on the call content of the target customer, and integrate the call duration and the number of active questions asked into call feature information;

[0049] S5: Calculate the target customer's intended transaction coefficient based on the keyword sequence and call feature information;

[0050] S6: Adjust the outbound calling strategy for target customers based on the intended transaction coefficient.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. The present invention obtains basic customer information and call information through the data acquisition module, and uses the voice recognition technology to perform keyword recognition and active question count on the target customer's call content through the data processing module to form call feature information, and calculates the target customer's intended transaction coefficient in combination with the keyword sequence. Finally, the outbound call control module dynamically adjusts the outbound call strategy according to the intended transaction coefficient. This method not only improves the accuracy of customer screening, but also optimizes the communication efficiency with potential high-intent customers, reduces invalid or inefficient outbound calls, thereby effectively improving the success rate of marketing activities and customer satisfaction, while reducing corporate operating costs.

[0053] 2. The present invention sets the first label to 1 when the customer's basic information meets the requirements of relevant products or services; sets the second label to 1 when the customer's call duration is greater than the effective call duration; and marks the customer whose sum of the first label and the second label is 2 as the target customer, which can effectively eliminate the customers who do not meet the product or service requirements and the customers who have inefficient calls, thereby helping to improve the accuracy of the intelligent outbound call system in promoting the customers.

[0054] 3. The present invention calculates the degree of keyword dispersion according to the position distribution of the keywords of each target customer in the call content, and determines the proportional coefficient a based on the degree of keyword dispersion, so that the obtained proportional coefficient a and the number of keywords are substituted into the formula to calculate the intended transaction coefficient to be more accurate, which is beneficial to improve the subsequent intelligent outbound call system to adjust the outbound call strategy for the customer according to the intended transaction coefficient, and further helps to increase the probability of customer transaction.

[0055] 4. The present invention classifies the questions actively asked by the target customers, sets corresponding weight coefficients for different categories of questions, and multiplies the number of questions in each category by the corresponding weight coefficient to calculate the proportional coefficient c, so that the number of active questions asked by the customer uses an appropriate proportional coefficient when calculating the intended transaction coefficient, which is conducive to more accurately reflecting the customer's transaction intention. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0057] Figure 1 The figure is an overall flow chart of the intelligent outbound calling method based on customer intention of the present invention;

[0058] Figure 2 It is a schematic diagram of the principle of the intelligent outbound calling system based on customer intention of the present invention;

[0059] Figure 3 The present invention is a flow chart for screening several customers based on basic information and call duration. DETAILED DESCRIPTION

[0060] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] See also Figure 1-Figure 3 , the first aspect of the present invention provides an intelligent outbound call system based on customer intention, including: a data processing module, and a data acquisition module and an outbound call control module connected thereto;

[0062] Data collection module: used to obtain basic information and call information of several customers; screen several customers based on basic information and call duration to obtain target customers; basic information includes age and gender; call information includes call duration, call content and call time;

[0063] Data processing module: perform keyword recognition on the target customer's call content based on speech recognition technology to obtain a keyword sequence; obtain the number of active questions based on the target customer's call content, and integrate the call duration and the number of active questions into call feature information; and,

[0064] Calculate the target customer's intended transaction coefficient based on keyword sequence and call feature information;

[0065] Outbound call control module: adjusts the outbound call strategy for target customers based on the intended transaction coefficient.

[0066] The present invention obtains basic customer information and call details through the data acquisition module, and uses the voice recognition technology and keyword recognition capabilities in the data processing module to achieve high-precision screening of target customers. At the same time, by deeply analyzing call characteristics such as call duration and the number of active questions, the intention transaction coefficient is accurately evaluated, which significantly improves the accuracy and efficiency of customer screening and reduces the cost of ineffective communication. The outbound call control module flexibly adjusts the strategy based on the intention transaction coefficient to ensure that marketing activities are accurately connected to customers with real needs, greatly improves the success rate and customer satisfaction, achieves efficient resource allocation and maximizes corporate economic benefits, effectively overcomes the blindness and inefficiency of traditional outbound marketing, and provides solid technical support for precision marketing.

