Service recommendation method and device, equipment, medium and program product

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a service recommendation method and device, equipment, a medium and a program product. The method comprises the following steps: acquiring a voice signal and / or an action signal of a user, and acquiring a user behavior classification according to the voice signal and / or the action signal; generating a cloud side recommendation service candidate set according to the user behavior classification through a cloud side pre-trained intelligent recommendation model, and generating a side recommendation service candidate set according to the user behavior classification through a pre-established user journey strategy library; and generating a target recommendation service according to the cloud side recommendation service candidate set and the side recommendation service candidate set, and pushing the target recommendation service to the user. According to the scheme of the embodiment, financial service recommendation can be realized at any time, and the user experience and the user stickiness can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a service recommendation method, device, equipment, medium and program product. Background Art

[0002] There are already a small number of electronic pet devices on the market that simulate the behavior of real pets. They can achieve rich interactions with users, such as making cute sounds when touched, which not only increases the fun of the interaction, but also makes the electronic pets more vivid and real.

[0003] Existing electronic pet devices do not have the function of recommending financial services to users. Users need to log in to mobile banking to handle business, and cannot provide users with financial services at any time, resulting in a poor user experience. Summary of the invention

[0004] The present invention provides a service recommendation method, device, equipment, medium and program product, which can realize financial service recommendation at any time and improve user experience and user stickiness.

[0005] According to one aspect of the present invention, a service recommendation method is provided, comprising:

[0006] Acquire a user's voice signal and / or action signal, and acquire a user behavior classification based on the voice signal and / or action signal;

[0007] Generate a candidate set of cloud-side recommended services based on the user behavior classification through a pre-trained intelligent recommendation model on the cloud side, and generate a candidate set of edge-side recommended services based on the user behavior classification through a pre-established user journey strategy library;

[0008] A target recommended service is generated according to the cloud-side recommended service candidate set and the edge-side recommended service candidate set, and the target recommended service is pushed to the user.

[0009] According to another aspect of the present invention, there is provided a service recommendation device, comprising:

[0010] A user behavior classification acquisition module is used to acquire a user's voice signal and / or action signal, and acquire a user behavior classification based on the voice signal and / or action signal;

[0011] A recommended service candidate set generation module is used to generate a cloud-side recommended service candidate set according to the user behavior classification through a cloud-side pre-trained intelligent recommendation model, and to generate an edge-side recommended service candidate set according to the user behavior classification through a pre-established user journey strategy library;

[0012] The target recommendation service generation module is used to generate a target recommendation service according to the cloud-side recommendation service candidate set and the edge-side recommendation service candidate set, and push the target recommendation service to the user.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the service recommendation method described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and the computer program is used to enable a processor to implement the service recommendation method described in any embodiment of the present invention when executed.

[0018] According to another aspect of the present invention, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the service recommendation method according to any embodiment of the present invention is implemented.

[0019] The technical solution of the embodiment of the present invention obtains the user's voice signal and / or action signal, and obtains the user behavior classification based on the voice signal and / or action signal; generates a cloud-side recommended service candidate set based on the user behavior classification through the cloud-side pre-trained intelligent recommendation model, and generates an edge-side recommended service candidate set based on the user behavior classification through the pre-established user journey strategy library; generates a target recommended service based on the cloud-side recommended service candidate set and the edge-side recommended service candidate set, and pushes the target recommended service to the user; determines the user behavior classification based on the voice signal and action signal, and generates a cloud-side recommended service candidate set and an edge-side recommended service candidate set based on the user behavior classification, and then determines the final target recommended service, which can realize financial service recommendations at any time and improve user experience and user stickiness.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 creative work.

[0022] Figure 1 is a flow chart of a service recommendation method provided according to Embodiment 1 of the present invention;

[0023] Figure 2 is a structural diagram of a service recommendation system provided according to Embodiment 1 of the present invention;

[0024] Figure 3 This is a schematic diagram of a network access process of an electronic pet intelligent terminal provided according to the first embodiment of the present invention;

[0025] Figure 4 is a flowchart of another service recommendation method provided according to the first embodiment of the present invention;

[0026] Figure 5 is a structural diagram of a service recommendation device provided according to Embodiment 2 of the present invention;

[0027] Figure 6 It is a structural diagram of an electronic device for implementing the service recommendation method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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 should fall within the scope of protection of the present invention.

