Intelligent response processing method and system

Abstract

The invention belongs to the technical field, and provides an intelligent response processing method and system. The method comprises the steps of collecting information input by a current user, extracting key demands and preferences of the current user, and determining user demands; inputting the determined user demand into a demand classification model to obtain a demand classification result, and when the result is a suggested demand, responding to the customer by matching a preset reply; when the result is a recommendation demand, determining a pre-pushed commodity meeting the recommendation demand according to a keyword in the recommendation demand, further performing comprehensive evaluation on the pre-pushed commodity, and determining a pushed commodity; and determining the customer category of the current user according to the consumption information data of the current user, performing sorting calculation on the determined pushed commodities, determining the optimal commodity as response information, and pushing the response information to the current user. According to the method, the retrieval problem of the customer can be effectively identified, and customized and dynamically calculated and adjusted commodity recommendation services can be provided on the basis of fully understanding customer demands.

Description

Technical Field

[0001] The present invention relates to the field of intelligent recognition technology, and in particular to an intelligent response processing method and system. Background Art

[0002] With the rapid development of artificial intelligence (AI) technology, especially with the progress in natural language processing (NLP) and machine learning, online customer service systems have been widely used in various industries. At present, many online customer service systems have achieved automated responses through intelligent means. These systems usually rely on AI algorithms, especially deep learning and dialogue systems, to be able to identify and understand customer questions in real time. For example, common chatbots and virtual assistants use NLP technology to parse text information entered by customers, extract key intent from it, and provide corresponding answers. This technology performs well in handling common questions such as inquiring order status, operating guides, and basic information consultation, which not only improves customer service efficiency, but also significantly reduces labor costs.

[0003] In the existing online customer service system, when the customer's questions involve specific product needs, the artificial intelligence system often performs unsatisfactorily. Although AI can identify the needs raised by customers and provide relevant product information, it still has many defects in recommending reasonable store products. Secondly, the existing customer service system lacks dynamic adjustment and real-time response capabilities in the product recommendation process. This not only affects the customer experience, but also leads to the loss of potential sales opportunities.

[0004] Therefore, it is necessary to provide an intelligent answer processing method and system that can effectively identify customer questions and provide personalized and dynamically adjusted product recommendation services based on a full understanding of customer needs. Summary of the invention

[0005] The present invention aims to provide an intelligent answer processing method and system to solve the technical problems of poor customer experience and loss of potential sales opportunities caused by the lack of dynamic adjustment and real-time response capabilities in the existing product recommendation process. The technical problem to be solved by the present invention is achieved through the following technical solution.

[0006] The first aspect of the present invention provides an intelligent answer processing method, including: collecting information input by the current user, using natural language processing technology to parse the information input by the user, extracting the key needs and preferences of the current user, performing semantic understanding through a pre-trained language model, and determining user needs; inputting the determined user needs into a demand classification model to obtain a demand classification result, wherein when the demand classification result is a suggested demand, answering the customer by matching a preset reply; when the demand classification result is a recommended demand, filtering the related products involved in the keywords according to the keywords in the recommended demand, determining the pre-promoted products that meet the recommended demand, further comprehensively evaluating the pre-promoted products, and determining the pushed products; determining the customer category of the current user according to the consumption information data of the current user; sorting and calculating the determined pushed products according to the customer category corresponding to the current user, determining the best product as the answer information, and pushing it to the current user.

[0007] The second aspect of the present invention provides an intelligent answer processing system, which uses the method described in the first aspect of the present invention, and the intelligent answer processing system includes: a collection and processing module, which is used to collect information input by the current user, use natural language processing technology to parse the information input by the user, extract the key needs and preferences of the current user, and perform semantic understanding through a pre-trained language model to determine the user needs; a first model calculation module, which inputs the determined user needs into a demand classification model to obtain a demand classification result, wherein when the demand classification result is a recommended demand, the customer is answered by matching a preset reply; when the demand classification result is a recommended demand, according to the keywords in the recommended demand, the relevant products involved in the keywords are filtered layer by layer to determine the pre-promoted products that meet the recommended demand, and further comprehensively evaluate the pre-promoted products to determine the pushed products; a category determination module, which determines the customer category of the current user according to the consumption information data of the current user; a calculation and push module, which sorts and calculates the determined pushed products according to the customer category corresponding to the current user, determines the best product as the answer information, and pushes it to the current user.

[0008] The third aspect of the present invention provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the intelligent response processing method described in the first aspect of the present invention.

[0009] A fourth aspect of the present invention provides a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the intelligent response processing method described in the first aspect of the present invention.