[0067] In this embodiment, several customers are screened based on basic information and call duration, including:

[0068] Extract basic information and call duration of several customers;

[0069] Determine whether the age and gender in the basic information of the customer meet the requirements of the product or service; if yes, set the first tag of the corresponding customer to 1; if no, set the first tag of the corresponding customer to 0;

[0070] Determine whether the call duration is greater than the preset effective call duration; if yes, set the second tag of the corresponding customer to 1; if no, set the second tag of the corresponding customer to 0;

[0071] The customer whose sum of the first label and the second label is 2 is marked as the target customer.

[0072] For example, assuming that the basic information of customer 1 shows that the age is 27 and the gender is male, and the requirements of the relevant products are that the age requirement is 18-70 years old and the gender is not limited, then the customer meets the requirements and the first label is set to 1; the call duration between the intelligent outbound call system and customer 1 is 64s, and the preset effective call duration is 30s. Since the call duration of customer 1 is longer than the effective call duration, the second label is set to 1; since the sum of the first label and the second label is 2, customer 1 is marked as a target customer.

[0073] The present invention sets the first label to 1 when the customer's basic information meets the requirements of relevant products or services; sets the second label to 1 when the customer's call duration is greater than the effective call duration; and marks the customer whose sum of the first label and the second label is 2 as the target customer, which can effectively eliminate the customers who do not meet the product or service requirements and the customers who have inefficient calls, thereby helping to improve the accuracy of the intelligent outbound call system in promoting the customers.

[0074] In this embodiment, keyword recognition is performed on the target customer's call content based on speech recognition technology, including:

[0075] A1: Extract the conversation contents of several customers;

[0076] A2: Convert the call content into text messages;

[0077] A3: Process the text information based on the word segmentation software to obtain several word segments;

[0078] A4: Input several segmented words into the preset keyword database for matching;

[0079] A5: Mark the successfully matched segmented words as keywords, and integrate all keywords into a keyword sequence.

[0080] In this embodiment, obtaining the number of active questions based on the call content of the target customer includes:

[0081] Exemplarily, voice information is extracted from the call content of each target customer to obtain the voiceprint features of the outbound call system; Mel-frequency cepstral coefficients (MFCC) are used to identify sentences belonging to the outbound call system in the voice information to obtain the customer's voice; several question word information is identified from the customer's voice based on automatic speech recognition technology (ASR); the total number of several question word information of the customer is marked as the number of active questions.

[0082] In this embodiment, the target customer's intended transaction coefficient is calculated based on the keyword sequence and call feature information, including:

[0083] Extract keyword sequences and call feature information of several target customers; count the number of keywords in the keyword sequence of each target customer; calculate the intended transaction coefficient YCXi of the corresponding target customer by the formula YCXi=a×GJSi+b×ln(1+TSCi)+c×ZWSi; wherein GJSi is the number of keywords in the keyword sequence of target customer i, TSCi is the call duration in the call feature information of target customer i, and ZWSi is the number of active questions in the call feature information of target customer i; a, b, c are proportional coefficients greater than 0; ln() is a logarithmic function with a natural constant as the base; i=1, 2, 3, ..., n, and n is the total number of target customers.

[0084] Exemplarily, the proportional coefficients are set as a=1.948, b=2, c=1.7; the number of keywords in the keyword sequence of target customer 1 GJS1=5, the call duration in the call feature information of target customer 1 TSC1=64s, and the number of active questions in the call feature information of target customer i ZWS1=6 times; the intended transaction coefficient of target customer 1 is calculated by the formula to be YCX1≈28.29.

[0085] In this embodiment, the method for obtaining the proportional coefficient a includes:

[0086] Obtaining the position of each keyword in the keyword sequence of each target customer in the call content, and obtaining the keyword dispersion degree of the corresponding target customer based on the position of each keyword in the call content;

[0087] The proportionality coefficient a is calculated by the formula a=0.4×si; wherein si is the keyword dispersion degree of the target customer i.

[0088] Exemplarily, the keyword dispersion degree s1 of target customer 1 is set to 4.87; the proportional coefficient a=0.4×4.87=1.948 is calculated by the formula.