[0029] It should be noted that the terms "first", "second", "target", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] Embodiment 1

[0031] Figure 1 A flowchart of a service recommendation method is provided for the first embodiment of the present invention. This embodiment is applicable to the case of recommending financial services to users through electronic pet devices. The method can be executed by a service recommendation device, which can be implemented in the form of hardware and / or software. Typically, the service recommendation device can be configured in an electronic device, such as a computer device or a server. Figure 1 As shown, the method includes:

[0032] S110: Acquire a voice signal and / or action signal of a user, and acquire a user behavior classification according to the voice signal and / or action signal.

[0033] In this embodiment, the structure of the service recommendation system can be as follows: Figure 2 As shown in the figure, the service recommendation system can adopt the overall architecture of "end + edge + cloud", where the end refers to the electronic pet smart terminal, the edge refers to the edge computing node, and the cloud refers to the bank's internal distributed computing system. Specifically, the distributed computing system is used to build a cloud-side intelligent recommendation model based on the customer historical behavior data in the data lake and customer tags based on the big data platform and machine learning platform, and intelligently sort the recommended content according to the customer's interest level (the recommended content can include investment and financial products, operating activities, financial information, partner marketing information, etc.) to provide intelligent services to the outside world. The edge computing node is used to collect and process the recent user interaction behaviors uploaded by the electronic pet smart terminal, and then generate edge-side recommendation services based on the user journey strategy. The electronic pet smart terminal is used to integrate the recommendation services on the cloud side and the edge side to obtain the final target recommendation service, and push the target recommendation service to the user in the form of voice or other means.

[0034] Among them, the network access process of the electronic pet intelligent terminal can be as follows: Figure 3 As shown. The electronic pet smart terminal has a unique device number. After accessing the Internet, it passes through the firewall and the virtual private network (VPN) to access the bank's internal network, and then passes through the bank server for identity authentication to establish a connection and conduct a session. Secondly, the communication message between the smart terminal and the bank server needs to go through encryption and decryption steps. The bank can only proceed to the next step of session establishment when it recognizes that the device number in the message is in a normal state. For some financial services, user identity authentication needs to be enhanced. The enhancement methods include fingerprint authentication, voice + custom questions + custom answers, and face recognition.

[0035] Moreover, for risk control purposes, risk control rules for smart terminals can be set. For example, when the same terminal customer authentication fails more than three times, the terminal status will be set to frozen. The terminal can be played with as a pet, but financial services cannot be used. Alternatively, each terminal can only be bound to three family sharing persons. If it is detected that the business request sent is from a non-bound customer, the transaction will be rejected.

[0036] The electronic pet intelligent terminal can be composed of a shell, a motion sensor, a microphone, a speaker, a processor, a memory, a control program, a connection module, and a power module. The shell is similar to the cute shape of a kitten or a piglet, and the outer skin and decorations are detachable and replaceable. The face has a touch screen for capturing the customer's touch behavior and fingerprint authentication. There is a camera at the bottom, which can be in a lifting form and is only used for face authentication and turned off when not in use. The motion sensor covers the entire shell and is used to capture the user's interactive actions such as caressing. The microphone is used to capture the user's voice commands. The speaker is used to broadcast information or interact with the user by voice. The processor can be a microprocessor or a central processing unit, which is responsible for user voice semantic analysis, controlling the pet's behavior and response, and managing the pet's status, and is responsible for processing network connections and application logic. The memory is used to store the operating system, application programs, and user data. The control program is the brain of the electronic pet, which contains the pet's behavior logic, user interface, and a dedicated secure connection with the bank. The control program can be pre-installed on the device or downloaded through updates. The connection module is used to connect to the Internet via Wi-Fi. The power module can be a power adapter or a built-in battery.

[0037] The gameplay of the electronic pet smart terminal may include (1) feeding, where users can feed the electronic pet through buttons or touch screens to simulate the basic needs of taking care of pets; (2) playing, where users can build relationships with electronic pets by playing, which may include simple clicks or more complex games; (3) training, where electronic pets may need training to learn new skills or behaviors, and users can train through specific instructions or interactions; (4) petting, with touch sensors, users can interact with pets through actual touch or simulated petting; (5) dressing up, where users can change different outfits or accessories for electronic pets to personalize their appearance; (6) socializing, where users can interact socially with other users' pets; (7) emotional feedback, where different emotional states are displayed, such as happiness, sadness, or anger, and users need to influence these emotions through interaction; (8) naming: users can customize the names of their pets; and (9) points. The more users interact with electronic pets, the more activity points they will get, and the points can be exchanged for benefits such as e-vouchers or accessories for dressing up pets.

