Information recommendation method and device based on large model, electronic equipment and storage medium

Abstract

The invention provides an information recommendation method and device based on a large model, electronic equipment and a storage medium, and relates to the field of artificial intelligence such as deep learning, large models and intelligent recommendation. The method comprises the following steps: acquiring feature information of different dimensions of an object to be recommended; according to the feature information, initial recommendation information is generated through a recommendation model, the initial recommendation information comprises at least one target recommendation object selected from all candidate recommendation objects, and the recommendation model is a large model; and generating target recommendation information according to the initial recommendation information, and recommending the target recommendation information to the to-be-recommended object.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, particularly to the fields of deep learning, large models, and intelligent recommendation, and more particularly to information recommendation methods, devices, electronic devices, and storage media based on large models. Background Art

[0002] At present, many platforms will recommend information to users. For example, telecom operator platforms or Internet service provider platforms can recommend network service products to users to improve users' efficiency in obtaining information. Summary of the Invention

[0003] The present disclosure provides a large model-based information recommendation method, device, electronic device, and storage medium.

[0004] An information recommendation method based on a large model, comprising:

[0005] Obtain feature information of different dimensions of the object to be recommended;

[0006] generating initial recommendation information using a recommendation model based on the feature information, wherein the initial recommendation information includes: at least one target recommendation object selected from each candidate recommendation object, and the recommendation model is a large model;

[0007] Generate target recommendation information based on the initial recommendation information, and recommend the target recommendation information to the object to be recommended.

[0008] An information recommendation device based on a large model, comprising: a feature acquisition module, an initial recommendation module, and a target recommendation module;

[0009] The feature acquisition module is used to obtain feature information of different dimensions of the object to be recommended;

[0010] The initial recommendation module is configured to generate initial recommendation information using a recommendation model based on the feature information, wherein the initial recommendation information includes at least one target recommendation object selected from each candidate recommendation object, and the recommendation model is a large model;

[0011] The target recommendation module is used to generate target recommendation information according to the initial recommendation information, and recommend the target recommendation information to the object to be recommended.

[0012] An electronic device, comprising:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the method as described above.

[0016] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the method as described above.

[0017] A computer program product comprises a computer program / instruction, which implements the above method when executed by a processor.

[0018] 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 disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0020] Figure 1 This is a flowchart of the first embodiment of the information recommendation method based on a large model described in the present disclosure;

[0021] Figure 2 This is a flowchart of the second embodiment of the large model-based information recommendation method described in the present disclosure;

[0022] Figure 3 Schematic diagram of the structure of an embodiment 300 of an information recommendation device based on a large model according to the present disclosure;

[0023] Figure 4 FIG. 4 is a schematic block diagram of an electronic device 400 that can be used to implement an embodiment of the present disclosure. DETAILED DESCRIPTION

[0024] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0025] Furthermore, it should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " as used herein generally indicates that the associated objects are in an "or" relationship.

[0026] Figure 1 This is a flow chart of the first embodiment of the information recommendation method based on the big model described in this disclosure. Figure 1 As shown, the following specific implementation methods are included.

[0027] In step 101, feature information of different dimensions of an object to be recommended is obtained.

[0028] In step 102, initial recommendation information is generated using a recommendation model based on the feature information. The initial recommendation information includes at least one target recommendation object selected from each candidate recommendation object. The recommendation model is a large model.

[0029] In step 103, target recommendation information is generated based on the initial recommendation information, and the target recommendation information is recommended to the object to be recommended.

[0030] Taking online service products as an example, traditional recommendation methods typically only perform rough grouping of users based on simple user tags. These recommendations then serve as a guide for recommending online service products to users based on the grouping results. For example, if a user is identified as belonging to group A, the corresponding product packages are recommended. Alternatively, collaborative filtering algorithms can be used to recommend online service products that similar users have purchased. However, these methods fail to provide personalized recommendations, resulting in poor recommendation accuracy.

[0031] By adopting the solution described in the above method embodiment, information can be recommended to the target based on the feature information of the target in different dimensions, thereby improving the pertinence of the recommendation and enabling the recommendation results to meet the multi-dimensional personalized needs of the target, that is, improving the accuracy of the recommendation results. Moreover, the recommendation model can be used to determine the initial recommendation information, and then the final target recommendation information can be generated based on the initial recommendation information. Accordingly, the powerful reasoning ability of the recommendation model can further improve the accuracy of the recommendation results. The target can be the target user to be recommended.