[0089] The present invention calculates the dispersion degree of keywords according to the distribution of the keywords of each target customer in the call content, and determines the proportional coefficient a based on the dispersion degree, so that the intended transaction coefficient calculated by substituting the proportional coefficient a and the number of keywords into the formula is more accurate. This method significantly improves the calculation accuracy of the intended transaction coefficient, so that the subsequent intelligent outbound call system can adjust the outbound call strategy for customers based on the more accurate intended transaction coefficient, further improve the probability of customer transactions, optimize the outbound call effect, and improve customer experience and overall operational efficiency.

[0090] In this embodiment, the keyword dispersion degree of the corresponding target customer is obtained based on the position of each keyword in the call content, including:

[0091] Get the distance between adjacent keywords in the call content of each target customer; divide the total number of words in the call content of each target customer by the total number of words in the corresponding keywords in the call content to get the average distance between keywords corresponding to each target customer; use the standard deviation formula Calculate the keyword dispersion degree of the corresponding target customer si; where GJi,k is the number of words between the kth keyword and the k+1th keyword; M is the average keyword spacing of the target customer i; k=1, 2, 3, ..., m, m is the total number of keywords in the keyword sequence of the target customer i.

[0092] It should be noted that the degree of keyword dispersion can reflect the distribution of keywords in the call content; when the degree of keyword dispersion is smaller, the keyword distribution is more concentrated, which is not conducive to customers' understanding in a short time, thus resulting in a decrease in the probability of customer transactions; when the degree of keyword dispersion is greater, the key distribution is more uniform, and customers have more time to understand or think about the keywords, which is conducive to increasing the probability of customer transactions.

[0093] Exemplarily, the total number of keywords in the call content of target customer 1 is set to 5, and the total number of words in the call content is 120. The total number of words in the call content of each target customer is divided by the total number of words of the keywords in the corresponding call content, and the average keyword spacing M1=24 corresponding to each target customer is obtained; the spacing between adjacent keywords is GJ1=18, GJ2=23, GJ3=21, GJ4=31, and the keyword dispersion degree s1≈4.87 of target customer 1 is calculated by the standard deviation formula.

[0094] In this embodiment, the method for obtaining the proportional coefficient b includes:

[0095] T1: Obtain the calling time from the call information of each target customer;

[0096] T2: Determine whether the dialing time is between 17:00 and 19:00; if yes, set the value of the proportional coefficient b to 2; if no, jump to T3;

[0097] T3: Determine whether the dialing time is between 9:00 and 11:00 or between 14:00 and 17:00; if yes, set the value of the proportional coefficient b to 1.5; if no, set the value of the proportional coefficient b to 1.

[0098] Exemplarily, the dialing time for the intelligent outbound call system to talk to the target customer 1 is set to 17:23. Since the dialing time is between 17:00 and 19:00, the value of the proportional coefficient b is set to b1=2.

[0099] It should be noted that the calling time of the outbound call system has a certain impact on increasing customer transactions; for example, when the calling time is between 17:00 and 19:00, most customers are off work. At this time, when the outbound call system calls the customer, the probability of the customer answering the call is higher, so the proportional coefficient b is set to a larger b3.

[0100] The present invention sets a corresponding proportional coefficient b according to the time period of the outbound call system's dialing time for each target customer, so that the customer's call duration uses an appropriate proportional coefficient when calculating the intended transaction coefficient, which is conducive to more accurately reflecting the customer's transaction intention.

[0101] In this embodiment, the method for obtaining the proportionality coefficient c includes:

[0102] Obtain the questions actively asked by each target customer, and classify the questions into price inquiry, parameter inquiry, and after-sales inquiry. Calculate the proportional coefficient c using the formula c=0.5×JGi+0.2×CSi+0.3×SHi; wherein JGi is the number of price inquiry questions asked by target customer i, CSi is the number of parameter inquiry questions asked by target customer i, and SHi is the number of after-sales inquiry questions asked by target customer i.

[0103] Exemplarily, the number of active questions asked by target customer 1 is set to 6 times, the number of price-related questions asked by target customer 1 JG1=1, the number of parameter-related questions asked by target customer 1 CS1=3, and the number of after-sales-related questions asked by target customer 1 SH1=2; the proportional coefficient c=1.7 is calculated by the formula.