[0038] The financial services that electronic pet smart terminals can provide include: (1) customer asset inquiries, such as inquiries about the account balance of a certain card, total customer assets, and various investment products held; (2) personalized information push, which can be customized to broadcast daily reports on the floating profit and loss of the customer's investment products, gold market conditions, stock market conditions, and market information, etc., and can also be pushed in response to customer instructions; (3) preferential activity inquiries, which inform customers of various bank card activities they can participate in through voice interaction, including preferential activities near the location; (4) proactive services, when customers enjoy petting cats When the customer receives a gift, the customer is informed that a certain product has earned * yuan today, thereby increasing the customer's happiness. When the customer's salary arrives in the account, the customer is informed of how much profit he or she has earned from using a certain product if he or she has retained his or her salary last month, and is recommended to continue purchasing the product. When the customer is feeding the gift, the customer is informed that a certain merchant is distributing coupons, and the gift is fed back to the customer, etc., and other proactive service scenarios are also available. (5) The piggy bank function is like an old-fashioned piggy bank. When you put money in it, you can hear the happy sound of coins colliding. It can also make a sound similar to "money" to tell the customer that he or she has won X yuan today (investment income). This brings more emotional value to the customer in terms of interactivity and experience.

[0039] It should be noted that the information collected is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data comply with the relevant laws, regulations and standards of relevant countries and regions, take necessary confidentiality measures, do not violate public order and good customs, and provide corresponding operation portals for users to choose to authorize or refuse. Moreover, all big data analysis based on personal information and the use of artificial intelligence technology for automated decision-making provide users with corresponding operation portals for users to choose to agree or refuse the automated decision-making results; if the user chooses to refuse, the expert decision-making process will be entered.

[0040] Specifically, the user's motion signal can be acquired through a camera, or the user's voice signal can be acquired through a microphone; then, the motion signal or voice signal can be feature extracted to obtain motion features or conversation features; finally, the motion features or conversation features can be input into a pre-trained user behavior recognition model to obtain the current user behavior classification. The behavior recognition model can be established based on methods such as neural networks and machine learning, and trained using corresponding labeled samples.

[0041] Optionally, obtaining user behavior classification based on the voice signal may include:

[0042] Preprocessing the speech signal to obtain a preprocessed speech signal, and extracting features from the preprocessed speech signal to obtain conversation features;

[0043] The session features are input into a pre-trained first behavior recognition model, and the user behavior classification output by the first behavior recognition model is obtained.

[0044] Among them, preprocessing may include operations such as denoising, enhancement and framing. In this embodiment, after obtaining the preprocessed voice signal, feature extraction may be performed on the preprocessed voice signal. Specifically, the preprocessed voice signal may be cleaned, segmented and vectorized to extract conversation features, for example, recording the total duration of the conversation and the number of question and answer rounds, determining whether financial keywords (such as financial management, funds, activities, discounts, discounts, etc.) appear, whether modal particles appear, extracting the frequency of keywords of the same type, and determining whether a service breakpoint occurs (such as a long pause by the customer, the customer saying "forget it", etc.).

[0045] Then, the session features can be input into the pre-trained first behavior recognition model to obtain the current user behavior classification, wherein the user behavior classification may include user demands and user attitudes.

[0046] The advantage of the above setting is that the user's real-time behavior can be identified based on the user's voice signal, and financial service recommendations can be implemented based on voice interaction.

[0047] Optionally, before inputting the conversation feature into the pre-trained first behavior recognition model, the following steps may also be included:

[0048] Obtain tone word samples and service word samples, and establish an initial behavior recognition model based on the neural network method;

[0049] The modal particle samples and the service word samples are used to perform supervised training on the initial behavior recognition model to obtain a trained first behavior recognition model.

[0050] Among them, modal particles can include a series of words that describe different tones, such as adjectives, adverbs, verb phrases, etc., which can express positive, negative, negative, affirmative and other emotional tendencies in the text. These modal particles can be used to judge the customer's attitude. Service words can be financial service keywords built based on the in-line software search vocabulary and customer service vocabulary, for example, they can include financial management, funds, activities, discounts, discounts, etc. In this embodiment, modal particles and service words can be collected and manually annotated to generate corresponding modal particle samples and service word samples.