[0032] In some embodiments of the present disclosure, candidate recommendation objects may include network service products. Accordingly, obtaining feature information of different dimensions of the recommended object may include obtaining account feature information of the network account corresponding to the recommended object and obtaining network problem description information of the recommended object. Specifically, feature information of the recommended object in the account dimension and network usage pain point dimension may be obtained separately.

[0033] The content included in the account feature information and the network problem description information can be determined according to actual needs.

[0034] For example, account feature information may include the number of members using the account (such as 3 people), broadband speed, the number of smart devices using the account (such as smart phones, tablets, smart speakers, etc.), and historical repair key information, etc. Among them, the historical repair key information can be generated based on the historical repair records of the recommended object obtained within the most recent first predetermined time period. For example, the most recent first predetermined time period may refer to the most recent three months. Assuming that a total of 8 historical repair records are obtained, these historical repair records can be processed by key information extraction to obtain historical repair key information. The historical repair key information may include the number of repairs and the reason for the repairs, etc. In addition, there is no restriction on the method by which the recommended object reports the repair. For example, the repair can be reported by telephone (voice) or text. Each time a repair is reported, a historical repair record can be generated to record the problems reported by the recommended object.

[0035] The network problem description information refers to the demand pain point information of the object to be recommended, which can be obtained based on the dynamic scene data of the object to be recommended that is parsed in real time. For example, if it is determined through base station positioning or other methods that the problem of video freeze at night occurs, then it can be recorded. If it is determined that the problem of abnormal device offline alarm occurs, it can also be recorded. Accordingly, based on these problems recorded within the most recent second predetermined time period, the network problem description information of the object to be recommended can be analyzed and generated. The values ​​of the second predetermined time period and the first predetermined time period can be the same or different. For example, the most recent second predetermined time period can refer to the most recent month. The network problem description information may include the frequency of network freezes and the duration of device offline, among which the device offline duration can refer to the average abnormal offline duration.

[0036] It can be seen that the above-mentioned feature information describes the recommended objects from different dimensions. Accordingly, combining these feature information to recommend information to the recommended objects can improve the accuracy of the recommendation results and dynamically match the network usage pain points of the recommended objects, that is, dynamically match the scenario requirements of the recommended objects, thereby further improving the accuracy of the recommendation results.

[0037] In addition, it should be noted that the characteristic information and other information in the embodiments described in this disclosure are not targeted at a specific user and are not intended to reflect the personal information of a specific user. Moreover, the characteristic information can be obtained through various public, legal and compliant means, such as obtaining it from the user with the user's knowledge and authorization. In the technical solutions of this disclosure, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved are all in compliance with relevant laws and regulations and do not violate public order and good morals.

[0038] Based on the feature information, a recommendation model may be used to generate initial recommendation information, and the initial recommendation information may include: at least one target recommendation object selected from each candidate recommendation object.

[0039] In some embodiments of the present disclosure, feature information of different dimensions may be fused with a first prompt template to obtain a first target prompt, and then the first target prompt may be input into a recommendation model to obtain output initial recommendation information.

[0040] The recommendation model can be obtained by fine-tuning and training a predetermined generative artificial intelligence model, and the predetermined generative artificial intelligence model can be a 4th generation generative pre-trained transformer (GPT-4, Generative Pre-trained Transformer 4) model, etc.

[0041] In addition, fusing feature information of different dimensions with the first prompt word template may include: filling the feature information of different dimensions into corresponding slots of the first prompt word template respectively, where slots can be understood as areas to be filled in the template, obtaining a pre-generated first prompt word template, and filling the obtained feature information of different dimensions into corresponding slots of the first prompt word template respectively, thereby obtaining the first target prompt word.

[0042] For example, the first target prompt word may include the following:

[0043] prompt=f"""

[0044] Account characteristic information: {number of members}_{broadband speed}_{number of smart devices}_{historical repair report key information};

[0045] Network problem description: {network freeze frequency}_{device offline duration}""".

[0046] It can be seen that by adopting the above processing method, the required first target prompt word can be generated simply and efficiently through operations such as slot filling.

[0047] In some embodiments of the present disclosure, recommendation requirement information corresponding to the object to be recommended can also be obtained, and then the feature information of different dimensions, the recommendation requirement information and the second prompt word template can be integrated to obtain the second target prompt word, and then the second target prompt word can be input into the recommendation model to obtain the output initial recommendation information.