[0104] The present invention classifies the questions actively asked by target customers, sets corresponding weight coefficients for different categories of questions, and multiplies the number of questions in each category by the corresponding weight coefficient to calculate the proportional coefficient c, so that the number of active questions asked by customers uses an appropriate proportional coefficient when calculating the intended transaction coefficient, which is conducive to more accurately reflecting the customer's transaction intention.

[0105] In this embodiment, the outbound calling strategy for target customers is adjusted based on the intended transaction coefficient, including:

[0106] Extract the intended transaction coefficient of each target customer; determine whether the intended transaction coefficient is greater than a preset intention threshold; if yes, mark the intention level of the corresponding target customer as level one; if no, mark the intention level of the corresponding target customer as level two;

[0107] Set the outbound call strategy for target customers with a first-level intention level to arrange a human agent to call back the target customer:

[0108] The outbound calling strategy for target customers with the second level of intent is set to arrange an outbound calling robot to call back the target customers.

[0109] Exemplarily, the target customer 1's intention transaction coefficient is set to 28.29, and the preset intention threshold is 25; because the target customer 1's intention transaction coefficient is greater than the preset intention threshold, the target customer 1's intention level is marked as level one; the target customer 1's outbound call strategy is set to arrange a manual agent to call the target customer back.

[0110] A second aspect of the present invention provides an intelligent outbound calling method based on customer intention, comprising:

[0111] S1: Obtain basic information and call information of several customers;

[0112] S2: Screen several customers based on basic information and call duration to obtain target customers;

[0113] S3: Perform keyword recognition on the target customer's call content based on speech recognition technology to obtain a keyword sequence;

[0114] S4: Obtain the number of active questions asked based on the call content of the target customer, and integrate the call duration and the number of active questions asked into call feature information;

[0115] S5: Calculate the target customer's intended transaction coefficient based on the keyword sequence and call feature information;

[0116] S6: Adjust the outbound calling strategy for target customers based on the intended transaction coefficient.

[0117] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0118] Working principle of the present invention:

[0119] The present invention obtains basic information and call information of several customers; screens several customers based on the basic information and call duration to obtain target customers; performs keyword recognition on the call content of the target customers based on speech recognition technology to obtain a keyword sequence; obtains the number of active questions based on the call content of the target customers, and integrates the call duration and the number of active questions into call feature information; calculates the intended transaction coefficient of the target customers based on the keyword sequence and the call feature information; and adjusts the outbound call strategy for the target customers based on the intended transaction coefficient.

[0120] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent outbound calling system based on customer intention, comprising: The data processing module, and the data acquisition module and outbound call control module connected thereto are characterized in that: The data collection module is used to obtain basic information and call information of several customers; screen several customers based on the basic information and call duration to obtain target customers; wherein the basic information includes age and gender; the call information includes call duration, call content and call time; The data processing module: performs keyword recognition on the target customer's call content based on speech recognition technology to obtain a keyword sequence; obtains the number of active questions based on the target customer's call content, and integrates the call duration and the number of active questions into call feature information; and Calculate the target customer's intended transaction coefficient based on keyword sequence and call feature information; The outbound call control module adjusts the outbound call strategy for target customers based on the intended transaction coefficient.

2. According to claim 1, an intelligent outbound calling system based on customer intention is characterized in that: The screening of several customers based on basic information and call duration includes: Extract basic information and call duration of several customers; Determine whether the age and gender in the basic information of the customer meet the requirements of the product or service; if yes, set the first tag of the corresponding customer to 1; if no, set the first tag of the corresponding customer to 0; Determine whether the call duration is greater than the preset effective call duration; if yes, set the second tag of the corresponding customer to 1; if no, set the second tag of the corresponding customer to 0; The customer whose sum of the first label and the second label is 2 is marked as the target customer.

3. According to claim 1, the intelligent outbound calling system based on customer intention is characterized in that: The keyword recognition of the target customer's call content based on the speech recognition technology includes: A1: Extract the conversation contents of several customers; A2: Convert the call content into text messages; A3: Process the text information based on the word segmentation software to obtain several word segments; A4: Input several segmented words into the preset keyword database for matching; A5: Mark the successfully matched segmented words as keywords, and integrate all keywords into a keyword sequence.