[0010] Beneficial effects of the present invention:

[0011] Compared with the prior art, the present invention can effectively identify and determine user needs by extracting the key needs and preferences of the current user, input the demand classification model, obtain the demand classification result, and when the demand classification result is a recommended demand, the customer is answered by matching the preset reply; when the demand classification result is a recommended demand, the relevant commodities involved in the keywords are filtered layer by layer according to the keywords in the recommended demand, and the pre-pushed commodities that meet the recommended demand are determined, and the pre-pushed commodities are further comprehensively evaluated, and the pushed commodities to be pushed can be dynamically calculated and adjusted. The customer category of the current user is determined according to the consumption information data of the current user, and the determined pushed commodities are sorted and calculated, and the best commodity is determined as the response information, and the commodity of the customized user type can be provided to be pushed to the current user. This method can not only effectively identify the customer's search problem, but also provide customized, dynamically calculated and adjusted commodity recommendation services based on a full understanding of customer needs. At the same time, by integrating multi-channel data, a comprehensive analysis and real-time response to customer behavior can be achieved, so as to better meet the personalized needs of customers and improve the overall service level of customer responses. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a flowchart of an example of the intelligent answer processing method of the present invention;

[0013] Figure 2 is a partial flow chart of an implementation method of the intelligent response processing method of the present invention;

[0014] Figure 3 This is a schematic diagram of user sub-classification according to an embodiment of the intelligent answer processing method of the present invention;

[0015] Figure 4 is a schematic diagram of an example of the intelligent response processing system of the present invention;

[0016] Figure 5 is a schematic structural diagram of an electronic device embodiment according to the present invention;

[0017] Figure 6 It is a schematic diagram of the structure of a computer-readable medium embodiment according to the present invention. DETAILED DESCRIPTION

[0018] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0019] In view of the above problems, the present invention provides an intelligent answer processing method, which extracts the key needs and preferences of the current user, can effectively identify and determine the user needs, input the demand classification model, obtain the demand classification result, and when the demand classification result is a recommended demand, the customer is answered by matching the preset reply; when the demand classification result is a recommended demand, the relevant commodities involved in the keywords are filtered layer by layer according to the keywords in the recommended demand, and the pre-promoted commodities that meet the recommended demand are determined, and the pre-promoted commodities are further comprehensively evaluated, and the pushed commodities to be pushed can be dynamically calculated and adjusted. The customer category of the current user is determined according to the consumption information data of the current user, and the determined pushed commodities are sorted and calculated, and the best commodity is determined as the answer information, and the commodity of the customized user type can be provided to be pushed to the current user. This method can not only effectively identify the customer's search problem, but also provide customized, dynamically calculated and adjusted commodity recommendation services based on a full understanding of the customer's needs. At the same time, by integrating multi-channel data, a comprehensive analysis and real-time response to customer behavior can be achieved, so as to better meet the personalized needs of customers and improve the overall service level of customer responses.

[0020] It should be noted that the method of the present invention is widely used, and is particularly suitable for the interactive processing process or response processing process of internal technical data between users in an enterprise department. In addition, the present invention can also be applied to the interactive processing process or response processing process of data information such as landmark buildings, roads, etc. obtained by various sensors in urban intelligent construction scenarios. In addition, it can also be applied to telemedicine, collecting motion data of remote doctors, etc. In the above three application scenarios, data information recommendations are provided to users, and the corresponding pushed products can also be understood as digital information. In addition, it is also applicable to the interactive processing process or response processing process of purchase searches in shopping scenarios.

[0021] Example 1

[0022] Refer to the following Figure 1 , Figure 2 and Figure 3 , the intelligent response processing method of the present invention will be described in detail.

[0023] Figure 1 It is a step flow chart of an example of the intelligent response processing method of the present invention.

[0024] First, in step S101, the information input by the current user is collected, and the information input by the user is parsed using natural language processing technology to extract the key needs and preferences of the current user, and semantic understanding is performed through a pre-trained language model to determine the user needs.

[0025] Specifically, key requirements and preferences are extracted, wherein the key requirements are specifically the required subject, i.e., the subject, such as freeze-dried, cat food, and yogurt, and the preferences are specifically qualifiers for the required subject, such as zero sugar.

[0026] In a specific implementation, information about freeze-dried cat food, zero-sugar yogurt, etc. input by the current user is collected from a shopping platform. After parsing, the following key requirements (e.g., freeze-dried, cat food, yogurt) and preferences (e.g., zero sugar, zero fat, no additives) are extracted. Therefore, it is determined that the user needs to buy cat food and yogurt related to cats.

[0027] In another specific embodiment, in the scenario of urban intelligent construction, the user requests data information such as landmark buildings, roads, etc., and the central data system stores the above data information. The central data system is integrated with a demand classification model to identify the key needs and preferences of the user. For example, when the central data system receives the information input by the current user (for example, requesting the height information of building a on road 1), the requested height information is returned to the current user. Specifically, the information input by the user is parsed using natural language processing technology to extract the key needs (such as road 1, building a) and preferences (such as height information) of the current user to determine that the current user wants to obtain the height information of building a on road 1.