[0051] Specifically, an initial behavior recognition model can be established according to a preset neural network method and initial model parameters; then, the initial behavior recognition model can be supervised trained using interjection samples, service word samples, and conversation data of existing human agent customer service; when it is detected that a preset training termination condition is met, for example, a specified number of training iterations is reached, a loss function value is less than or equal to a preset loss threshold, etc., a first behavior recognition model that has been trained is obtained. This embodiment does not specifically limit the type of neural network method.

[0052] The advantage of the above setting is that it can improve the accuracy of user behavior classification, and thus improve the accuracy of service recommendation.

[0053] Optionally, obtaining user behavior classification based on the action signal may include:

[0054] Extracting features from the motion signal to obtain motion intensity features;

[0055] Inputting the action intensity feature into a pre-trained second behavior recognition model, and obtaining a user behavior classification output by the second behavior recognition model;

[0056] Among them, the second behavior recognition model is established based on a deep neural network method.

[0057] In this embodiment, in order to identify customer interactive behaviors and analyze customer preferences for products, customers' interactive behaviors such as stroking electronic pets can be collected, and the interactive behavior data can be used for behavioral recognition using a deep neural network model, thereby using the interactive behavior data for marketing activities to make customers sticky to the product.

[0058] Specifically, the built-in sensor of the electronic pet intelligent terminal can be used to collect the electrical signals of interactive behaviors such as stroking, patting, and stroking, and the electrical signals can be converted into digital signals after amplification, filtering, and other processing to obtain action signals. Then, the action signal can be converted into a vector, and further converted into a one-hot encoding for storage. The storage structure can be Act1 [duration, starting position, end position, action intensity, speed, acceleration]. Furthermore, the action signal can be feature extracted to obtain the action intensity feature.

[0059] Among them, the action intensity feature extraction requires the use of Fourier transform to further process the action signal. For example, the action signal can be regarded as a time series function f(t), where t represents time and f(t) represents the position of the center of force at time t. Through Fourier transform, the time series function f(t) can be decomposed into a superposition of a series of sine (or cosine) fluctuations. Each sine fluctuation corresponds to a specific frequency and amplitude, which reflect the movement intensity of the electronic pet at different frequencies.

[0060] Secondly, a second behavior recognition model can be pre-trained. During product testing, a deep neural network algorithm is used to train various interactive behavior data. During the training process, the model performance can be improved by adjusting the model parameters. After production application, the action intensity features corresponding to the real-time collected action signals can be input into the second behavior recognition model to obtain user behavior classification. Among them, the user behavior classification can include the degree of customer preference.

[0061] The advantage of the above setting is that it can expand the applicable scenarios of the service recommendation method and can make financial service recommendations based on customer interaction behaviors.

[0062] S120. Generate a cloud-side recommended service candidate set based on the user behavior classification through the cloud-side pre-trained intelligent recommendation model, and generate an edge-side recommended service candidate set based on the user behavior classification through the pre-established user journey strategy library.

[0063] Specifically, the user behavior classification can be input into the intelligent recommendation model pre-deployed on the cloud side, and the cloud-side recommended services and corresponding probability values ​​output by the intelligent recommendation model can be obtained; then, the cloud-side recommended services can be composed of candidate sets. At the same time, based on the user journey strategy library pre-deployed on the edge side, the target user journey that matches the user behavior classification can be screened, and the edge-side recommended services can be determined based on the target user journey, and then the edge-side recommended services can be composed of candidate sets of edge-side recommended services.

[0064] Optionally, a pre-established user journey strategy library is used to generate a candidate set of side recommendation services according to the user behavior classification, which may include:

[0065] Obtaining a historical behavior classification corresponding to the user, and obtaining a current user journey based on the historical behavior classification and the user behavior classification;

[0066] Acquire multiple user journey strategies in the user journey strategy library, and perform matching detection on the current user journey and each of the user journey strategies to obtain a target user journey strategy that successfully matches the current user journey;

[0067] According to the target user journey strategy, a candidate set of side recommendation services is generated.

[0068] Among them, the user journey can be a behavior chain composed of the user's financial service acquisition data combined with the interactive behavior of the electronic pet smart terminal; for example, journey one: the first type of interactive behavior-financial information acquisition-customer asset query-product recommendation-product repurchase-second type of interactive behavior; journey two: push consumption promotion activities based on location information-multiple rounds of conversations-receive coupons through mobile software-consumption-third type of interactive behavior.