[0048] Compared with the first prompt word template, the second prompt word template may include not only slots corresponding to feature information of different dimensions, but also slots corresponding to recommendation requirement information. Accordingly, the feature information of different dimensions and the recommendation requirement information may be filled into the corresponding slots of the second prompt word template respectively, thereby obtaining the second target prompt word.

[0049] For example, the second target prompt word may include the following:

[0050] prompt=f"""

[0051] Account characteristic information: {number of members}_{broadband speed}_{number of smart devices}_{historical repair report key information};

[0052] Network problem description: {network freeze frequency}_{device offline duration};

[0053] Recommended requirements: Combination of ≤ 3 products, priority network upgrade + maintenance services, budget ≤ user's historical average consumption * 1.2".

[0054] The specific content included in the recommendation requirements can be determined based on actual needs and by the platform. For example, the aforementioned "Combination ≤ 3 products" means that the number of selected network service products must be less than or equal to 3, "Priority network upgrade + maintenance service" indicates the preferred type of network service product, and "Budget ≤ User's historical average spending * 1.2" means that the budget for the selected network service products must be less than or equal to the user's historical average spending * 1.2.

[0055] The recommendation requirement information can be used to further constrain the process of selecting target recommendation objects by the recommendation model, so that the selected target recommendation objects can better match the actual needs of the recommended objects, thereby further improving the accuracy of the recommendation results.

[0056] After obtaining the first target prompt word or the second target prompt word, it can be input into the recommendation model to obtain initial recommendation information output by the recommendation model.

[0057] For example, the initial recommendation information may include the following:

[0058] Fiber to the Room (FTTR) all-optical networking (main product) + smart home detection service (value-added package) + video membership quarterly card (hook benefit).

[0059] After obtaining the initial recommendation information, target recommendation information can be further generated based on the initial recommendation information. In some embodiments of the present disclosure, the initial recommendation information can be subjected to a rationality check. In response to determining that the check passes, target recommendation information can be generated based on the initial recommendation information, and the target recommendation information can be recommended to the target object.

[0060] In some embodiments of the present disclosure, the method of performing a rationality check on the initial recommendation information may include: counting the number M of target recommendation objects included in the initial recommendation information, in response to determining that the value of M is 1, determining that the check of the initial recommendation information has passed, in response to determining that the value of M is greater than 1, constructing object pairs based on the initial recommendation information, each object pair includes two different target recommendation objects, and each target recommendation object appears in M-1 object pairs, performing a rationality check on each object pair, and in response to determining that each object pair has passed the check, determining that the check of the initial recommendation information has passed.

[0061] That is to say, if the initial recommendation information only includes one target recommendation object, then it can be directly determined that the initial recommendation information has passed the inspection. If the initial recommendation information includes multiple target recommendation objects, then the object pairs can be constructed based on the initial recommendation information. For example, assuming that the initial recommendation information includes 3 target recommendation objects, for the sake of convenience, they are respectively called target recommendation object a, target recommendation object b and target recommendation object c. Then the following 3 object pairs can be constructed: target recommendation object a-target recommendation object b, target recommendation object a-target recommendation object c, target recommendation object b-target recommendation object c. Then, the rationality of the 3 object pairs can be checked separately. If it is determined that all 3 object pairs have passed the inspection, it can be determined that the inspection of the initial recommendation information has passed. Otherwise, as long as one object pair fails the inspection, it can be determined that the inspection of the initial recommendation information has failed.

[0062] By performing a rationality check, the rationality of the obtained initial recommendation information can be improved, and the rationality of the target recommendation information can be improved accordingly.

[0063] In some embodiments of the present disclosure, the following processing may be performed for each object pair: identifying whether there is a conflict between the two target recommended objects in the object pair; in response to the existence of a conflict, determining that the inspection of the object pair has failed; and in response to the absence of a conflict, determining that the inspection of the object pair has passed.

[0064] For example, the mutual exclusion rule engine can be called to determine whether there is a conflict between the two target recommendation objects in the object pair. The conflict usually refers to a conflict in business logic, such as the conflict between the two target recommendation objects "contract phone discount" and "number portability subsidy".

[0065] Accordingly, in some embodiments of the present disclosure, in response to determining that there are object pairs that fail the inspection, the initial recommendation information can be updated, and the rationality check can be performed on the updated initial recommendation information. In response to determining that the inspection passes, target recommendation information can be generated based on the updated initial recommendation information.

[0066] Through the above processing, it is possible to avoid recommending conflicting network service products to the recommended object, thereby further improving the accuracy of the recommendation results.