[0028] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0029] Next, in step S102, the determined user demand is input into the demand classification model to obtain a demand classification result, wherein when the demand classification result is a suggested demand, the customer is answered by matching the preset reply; when the demand classification result is a recommended demand, the related products involved in the keywords are filtered layer by layer according to the keywords in the recommended demand to determine the pre-promoted products that meet the recommended demand, and further a comprehensive evaluation is performed on the pre-promoted products to determine the pushed products.

[0030] Specifically, there are two types of demand classification: suggestion demand and recommendation demand. Suggestion demand means asking questions about a certain product; recommendation demand means recommending corresponding products based on customer needs.

[0031] In a specific embodiment, a demand classification model is established through deep learning methods, such as convolutional neural networks and recurrent neural networks.

[0032] Specifically, text information data marked with the above two demand classification labels (for example, the search information entered by the user on a shopping platform: cat food, 500g) is used to establish a training data set for training demand classification to obtain a trained demand classification model.

[0033] It should be noted that, in another specific implementation, text information with multiple demand classification labels (suggested demand is 1, recommended demand is 2, and preference type is 3) can also be used (for example, the search information entered by the user on a shopping platform: cat food, how many kilograms, brand a). In another specific implementation, the urban intelligent construction scenario can also use text information with multiple demand classification labels, and the multiple demand classification labels specifically include: suggested demand is 1, recommended demand is 2, and preference type is, for example, a multi-dimensional vector of (3, 4, 5) (i.e., multiple preference types, such as height, length, position, etc.). The above is only explained as an optional example and cannot be understood as a limitation of the present invention.

[0034] Specifically, the determined user demand (eg, freeze-dried food, cat food, yogurt) is input into a demand classification model to obtain a demand classification result.

[0035] For the demand classification results obtained, when the demand classification result is a recommended demand, the customer is answered by matching the preset reply.

[0036] For example, the preset responses of cat food and yogurt are matched with the preference information in the current user demand input (for example, calculated using cosine similarity). When the calculated value of the matching calculation is greater than the set value (for example, 95%), the preset response corresponding to the calculated value is used as the response information to answer or reply to the customer.

[0037] For the obtained demand classification results, when the demand classification result is a recommended demand, the related products involved in the keywords are filtered layer by layer according to the keywords in the recommended demand to determine the pre-promoted products that meet the recommended demand, and further comprehensively evaluate the pre-promoted products to determine the pushed products.

[0038] According to the keywords in the recommended requirements, the related products involved in the keywords are filtered layer by layer. For example, the search terms (such as cat food, freeze-dried, 3-6 months, 2 kg) are filtered in sequence to obtain 2 kg of freeze-dried cat food suitable for 3-6 month old kittens.

[0039] Through the above-mentioned layer-by-layer filtering method, it is possible to determine the pre-recommended products that meet the recommendation requirements.

[0040] Further conduct a comprehensive evaluation of the pre-promoted products and determine the products to be pushed.

[0041] For comprehensive evaluation, such as Figure 2 As shown, the specific steps include:

[0042] Step S201: determine s pre-promoted products, randomly select one of them as a target product, obtain all evaluations of the target product within a specified time period, and divide the specified time period into m interval time periods t.

[0043] For example, a product is randomly selected from s pre-recommended products as the target product, and all evaluations of the target product within a specified time period (for example, a time period 90 days ago from the current moment) are obtained. Then, 90 days are evenly divided into m interval time periods t, each interval time period t=90 / m, where m is a preset value, for example, determined according to the type of the target product and the number of evaluations.

[0044] The determined pushed products are marked as pre-pushed products.

[0045] S202: Selecting an interval time period from the m interval time periods t as a target interval time, and determining a favorable review rate of the target interval time.

[0046] For m interval time periods t, select one of the interval times as the target interval time, obtain the number Gj of evaluations and the number Fj of favorable comments of the target product within the target interval time t, and determine the favorable comment rate Hj of the target interval time t through the formula Hj=Fj / Gj, where 1≤j≤m.

[0047] Step S203: Repeat step S202 to determine the praise rates of m interval time periods, calculate the average of the m praise rates to further determine the deviation values ​​of the m praise rates, and sort and delete them according to the determined deviation values ​​until the first ratio is less than the first predetermined value, and then calculate the average of the remaining praise rates as the standard praise rate.

[0048] The m interval time periods are all taken as target interval time periods, that is, step S202 is repeatedly executed to determine the favorable review rates Hj of all interval time periods t, and then the average value Hp of the m favorable review rates is calculated.

[0049] It should be noted that in another embodiment, another step is included: repeating steps S201 to S203 until it is determined that all the pre-recommended products have a standard favorable review rate Hv of at least 100%. The above is only described as a preferred example and cannot be understood as a limitation of the present invention.