[0069] Specifically, the user's historical behavior classification and the current user behavior classification can be used to form the current user journey; then, the current user journey can be matched with each user journey strategy in the user journey strategy library to obtain a target user journey strategy that successfully matches the current user journey; finally, each target user journey strategy can form a side recommendation service candidate set. For example, the current user journey is behavior A-behavior B, the first user journey strategy is behavior A-behavior B-behavior C, and the second user journey strategy is behavior B-behavior A-behavior C. At this time, the current user journey successfully matches the first user journey strategy, and the first user journey strategy can be determined as the target user journey strategy.

[0070] The advantage of the above setting is that it can achieve accurate prediction of the user's next action, which can further improve the accuracy of service recommendations.

[0071] S130: Generate a target recommended service according to the cloud-side recommended service candidate set and the edge-side recommended service candidate set, and push the target recommended service to the user.

[0072] Specifically, the intersection of the cloud-side recommended service candidate set and the edge-side recommended service candidate set can be calculated to obtain the recommended services included in both sets as the target recommended services. Alternatively, the edge-side recommended services in the edge-side recommended service candidate set can be preferentially used as the target recommended services. Then, the target recommended services can be pushed to the user in the form of voice broadcast or the like.

[0073] Optionally, generating a target recommended service according to the cloud-side recommended service candidate set and the edge-side recommended service candidate set may include:

[0074] Determine whether the edge-side recommended service candidate set is empty, and if so, obtain each cloud-side recommended service and a corresponding probability value according to the cloud-side recommended service candidate set;

[0075] A target cloud-side recommended service corresponding to the maximum probability value is obtained from among the cloud-side recommended services to serve as the target recommended service.

[0076] In a specific example, first, it is determined whether the edge-side recommended service candidate set is empty. If not, the edge-side recommended service in the edge-side recommended service candidate set can be directly used as the target recommended service. If so, multiple cloud-side recommended services and the probability value corresponding to each cloud-side recommended service can be extracted from the cloud-side recommended service candidate set; further, the target cloud-side recommended service corresponding to the maximum probability value can be screened out from all cloud-side recommended services to be used as the target recommended service.

[0077] The advantage of the above setting is that it can improve the robustness of service recommendations, ensure that at least one recommended service can be given, and improve user experience.

[0078] The technical solution of the embodiment of the present invention obtains the user's voice signal and / or action signal, and obtains the user behavior classification based on the voice signal and / or action signal; generates a cloud-side recommended service candidate set based on the user behavior classification through the cloud-side pre-trained intelligent recommendation model, and generates an edge-side recommended service candidate set based on the user behavior classification through the pre-established user journey strategy library; generates a target recommended service based on the cloud-side recommended service candidate set and the edge-side recommended service candidate set, and pushes the target recommended service to the user; determines the user behavior classification based on the voice signal and action signal, and generates a cloud-side recommended service candidate set and an edge-side recommended service candidate set based on the user behavior classification, and then determines the final target recommended service, which can realize financial service recommendations at any time and improve user experience and user stickiness.

[0079] In a specific implementation of this embodiment, the process of the service recommendation method can be as follows: Figure 4 As shown. The end side is used to collect data and transmit the collected real-time user behavior to the edge side and cloud side, as well as to provide the final target recommendation service by integrating the recommendation services of the cloud side and the edge side. The cloud side is used to store user behavior data (including all user attributes and all user history), and can train intelligent recommendation models based on the stored data, and use the trained intelligent recommendation model to provide cloud-side recommendation services. The edge side is used to store short-term user behavior data, and to establish and update the user journey strategy library based on the user behavior data, and to provide edge recommendation services based on the user's real-time behavior based on the user journey strategy library.

[0080] The solution of this embodiment can stimulate customer enthusiasm, obtain richer customer data, and enhance customer experience by using instant intelligent digital capabilities and proactive service capabilities by adding a financial service recommendation function in the electronic pet smart terminal.