[0067] In some embodiments of the present disclosure, updating the initial recommendation information may include regenerating the initial recommendation information using a recommendation model. Other updating methods may also be employed if desired, such as performing the following processing on each pair of objects that failed the inspection: deleting one of the target recommended objects in the pair from the initial recommendation information.

[0068] For example, among the above-mentioned target recommendation objects a, b and c, if there is a conflict between target recommendation objects b and c, then the target recommendation object b or the target recommendation object c can be deleted, or the recommendation model can be used to directly regenerate the initial recommendation information. The specific method to be adopted can be determined according to actual needs, which is very flexible and convenient. Preferably, the latter method can be adopted to make the number of target recommendation objects and other information more in line with the recommendation requirements.

[0069] After obtaining the initial recommendation information that passes the rationality check, the target recommendation information can be generated based on the initial recommendation information. In some embodiments of the present disclosure, the initial recommendation information and the feature information of different dimensions can be combined to generate a recommendation text, and the recommendation text can be determined as the final target recommendation information.

[0070] For example, the initial recommendation information and the feature information of different dimensions may be input into the result generation model to obtain the output recommendation text.

[0071] Through the above processing, the target recommended object in the initial recommendation information can be combined with the network problem description information of the object to be recommended, thereby improving the matching degree between the recommendation text and the target recommended object, making the recommendation text more scenario-based and logical, and enhancing the user experience of the object to be recommended, and improving the acceptance of the object to be recommended.

[0072] For example, the recommended text may be: 3 devices are detected to be stuck, it is recommended to upgrade FTTR to allocate exclusive frequency bands...

[0073] Afterwards, the recommendation text, i.e., the target recommendation information, can be recommended to the recommended object, and the specific recommendation method and recommendation timing are not limited.

[0074] Combined with the above introduction, Figure 2 This is a flow chart of the second embodiment of the information recommendation method based on a large model described in this disclosure. Figure 2 As shown, the following specific implementation methods are included.

[0075] In step 201, feature information of different dimensions of the object to be recommended is obtained.

[0076] For example, candidate recommendation objects may include: network service products, account feature information of the network account corresponding to the recommended object may be obtained, and network problem description information of the recommended object may be obtained.

[0077] In step 202, the recommendation requirement information corresponding to the object to be recommended is obtained, and the feature information of different dimensions, the recommendation requirement information and the second prompt word template are integrated to obtain the second target prompt word.

[0078] In step 203, the second target prompt word is input into the recommendation model to obtain output initial recommendation information.

[0079] In step 204 , a rationality check is performed on the initial recommendation information to determine whether the check passes. If so, step 205 is executed; otherwise, step 203 is repeated.

[0080] For example, the number M of target recommendation objects included in the initial recommendation information can be counted. In response to determining that the value of M is 1, it can be determined that the check of the initial recommendation information has passed. In response to determining that the value of M is greater than 1, object pairs can be constructed based on the initial recommendation information, each object pair includes two different target recommendation objects, and each target recommendation object appears in M-1 object pairs respectively, and the rationality of each object pair can be checked separately. In response to determining that each object pair has passed the check, it can be determined that the check of the initial recommendation information has passed.

[0081] Among them, for each object pair, the following processing can be performed respectively: identify whether there is a conflict between the two target recommended objects in the object pair, and in response to the existence of a conflict, determine that the inspection of the object pair has failed; in response to the absence of a conflict, determine that the inspection of the object pair has passed.

[0082] In step 205 , the initial recommendation information and the feature information of different dimensions are combined to generate a recommendation text, and the recommendation text is determined as the target recommendation information.

[0083] In step 206, the target recommendation information is recommended to the object to be recommended, and then the process ends.

[0084] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should know that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present disclosure. In addition, for parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0085] The above is an introduction to the method embodiment. The following is a further explanation of the solution disclosed in the present disclosure through an apparatus embodiment.

[0086] Figure 3 Schematic diagram of the structure of the information recommendation device embodiment 300 based on the large model described in this disclosure. Figure 3 As shown, it includes: a feature acquisition module 301, an initial recommendation module 302 and a target recommendation module 303.

[0087] The feature acquisition module 301 is used to acquire feature information of different dimensions of the object to be recommended.

[0088] The initial recommendation module 302 is configured to generate initial recommendation information using a recommendation model according to the feature information. The initial recommendation information includes at least one target recommendation object selected from the candidate recommendation objects. The recommendation model is a large model.

[0089] The target recommendation module 303 is configured to generate target recommendation information based on the initial recommendation information and recommend the target recommendation information to the object to be recommended.