[0050] Step S204: Select a specific number of top-ranked recommended products as the pushed products.

[0051] Use the following expression to calculate the deviation value U of m positive reviews:

[0052]

[0053] Wherein, U represents the deviation value of the favorable comment rate of the target interval time period; Hj represents the favorable comment rate of the jth target interval time period, j is a positive integer, specifically 1, 2, ..., m; Hp represents the average favorable comment rate of m target interval time periods;

[0054] Each time the highest praise rate is deleted, the calculation of the first ratio is performed once;

[0055] The first ratio is calculated using the following expression:

[0056]

[0057] Among them, B represents the ratio of the number SQ of deleted favorable reviews to m, that is, the first ratio; SQ represents the number of deleted favorable reviews; and m represents the number of target interval time periods.

[0058] The deviation value U is judged. When the calculated deviation value U is greater than or equal to the first preset value Q1, the highest ranked positive rate is deleted, and the deviation values ​​of the remaining positive rates except the deleted positive rate are recalculated until the first ratio is less than the first preset value Q1 and the calculated deviation value is less than the second preset value. The remaining positive rate is calculated as the standard positive rate.

[0059] Specifically, when the deviation value U is greater than or equal to the first preset value Q1, the values ​​are sorted in descending order according to |Hj-Hp|, the Hj with the largest sorting value is deleted in turn, and the deviation value U of the remaining Hj values ​​is recalculated until the deviation value U is less than the first preset value Q1, and the number of deleted Hj values ​​is counted.

[0060] When the ratio of the number of deleted Hj to m is greater than the second preset value Q2, the average value of the deleted Hj is calculated as the standard favorable review rate Hv, where 1≤v≤s.

[0061] When the deviation value U is smaller than the first preset value Q1, the average value Hp of the m favorable review rates is taken as the standard favorable review rate Hv.

[0062] Next, in step S103, the customer category of the current user is determined according to the consumption information data of the current user.

[0063] Specifically, the number of orders g1 of all successful purchases of the current user within the preset time T is obtained, and the number of orders g1 is compared with the third preset value. When the number of orders g1 is greater than or equal to the third preset value Q3, the user is determined to be a first frequency band customer. When the number of orders g1 is less than the third preset value Q3, the user is determined to be a second frequency band customer, wherein the preset time T is represented by a time period starting from the current moment and pushed forward n days.

[0064] Optionally, the predetermined time T is calculated from the current time and pushed forward three months, six months, nine months or twelve months.

[0065] For example, the number g1 of all successful purchase orders of the customer within a preset time T is obtained, and when the number g1 of successful purchase orders is compared with a third preset value Q3, when the number g1 of successful purchase orders is greater than or equal to the third preset value Q3, it is determined to be a first frequency band customer (for example, one or more high frequency customers corresponding to one or more segments, such as Figure 3 When the number of successful purchase orders g1 is less than the third preset value Q3, the user is determined to be a second frequency band customer (for example, one or more low-frequency customers corresponding to one or more frequency bands, such as Figure 3 As shown, there are low-frequency users c, low-frequency users d, and low-frequency users e), where the preset time T represents a time period starting from the current moment and moving forward n days, and the current day is not included in the calculation, and n is a preset value.

[0066] It should be noted that, in another embodiment, the period may also be two months, five months, ten months, one year, etc. The above is only described as an optional example and cannot be understood as a limitation of the present invention.

[0067] Next, the sum of the prices Dz of all orders of the current user within the preset time T is calculated, and the average price Bz of different commodities of the order quantity g1 is determined by the formula Bz=Dz / g1. When the average price Bz is compared with the fourth preset value Q4, when the average price Bz is greater than or equal to the fourth preset value Q4, it is determined to be a first-value customer (for example Figure 3 When the average price Bz is less than the fourth preset value, it is determined to be a second value customer (for example Figure 3 The low-priced user p1 and the low-priced user p2 are shown).

[0068] Next, in step S104, the determined pushed commodities are sorted and calculated according to the customer category corresponding to the current user, and the best commodity is determined to be pushed to the current user.

[0069] In a specific implementation, users are subdivided according to their consumption habits into high-frequency low-price customers, high-frequency high-price customers, low-frequency low-price customers, and low-frequency high-price customers. The specific process of determining users is described in detail below.

[0070] When it is determined that the current user is a first-band user and a first-value user, the pushed products are sorted from high to low according to the standard favorable comment rate, and the first-ranked pushed product is selected as the best product.

[0071] When the current user is determined to be a first-band user and a second-value user, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product:

[0072] Wi=α1×Hv+β1×Xi+θ1×Ri,

[0073] Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α1 represents the first calculation parameter when the current user is a first-band user and a second-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the current user is a first-band user and a second-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the current user is a first-band user and a second-value user.