[0081] Embodiment 2

[0082] Figure 5This is a schematic diagram of the structure of a service recommendation device provided in Embodiment 2 of the present invention. Figure 5 As shown, the device includes: a user behavior classification acquisition module 210, a recommended service candidate set generation module 220 and a target recommended service generation module 230; wherein,

[0083] A user behavior classification acquisition module 210 is used to acquire a user's voice signal and / or action signal, and acquire a user behavior classification based on the voice signal and / or action signal;

[0084] A recommended service candidate set generation module 220 is used to generate a cloud-side recommended service candidate set according to the user behavior classification through a cloud-side pre-trained intelligent recommendation model, and to generate an edge-side recommended service candidate set according to the user behavior classification through a pre-established user journey strategy library;

[0085] The target recommendation service generation module 230 is used to generate a target recommendation service according to the cloud-side recommendation service candidate set and the edge-side recommendation service candidate set, and push the target recommendation service to the user.

[0086] The technical solution of the embodiment of the present invention obtains the user's voice signal and / or action signal, and obtains the user behavior classification based on the voice signal and / or action signal; generates a cloud-side recommended service candidate set based on the user behavior classification through the cloud-side pre-trained intelligent recommendation model, and generates an edge-side recommended service candidate set based on the user behavior classification through the pre-established user journey strategy library; generates a target recommended service based on the cloud-side recommended service candidate set and the edge-side recommended service candidate set, and pushes the target recommended service to the user; determines the user behavior classification based on the voice signal and action signal, and generates a cloud-side recommended service candidate set and an edge-side recommended service candidate set based on the user behavior classification, and then determines the final target recommended service, which can realize financial service recommendations at any time and improve user experience and user stickiness.

[0087] Optionally, the user behavior classification acquisition module 210 is specifically used to preprocess the voice signal to obtain a preprocessed voice signal, and perform feature extraction on the preprocessed voice signal to obtain a conversation feature;

[0088] The session features are input into a pre-trained first behavior recognition model, and the user behavior classification output by the first behavior recognition model is obtained.

[0089] Optionally, the user behavior classification acquisition module 210 is further used to acquire modal word samples and service word samples, and establish an initial behavior recognition model according to a neural network method;

[0090] The modal particle samples and the service word samples are used to perform supervised training on the initial behavior recognition model to obtain a trained first behavior recognition model.

[0091] Optionally, the user behavior classification acquisition module 210 is specifically used to extract features from the action signal to obtain action intensity features;

[0092] Inputting the action intensity feature into a pre-trained second behavior recognition model, and obtaining a user behavior classification output by the second behavior recognition model;

[0093] Among them, the second behavior recognition model is established based on a deep neural network method.

[0094] Optionally, the recommended service candidate set generation module 220 is specifically used to obtain the historical behavior classification corresponding to the user, and obtain the current user journey according to the historical behavior classification and the user behavior classification;

[0095] Acquire multiple user journey strategies in the user journey strategy library, and perform matching detection on the current user journey and each of the user journey strategies to obtain a target user journey strategy that successfully matches the current user journey;

[0096] According to the target user journey strategy, a candidate set of side recommendation services is generated.

[0097] Optionally, the target recommended service generation module 230 is specifically configured to determine whether the edge-side recommended service candidate set is empty, and if so, to obtain each cloud-side recommended service and a corresponding probability value according to the cloud-side recommended service candidate set;

[0098] A target cloud-side recommended service corresponding to the maximum probability value is obtained from among the cloud-side recommended services to serve as the target recommended service.

[0099] The service recommendation device provided in the embodiment of the present invention can execute the service recommendation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0100] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0101] Embodiment 3

[0102] Figure 6A schematic diagram of the structure of an electronic device 30 that can be used to implement an embodiment of the present invention is shown. The electronic device 30 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 30 can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0103] like Figure 6 As shown, the electronic device 30 includes at least one processor 31, and a memory connected to the at least one processor 31, such as a read-only memory (ROM) 32, a random access memory (RAM) 33, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 31 can perform various appropriate actions and processes according to the computer program stored in the read-only memory 32 or the computer program loaded from the storage unit 38 to the random access memory 33. In the RAM 33, various programs and data required for the operation of the electronic device 30 can also be stored. The processor 31, the ROM 32 and the RAM 33 are connected to each other through a bus 34. The input / output (I / O) interface 35 is also connected to the bus 34.

[0104] A number of components in the electronic device 30 are connected to the I / O interface 35, including: an input unit 36, such as a keyboard, a mouse, etc.; an output unit 37, such as various types of displays, speakers, etc.; a storage unit 38, such as a disk, an optical disk, etc.; and a communication unit 39, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 39 allows the electronic device 30 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0105] The processor 31 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 31 include, but are not limited to, a central processing unit, a graphics processing unit, various special artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any appropriate processors, controllers, microcontrollers, etc. The processor 31 executes the various methods and processes described above, such as the service recommendation method.