[0090] By adopting the scheme described in the above-mentioned device embodiment, information can be recommended to the object to be recommended based on the feature information of different dimensions of the object to be recommended, thereby improving the targetedness of the recommendation and enabling the recommendation results to meet the multi-dimensional personalized needs of the object to be recommended, that is, improving the accuracy of the recommendation results. Moreover, the recommendation model can be used to determine the initial recommendation information, and then the final target recommendation information required can be generated based on the initial recommendation information. Accordingly, with the help of the powerful reasoning ability of the recommendation model, the accuracy of the recommendation results can be further improved.

[0091] In some embodiments of the present disclosure, candidate recommendation objects may include: network service products. Accordingly, the way in which the feature acquisition module 301 obtains feature information of different dimensions of the object to be recommended may include: obtaining account feature information of the network account corresponding to the object to be recommended, and obtaining network problem description information of the object to be recommended.

[0092] According to the feature information, the initial recommendation module 302 may generate initial recommendation information using a recommendation model. The initial recommendation information may include: at least one target recommendation object selected from each candidate recommendation object.

[0093] In some embodiments of the present disclosure, the initial recommendation module 302 may fuse feature information of different dimensions with the first prompt word template to obtain a first target prompt word, and then input the first target prompt word into a recommendation model to obtain output initial recommendation information.

[0094] Alternatively, in some embodiments of the present disclosure, the initial recommendation module 302 may obtain the recommendation requirement information corresponding to the object to be recommended, and then may fuse the feature information of different dimensions, the recommendation requirement information and the second prompt word template to obtain the second target prompt word, and then may input the second target prompt word into the recommendation model to obtain the output initial recommendation information.

[0095] After obtaining the initial recommendation information, the target recommendation module 303 may further generate target recommendation information based on the initial recommendation information. In some embodiments of the present disclosure, the target recommendation module 303 may perform a rationality check on the initial recommendation information. In response to determining that the rationality check passes, the target recommendation information may be generated based on the initial recommendation information, and the target recommendation information may be recommended to the target object.

[0096] In some embodiments of the present disclosure, the method in which the target recommendation module 303 performs a rationality check on the initial recommendation information may include: counting the number M of target recommendation objects included in the initial recommendation information, in response to determining that the value of M is 1, determining that the check of the initial recommendation information has passed, in response to determining that the value of M is greater than 1, constructing object pairs based on the initial recommendation information, each object pair includes two different target recommendation objects, and each target recommendation object appears in M-1 object pairs, performing a rationality check on each object pair, and in response to determining that each object pair has passed the check, determining that the check of the initial recommendation information has passed.

[0097] In some embodiments of the present disclosure, for each object pair, the target recommendation module 303 may perform the following processing respectively: identify whether there is a conflict between the two target recommendation objects in the object pair; in response to the existence of a conflict, determine that the inspection of the object pair has failed; in response to the absence of a conflict, determine that the inspection of the object pair has passed.

[0098] In some embodiments of the present disclosure, the target recommendation module 303 may update the initial recommendation information in response to determining that there are object pairs that fail the inspection, and may perform a rationality check on the updated initial recommendation information. In response to determining that the inspection passes, target recommendation information may be generated based on the updated initial recommendation information.

[0099] In some embodiments of the present disclosure, the target recommendation module 303 may update the initial recommendation information by regenerating the initial recommendation information using a recommendation model.

[0100] After obtaining the initial recommendation information that passes the rationality check, the target recommendation module 303 may generate target recommendation information based on the initial recommendation information. In some embodiments of the present disclosure, the target recommendation module 303 may combine the initial recommendation information and the feature information of different dimensions to generate a recommendation text, and may determine the recommendation text as the final target recommendation information required.

[0101] Figure 3 The specific working process of the device embodiment shown can refer to the relevant description in the aforementioned method embodiment and will not be repeated here.

[0102] The solutions described in this disclosure can be applied to the field of artificial intelligence, particularly in areas such as deep learning, large models, and intelligent recommendation. Artificial intelligence is the study of how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). It involves both hardware and software technologies. Artificial intelligence hardware technologies generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing. Artificial intelligence software technologies mainly include computer vision, speech recognition, natural language processing, machine learning / deep learning, big data processing, and knowledge graph technology.

[0103] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0104] Figure 4 A schematic block diagram of an electronic device 400 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, 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 disclosure described and / or claimed herein.