[0074] When the current user is determined to be a second-band user and a first-value user, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product:

[0075] Wi=α3×Hv+β3×Xi+θ3×Ri,

[0076] Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α3 represents the first calculation parameter when the current user is a second-band user and a first-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the current user is a second-band user and a first-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the current user is a second-band user and a first-value user.

[0077] When the current user is determined to be a second-band user and a second-value user, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product:

[0078] Wi=α2×Hv+β2×Xi+θ2×Ri,

[0079] Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α3 represents the first calculation parameter when the user is a second-band user and a second-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the user is a second-band user and a second-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the user is a second-band user and a second-value user.

[0080] After calculation, the first ranked product is selected as the best product to be pushed to the current user, for example, in a dialog box pushed to the current user, where α1, β1, and θ1 are preset factors, α2, β2, and θ2 are preset factors, and α3, β3, and θ3 are preset factors.

[0081] It should be noted that the above-mentioned preset factors can be used to adaptively adjust the evaluation parameters according to the needs of platform sales volume and sales revenue. For example, it is necessary to recommend products with high sales volume, general evaluation but low price to a certain type of users. The corresponding sales volume parameter is larger, and the corresponding evaluation parameter is taken in the middle, such as 85%, and the corresponding price parameter has a greater impact. When the strategy needs to be modified, these preset factors will also be adjusted accordingly. Specifically, first, moderate adjustments are made based on experience, and then appropriate adjustments are made based on the modified user feedback, that is, the preset factors adjusted multiple times are used for calculation. For different application scenarios, the above-mentioned preset factors will also be determined by multiple adjustments. The above is only explained as an optional example and cannot be understood as a limitation of the present invention.

[0082] Compared with the prior art, the present invention can effectively identify and determine user needs by extracting the key needs and preferences of the current user, input the demand classification model, obtain the demand classification result, and when the demand classification result is a recommended demand, the customer is answered by matching the preset reply; when the demand classification result is a recommended demand, the relevant commodities involved in the keywords are filtered layer by layer according to the keywords in the recommended demand, and the pre-pushed commodities that meet the recommended demand are determined, and the pre-pushed commodities are further comprehensively evaluated, and the pushed commodities to be pushed can be dynamically calculated and adjusted. The customer category of the current user is determined according to the consumption information data of the current user, and the determined pushed commodities are sorted and calculated, and the best commodity is determined as the response information, and the commodity of the customized user type can be provided to be pushed to the current user. This method can not only effectively identify the customer's search problem, but also provide customized, dynamically calculated and adjusted commodity recommendation services based on a full understanding of customer needs. At the same time, by integrating multi-channel data, a comprehensive analysis and real-time response to customer behavior can be achieved, so as to better meet the customized needs of customers and improve the overall service level of customer responses.

[0083] Example 2

[0084] The following is a system embodiment of the present invention, which can be used to implement the method embodiment of the present invention. For details not disclosed in the system embodiment of the present invention, please refer to the method embodiment of the present invention.

[0085] Figure 4 It is a structural diagram of an example of the intelligent answer processing system according to the present invention.

[0086] The following will refer to Figure 4, an intelligent response processing system is described, which uses the intelligent response processing method described in the first aspect of the present invention.

[0087] like Figure 4 As shown, the intelligent answer processing system 400 includes a collection processing module 410 , a first model calculation module 420 , a category determination module 430 and a calculation push module 440 .

[0088] In a specific embodiment, the collection and processing module 410 is used to collect the information input by the current user, use natural language processing technology to parse the information input by the user, extract the key needs and preferences of the current user, perform semantic understanding through the pre-trained language model, and determine the user needs. The first model calculation module 420 inputs the determined user needs into the demand classification model to obtain the demand classification result, wherein, when the demand classification result is a recommended demand, the customer is answered by matching the preset reply; when the demand classification result is a recommended demand, the relevant commodities involved in the keywords are filtered layer by layer according to the keywords in the recommended demand, and the pre-promoted commodities that meet the recommended demand are determined, and the pre-promoted commodities are further comprehensively evaluated to determine the pushed commodities. The category determination module 430 determines the customer category of the current user based on the consumption information data of the current user. The calculation and push module 440 sorts and calculates the determined pushed commodities according to the customer category corresponding to the current user, determines the best commodity as the response information, and pushes it to the current user.

[0089] According to an optional implementation manner, further comprehensively evaluating the pre-promoted products to determine the pushed products includes performing the following steps:

[0090] Step S201: determine s pre-promoted products, randomly select one of them as a target product, obtain all evaluations of the target product within a specified time period, and divide the specified time period into m interval time periods t.

[0091] Step S202: selecting an interval time period from the m interval time periods t as a target interval time, and determining a favorable review rate of the target interval time.