4. According to claim 1, the intelligent outbound calling system based on customer intention is characterized in that: The method of obtaining the number of active questions based on the call content of the target customer includes: B1: Extract the voice information from the call content of each target customer and obtain the voiceprint features of the outbound call system; B2: Based on voiceprint recognition technology, the sentences with voiceprint features belonging to the outbound call system in the voice information are removed to obtain the customer's voice; B3: Recognize some question words from the customer's voice based on voice recognition technology; B4: Mark the total number of customer's question words as the number of active questions.

5. According to claim 1, the intelligent outbound calling system based on customer intention is characterized in that: The calculation of the target customer's intended transaction coefficient based on the keyword sequence and call feature information includes: Extract keyword sequences and call feature information of several target customers; count the number of keywords in the keyword sequence of each target customer; calculate the intended transaction coefficient YCXi of the corresponding target customer by the formula YCXi=a×GJSi+b×ln(1+TSCi)+c×ZWSi; wherein GJSi is the number of keywords in the keyword sequence of target customer i, TSCi is the call duration in the call feature information of target customer i, and ZWSi is the number of active questions in the call feature information of target customer i; a, b, c are proportional coefficients greater than 0; ln() is a logarithmic function with a natural constant as the base; i=1, 2, 3, ..., n, and n is the total number of target customers.

6. According to claim 5, an intelligent outbound calling system based on customer intention is characterized in that: The method for obtaining the proportional coefficient a includes: Obtaining the position of each keyword in the keyword sequence of each target customer in the call content, and obtaining the keyword dispersion degree of the corresponding target customer based on the position of each keyword in the call content; The proportionality coefficient a is calculated by the formula a=0.4×si; wherein si is the keyword dispersion degree of the target customer i.

7. The intelligent outbound calling system based on customer intention according to claim 6, characterized in that: The obtaining of the keyword dispersion degree of the corresponding target customer based on the position of each keyword in the call content includes: Get the distance between adjacent keywords in the call content of each target customer; divide the total number of words in the call content of each target customer by the total number of words in the corresponding keywords in the call content to get the average distance between keywords corresponding to each target customer; use the standard deviation formula Calculate the keyword dispersion degree of the corresponding target customer si; where GJi,k is the number of words between the kth keyword and the k+1th keyword; M is the average keyword spacing of the target customer i; k=1, 2, 3, ..., m, m is the total number of keywords in the keyword sequence of the target customer i.

8. According to claim 5, the intelligent outbound calling system based on customer intention is characterized in that: The method for obtaining the proportional coefficient c includes: Obtain the questions actively asked by each target customer, and classify the questions into price inquiry, parameter inquiry, and after-sales inquiry. Calculate the proportional coefficient c using the formula c=0.5×JGi+0.2×CSi+0.3×SHi; wherein JGi is the number of price inquiry questions asked by target customer i, CSi is the number of parameter inquiry questions asked by target customer i, and SHi is the number of after-sales inquiry questions asked by target customer i.

9. The intelligent outbound calling system based on customer intention according to claim 1, characterized in that: The outbound calling strategy for target customers is adjusted based on the intended transaction coefficient, including: Extract the intended transaction coefficient of each target customer; determine whether the intended transaction coefficient is greater than a preset intention threshold; if yes, mark the intention level of the corresponding target customer as level one; if no, mark the intention level of the corresponding target customer as level two; Set the outbound call strategy for target customers with a first-level intention level to arrange a human agent to call back the target customer: The outbound calling strategy for target customers with the second level of intent is set to arrange an outbound calling robot to call back the target customers.

10. An intelligent outbound calling method based on customer intention, based on the intelligent outbound calling system based on customer intention according to any one of claims 1 to 9, characterized in that: include: S1: Obtain basic information and call information of several customers; S2: Screen several customers based on basic information and call duration to obtain target customers; S3: Perform keyword recognition on the target customer's call content based on speech recognition technology to obtain a keyword sequence; S4: Obtain the number of active questions asked based on the call content of the target customer, and integrate the call duration and the number of active questions asked into call feature information; S5: Calculate the target customer's intended transaction coefficient based on the keyword sequence and call feature information; S6: Adjust the outbound calling strategy for target customers based on the intended transaction coefficient.

Images