[0105] like Figure 4As shown, the electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 into a random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the electronic device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0106] Multiple components in the electronic device 400 are connected to the I / O interface 405, including an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a magnetic disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the electronic device 400 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0107] The computing unit 401 can be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI, Artificial Intelligence) computing chips, various computing units that run machine learning model algorithms, digital signal processors (DSP, Digital Signal Processing), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 401 performs the various methods and processes described above, such as the methods described in the present disclosure. For example, in some embodiments, the methods described in the present disclosure can be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the method described in the present disclosure can be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to execute the method described in the present disclosure in any other appropriate manner (for example, by means of firmware).

[0108] 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 (FPGAs), application specific integrated circuits (ASICs), application specific standard parts (ASSPs), system on chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can 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.

[0109] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0110] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can 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 (EPROM), a 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 of the foregoing.

[0111] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0112] The systems and techniques described herein can be implemented in a computing system that includes back-end 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 front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with embodiments of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0113] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0114] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0115] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A large model-based information recommendation method, comprising: Obtain feature information of different dimensions of the object to be recommended; generating initial recommendation information using a recommendation model based on the feature information, wherein the initial recommendation information includes: at least one target recommendation object selected from each candidate recommendation object, and the recommendation model is a large model; Generate target recommendation information based on the initial recommendation information, and recommend the target recommendation information to the object to be recommended.

2. The method according to claim 1, wherein The candidate recommendation objects include: network service products; The obtaining of characteristic information of different dimensions of the object to be recommended includes: obtaining account characteristic information of the network account corresponding to the object to be recommended, and obtaining network problem description information of the object to be recommended.

3. The method according to claim 1, wherein Generating initial recommendation information using a recommendation model according to the feature information includes: fusing the feature information of different dimensions with the first prompt word template to obtain a first target prompt word; The first target prompt word is input into the recommendation model to obtain the output initial recommendation information.

4. The method according to claim 1, wherein Generating initial recommendation information using a recommendation model according to the feature information includes: Obtaining recommendation requirement information corresponding to the object to be recommended; fusing the feature information of different dimensions, the recommendation requirement information, and the second prompt word template to obtain a second target prompt word; The second target prompt word is input into the recommendation model to obtain the output initial recommendation information.

5. The method according to claim 1, wherein Generating target recommendation information according to the initial recommendation information includes: Performing a rationality check on the initial recommendation information; In response to determining that the check is passed, the target recommendation information is generated according to the initial recommendation information.

6. The method according to claim 5, wherein: The rationality check of the initial recommendation information includes: Counting the number M of target recommendation objects included in the initial recommendation information; In response to determining that the value of M is 1, determining that the check of the initial recommendation information passes; In response to determining that the value of M is greater than 1, object pairs are constructed according to the initial recommendation information, each object pair includes two different target recommendation objects, and each target recommendation object appears in M-1 object pairs respectively, and rationality checks are performed on each object pair respectively. In response to determining that each object pair has passed the check, it is determined that the check of the initial recommendation information has passed.

7. The method according to claim 6, wherein: The rationality check of each object pair includes: For each object pair, the following processing is performed respectively: identifying whether there is a conflict between the two target recommended objects in the object pair; in response to the existence of a conflict, determining that the inspection of the object pair has failed; in response to the absence of a conflict, determining that the inspection of the object pair has passed.

8. The method according to claim 7, further comprising: In response to determining that there is an object pair that fails the inspection, updating the initial recommendation information; A rationality check is performed on the updated initial recommendation information, and in response to determining that the check passes, the target recommendation information is generated according to the updated initial recommendation information.

9. The method according to claim 8, wherein The updating of the initial recommendation information includes: regenerating the initial recommendation information using the recommendation model.

10. The method according to claim 1, wherein Generating target recommendation information according to the initial recommendation information includes: A recommendation text is generated by combining the initial recommendation information and the characteristic information, and the recommendation text is determined as the target recommendation information.

11. An information recommendation device based on a large model, comprising: Feature acquisition module, initial recommendation module and target recommendation module; The feature acquisition module is used to obtain feature information of different dimensions of the object to be recommended; The initial recommendation module is configured to generate initial recommendation information using a recommendation model based on the feature information, wherein the initial recommendation information includes at least one target recommendation object selected from each candidate recommendation object, and the recommendation model is a large model; The target recommendation module is used to generate target recommendation information according to the initial recommendation information, and recommend the target recommendation information to the object to be recommended.

12. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 10.

13. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 10.

14. A computer program product comprising a computer program / instructions, which implement the method according to any one of claims 1 to 10 when executed by a processor.

Images