[0092] Step S203: Repeat step S202 to determine the praise rates of m interval time periods, calculate the average of the m praise rates to further determine the deviation values ​​of the m praise rates, and sort and delete them according to the determined deviation values ​​until the first ratio is less than the first predetermined value, and then calculate the average of the remaining praise rates as the standard praise rate.

[0093] Step S204: Select a specific number of top-ranked recommended products as the pushed products.

[0094] Use the following expression to calculate the deviation value U of m positive reviews:

[0095]

[0096] Where U represents the deviation value of the favorable comment rate in the target interval time period; H j represents the praise rate of the jth target interval, where j is a positive integer, specifically 1, 2, ..., m; H p represents the average of the favorable reviews in m target intervals;

[0097] When the calculated deviation value U is greater than or equal to the first preset value, the highest ranked praise rate is deleted, and the deviation values ​​of the remaining praise rates except the deleted praise rate are recalculated until the first ratio is less than the first preset value and the calculated deviation value is less than the second preset value, and the remaining praise rates are calculated as the standard praise rates.

[0098] According to an optional implementation manner, each time the highest ranking praise rate is deleted, the calculation of the first ratio is performed once;

[0099] The first ratio is calculated using the following expression:

[0100]

[0101] Among them, B represents the ratio of the number SQ of deleted favorable reviews to m, that is, the first ratio; SQ represents the number of deleted favorable reviews; and m represents the number of target interval time periods.

[0102] According to an optional implementation manner, the number g1 of all orders successfully purchased by the current user within a preset time T is obtained, and the order number g1 is compared with a third preset value. When the order number g1 is greater than or equal to the third preset value, the user is determined to be a first-band customer; when the order number g1 is less than the third preset value, the user is determined to be a second-band customer, wherein the preset time T is represented as a time period n days forward from the current moment; the total price Dz of all orders of the current user within the preset time T is calculated, and the average price Bz of different commodities with the order number g1 is determined by the formula Bz=Dz / g1. When the average price Bz is compared with a fourth preset value, when the average price Bz is greater than or equal to the fourth preset value, the user is determined to be a first-value customer, and when the average price Bz is less than the fourth preset value, the user is determined to be a second-value customer.

[0103] According to an optional implementation, when the demand classification result is a recommended demand, the related commodities involved in the keywords are filtered layer by layer according to the keywords in the recommended demand to determine the pre-recommended commodities that meet the recommended demand.

[0104] When it is determined that the current user is a first-band user and a first-value user, the pushed products are sorted from high to low according to the standard favorable comment rate, and the first-ranked pushed product is selected as the best product.

[0105] When the current user is determined to be a first-band user and a second-value user, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product:

[0106] Wi=α1×Hv+β1×Xi+θ1×Ri,

[0107] Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α1 represents the first calculation parameter when the current user is a first-band user and a second-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the current user is a first-band user and a second-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the current user is a first-band user and a second-value user.

[0108] When the current user is determined to be a user of the second frequency band and a user of the first value, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product:

[0109] Wi=α3×Hv+β3×Xi+θ3×Ri,

[0110] Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α3 represents the first calculation parameter when the current user is a second-band user and a first-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the current user is a second-band user and a first-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the current user is a second-band user and a first-value user.

[0111] When the current user is determined to be a second-band user and a second-value user, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product:

[0112] Wi=α2×Hv+β2×Xi+θ2×Ri,

[0113] Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α3 represents the first calculation parameter when the user is a second-band user and a second-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the user is a second-band user and a second-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the user is a second-band user and a second-value user.

[0114] The predetermined time T is calculated from the current time and pushed forward three months, six months, nine months, or twelve months.

[0115] It should be noted that due to Figure 4 Intelligent answer processing method and intelligent answer processing system executed by Figure 1 The smart response processing methods in the examples are roughly the same, so the description of the same parts is omitted.

[0116] Compared with the prior art, the present invention can effectively identify and determine user needs by extracting the key needs and preferences of the current user, input the demand classification model, obtain the demand classification result, and when the demand classification result is a recommended demand, the customer is answered by matching the preset reply; when the demand classification result is a recommended demand, the relevant commodities involved in the keywords are filtered layer by layer according to the keywords in the recommended demand, and the pre-pushed commodities that meet the recommended demand are determined, and the pre-pushed commodities are further comprehensively evaluated, and the pushed commodities to be pushed can be dynamically calculated and adjusted. The customer category of the current user is determined according to the consumption information data of the current user, and the determined pushed commodities are sorted and calculated, and the best commodity is determined as the response information, and the commodity of the customized user type can be provided to be pushed to the current user. This method can not only effectively identify the customer's search problem, but also provide customized, dynamically calculated and adjusted commodity recommendation services based on a full understanding of customer needs. At the same time, by integrating multi-channel data, a comprehensive analysis and real-time response to customer behavior can be achieved, so as to better meet the customized needs of customers and improve the overall service level of customer responses.