[0106] In some embodiments, the service recommendation method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 38. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 30 via the ROM 32 and / or the communication unit 39. When the computer program is loaded into the RAM 33 and executed by the processor 31, one or more steps of the service recommendation method described above may be performed. Alternatively, in other embodiments, the processor 31 may be configured to perform the service recommendation method in any other appropriate manner (e.g., by means of firmware).

[0107] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays, application specific integrated circuits, application specific standard products, systems on a chip, load programmable logic devices, computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0108] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0109] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0110] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device 30 having: a display device (e.g., a cathode ray tube or a liquid crystal display) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the electronic device 30. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0111] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local area networks, wide area networks, blockchain networks, and the Internet.

[0112] A computing system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other. The server may be a cloud server.

[0113] This embodiment may also include a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the service recommendation method provided by any embodiment of the present invention.

[0114] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0115] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A service recommendation method, characterized in that: include: Acquire a user's voice signal and / or action signal, and acquire a user behavior classification based on the voice signal and / or action signal; Generate a candidate set of cloud-side recommended services based on the user behavior classification through a pre-trained intelligent recommendation model on the cloud side, and generate a candidate set of edge-side recommended services based on the user behavior classification through a pre-established user journey strategy library; A target recommended service is generated according to the cloud-side recommended service candidate set and the edge-side recommended service candidate set, and the target recommended service is pushed to the user.

2. The method according to claim 1, characterized in that According to the voice signal, obtain the user behavior classification, including: Preprocessing the speech signal to obtain a preprocessed speech signal, and extracting features from the preprocessed speech signal to obtain conversation features; The session features are input into a pre-trained first behavior recognition model, and the user behavior classification output by the first behavior recognition model is obtained.

3. The method according to claim 2, characterized in that Before inputting the conversation feature into the pre-trained first behavior recognition model, the method further includes: Obtain tone word samples and service word samples, and establish an initial behavior recognition model based on the neural network method; The modal particle samples and the service word samples are used to perform supervised training on the initial behavior recognition model to obtain a trained first behavior recognition model.

4. The method according to claim 1, characterized in that: Based on the action signal, obtain the user behavior classification, including: Extracting features from the motion signal to obtain motion intensity features; Inputting the action intensity feature into a pre-trained second behavior recognition model, and obtaining a user behavior classification output by the second behavior recognition model; Among them, the second behavior recognition model is established based on a deep neural network method.

5. The method according to claim 1, characterized in that Through the pre-established user journey strategy library, according to the user behavior classification, a candidate set of side recommendation services is generated, including: Obtaining a historical behavior classification corresponding to the user, and obtaining a current user journey based on the historical behavior classification and the user behavior classification; Acquire multiple user journey strategies in the user journey strategy library, and perform matching detection on the current user journey and each of the user journey strategies to obtain a target user journey strategy that successfully matches the current user journey; According to the target user journey strategy, a candidate set of side recommendation services is generated.

6. The method according to claim 1, characterized in that Generating a target recommended service according to the cloud-side recommended service candidate set and the edge-side recommended service candidate set includes: Determine whether the edge-side recommended service candidate set is empty, and if so, obtain each cloud-side recommended service and a corresponding probability value according to the cloud-side recommended service candidate set; A target cloud-side recommended service corresponding to the maximum probability value is obtained from among the cloud-side recommended services to serve as the target recommended service.

7. A service recommendation device, characterized in that: include: A user behavior classification acquisition module is used to acquire a user's voice signal and / or action signal, and acquire a user behavior classification based on the voice signal and / or action signal; A recommended service candidate set generation module is used to generate a cloud-side recommended service candidate set according to the user behavior classification through a cloud-side pre-trained intelligent recommendation model, and to generate an edge-side recommended service candidate set according to the user behavior classification through a pre-established user journey strategy library; The target recommendation service generation module is used to generate a target recommendation service according to the cloud-side recommendation service candidate set and the edge-side recommendation service candidate set, and push the target recommendation service to the user.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor, and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the service recommendation method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is used to enable a processor to implement the service recommendation method according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The invention comprises a computer program, which implements the service recommendation method according to any one of claims 1 to 6 when executed by a processor.

Images