[0117] Example 3

[0118] Figure 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

[0119] like Figure 5 As shown, the electronic device is presented in the form of a general computing device. The processor may be one or more and work in coordination. The present invention does not exclude distributed processing, that is, the processor may be dispersed in different physical devices. The electronic device of the present invention is not limited to a single entity, but may also be the sum of multiple physical devices.

[0120] The memory stores a computer executable program, which is usually a machine-readable code. The computer-readable program can be executed by the processor to enable the electronic device to perform the method of the present invention, or at least part of the steps in the method.

[0121] The memory includes a volatile memory, such as a random access memory unit (RAM) and / or a cache memory unit, and may also be a non-volatile memory, such as a read-only memory unit (ROM).

[0122] Optionally, in this embodiment, the electronic device further includes an I / O interface, which is used for the electronic device to exchange data with an external device. The I / O interface can represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus structures.

[0123] It should be understood that Figure 6 The electronic device shown is only an example of the present invention, and the electronic device of the present invention may also include elements or components not shown in the above examples. For example, some electronic devices also include display units such as display screens, and some electronic devices also include human-computer interaction elements such as buttons, keyboards, etc. As long as the electronic device can execute the computer-readable program in the memory to implement the method of the present invention or at least part of the steps of the method, it can be considered as an electronic device covered by the present invention.

[0124] Through the above description of the implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by combining software with necessary hardware. Figure 6 As shown, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes a number of commands to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present invention.

[0125] The software product may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0126] The computer readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with a command execution system, device, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0127] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0128] The computer-readable medium carries one or more programs. When the one or more programs are executed by a device, the computer-readable medium implements the data interaction method of the present disclosure.

[0129] Those skilled in the art will appreciate that the above modules can be distributed in the device according to the description of the embodiment, or can be changed accordingly and only used in one or more devices different from the embodiment. The modules of the above embodiments can be combined into one module, or further divided into multiple sub-modules.

[0130] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several commands to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present invention.

[0131] It should be noted that the above detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present application belongs.

[0132] In the above detailed description, reference is made to the accompanying drawings, which form a part of this document. In the accompanying drawings, similar symbols typically identify similar components unless the context indicates otherwise. The illustrated embodiments described in the detailed description, drawings, and claims are not meant to be limiting. Other embodiments may be used, and other changes may be made, without departing from the spirit or scope of the subject matter presented herein.

[0133] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent answer processing method, characterized in that: include: Collect the information input by the current user, use natural language processing technology to parse the information input by the user, extract the key needs and preferences of the current user, perform semantic understanding through pre-trained language models, and determine user needs; The determined user needs are input into the demand classification model to obtain the demand classification result. When the demand classification result is a suggested need, the customer is answered by matching the preset reply; when the demand classification result is a recommended need, the related products involved in the keywords are filtered layer by layer according to the keywords in the recommended need to determine the pre-promoted products that meet the recommended needs, and the pre-promoted products are further comprehensively evaluated to determine the pushed products; Determine the customer category of the current user based on the consumption information data of the current user; According to the customer category corresponding to the current user, the determined pushed products are ranked and calculated, the best product is determined as the response information, and is pushed to the current user.

2. The intelligent response processing method according to claim 1, characterized in that: Further comprehensively evaluating the pre-promoted products and determining the promoted products includes performing the following steps: Step S201: determining s pre-promoted products, arbitrarily selecting one of the pre-promoted products as a target product, obtaining all evaluations of the target product within a specified time period, and dividing the specified time period into m interval time periods t; Step S202: selecting an interval time period from the m interval time periods t as a target interval time, and determining a favorable comment rate of the target interval time; Step S203: Repeat step S202 to determine the favorable comment rates of m interval time periods, calculate the average of the m favorable comment rates, further determine the deviation values ​​of the m favorable comment rates, and sort and delete them according to the determined deviation values ​​until the first ratio is less than the first predetermined value, and calculate the average of the remaining favorable comment rates as the standard favorable comment rate; Step S204: Select a specific number of top-ranked recommended products as the pushed products.

3. The intelligent response processing method according to claim 2, characterized in that: Further including: Use the following expression to calculate the deviation value U of m positive reviews: Where U represents the deviation value of the favorable comment rate in the target interval time period; H j represents the praise rate of the jth target interval, where j is a positive integer, specifically 1, 2, ..., m; H p represents the average of the favorable reviews in m target intervals; When the calculated deviation value U is greater than or equal to the first preset value, the highest ranked praise rate is deleted, and the deviation values ​​of the remaining praise rates except the deleted praise rate are recalculated until the first ratio is less than the first preset value and the calculated deviation value is less than the second preset value, and the remaining praise rates are calculated as the standard praise rates.

4. The intelligent response processing method according to claim 3, characterized in that: Each time the highest praise rate is deleted, the calculation of the first ratio is performed once; The first ratio is calculated using the following expression: Among them, B represents the ratio of the number SQ of deleted favorable reviews to m, that is, the first ratio; SQ represents the number of deleted favorable reviews; and m represents the number of target interval time periods.

5. The intelligent response processing method according to claim 2, characterized in that: Obtain the number g1 of all orders successfully purchased by the current user within the preset time T, and compare the number g1 with a third preset value. When the number g1 is greater than or equal to the third preset value, the user is determined to be a first-band customer; when the number g1 is less than the third preset value, the user is determined to be a second-band customer, wherein the preset time T is represented by a time period n days forward from the current moment; Calculate the total price Dz of all orders of the current user within the preset time T, and determine the average price Bz of different commodities with order quantity g1 through the formula Bz=Dz / g1. When the average price Bz is compared with the fourth preset value, when the average price Bz is greater than or equal to the fourth preset value, it is determined to be a first-value customer; when the average price Bz is less than the fourth preset value, it is determined to be a second-value customer.

6. The intelligent response processing method according to claim 2, characterized in that: When the demand classification result is a recommended demand, the related products involved in the keywords are filtered layer by layer according to the keywords in the recommended demand to determine the pre-recommended products that meet the recommended demand, including: When it is determined that the current user is a first-band user and a first-value user, the user is sorted from high to low according to the standard favorable comment rate, and the first-ranked push product is selected as the best product; When the current user is determined to be a first-band user and a second-value user, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product: Wi=α1×Hv+β1×Xi+θ1×Ri, Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α1 represents the first calculation parameter when the current user is a first-band user and a second-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the current user is a first-band user and a second-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the current user is a first-band user and a second-value user.

7. The intelligent response processing method according to claim 6, characterized in that: When the current user is determined to be a user of the second frequency band and a user of the first value, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product: Wi=α3×Hv+β3×Xi+θ3×Ri, Wherein, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard favorable comment rate of the i-th pre-promoted product; α3 represents the first calculation parameter when the current user is a second-band user and a first-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the second-band user is a first-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the second-band user is a first-value user; When the current user is determined to be a second-band user and a second-value user, the following expression is used to calculate the push coefficient of each of the specified number of pre-promoted products, and the pre-promoted product ranked first is selected as the best product: Wi=α2×Hv+β2×Xi+θ2×Ri, Among them, Wi represents the push coefficient of the i-th pre-promoted product; Hv represents the standard praise rate of the i-th pre-promoted product; α3 represents the first calculation parameter when the user is a second-band user and a second-value user; Xi represents the sales volume of the i-th pre-promoted product; β1 represents the second calculation parameter when the user is a second-band user and a second-value user; Ri represents the price of the i-th pre-promoted product; θ1 represents the third calculation parameter when the user is a second-band user and a second-value user.

8. The intelligent response processing method according to claim 5, characterized in that: Further including: The predetermined time T is calculated from the current time and pushed forward three months, six months, nine months, or twelve months.

9. An intelligent response processing system, characterized in that: The intelligent answer processing method according to any one of claims 1 to 8 is used, and the intelligent answer processing system comprises: The collection and processing module is used to collect the information input by the current user, parse the information input by the user using natural language processing technology, extract the key needs and preferences of the current user, and perform semantic understanding through the pre-trained language model to determine the user's needs; The first model calculation module inputs the determined user demand into the demand classification model to obtain the demand classification result. When the demand classification result is a suggested demand, the customer is answered by matching the preset reply; when the demand classification result is a recommended demand, the relevant products involved in the keywords are filtered layer by layer according to the keywords in the recommended demand to determine the pre-promoted products that meet the recommended demand, and the pre-promoted products are further comprehensively evaluated to determine the pushed products; A category determination module determines the customer category of the current user based on the consumption information data of the current user; The calculation and push module performs sorting calculation on the determined pushed products according to the customer category corresponding to the current user, determines the best product as the response information, and pushes it to the current user.

10. The intelligent response processing system according to claim 8, characterized in that: Further comprehensive evaluation of the pre-promoted products is conducted to determine the products to be promoted, including: Step S201: determining s pre-promoted products, arbitrarily selecting one of the pre-promoted products as a target product, obtaining all evaluations of the target product within a specified time period, and dividing the specified time period into m interval time periods t; Step S202: selecting an interval time period from the m interval time periods t as a target interval time, and determining a favorable comment rate of the target interval time; Step S203: Repeat step S202 to determine the favorable comment rates of m interval time periods, calculate the average of the m favorable comment rates, further determine the deviation values ​​of the m favorable comment rates, and sort and delete them according to the determined deviation values ​​until the first ratio is less than the first predetermined value, and calculate the average of the remaining favorable comment rates as the standard favorable comment rate; Step S204: Select a specific number of top-ranked recommended products as the pushed products.

Images