Recommendation method and device, electronic equipment and storage medium

Abstract

The invention provides a recommendation method and device, electronic equipment and a storage medium, belongs to the technical field of artificial intelligence, and particularly relates to the technical field of deep learning, cloud computing and intelligent recommendation. According to the specific implementation scheme, in response to an individual recommendation request for a target article, preference information, about the target article, of the target object is obtained, and the preference information is used for representing respective preference degrees of the target object for multiple quality categories of the target article; determining respective expected quality information of a plurality of individuals based on respective multi-modal features of the plurality of individuals of the target article included in the to-be-recommended region; and determining an individual to be recommended from the plurality of individuals based on the preference information and the respective expected quality information of the plurality of individuals.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, in particular to the field of deep learning, cloud computing, and intelligent recommendation technology, and more specifically to a recommendation method, device, electronic device, and storage medium. Background Art

[0002] Non-standardized items, i.e., items whose individual products do not have uniform quality, usually have the characteristics of diversified varieties and non-uniform criteria for distinguishing each variety. When purchasing non-standardized items, users need to rely on personal experience to complete the purchase of these non-standardized items. If the user's personal experience is insufficient, they may purchase unsatisfactory items, thus affecting the user's shopping experience. Summary of the invention

[0003] The present disclosure provides a recommendation method, device, electronic device and storage medium.

[0004] According to one aspect of the present disclosure, a recommendation method is provided, comprising: in response to an individual recommendation request for a target item, obtaining preference information of the target object regarding the target item, the preference information being used to characterize the target object's degree of preference for each of multiple quality categories of the target item; determining expected quality information of each of the multiple individuals of the target item included in an area to be recommended based on multimodal features of each of the multiple individuals; and determining an individual to be recommended from the multiple individuals based on the preference information and the expected quality information of each of the multiple individuals.

[0005] According to another aspect of the present disclosure, a recommendation device is provided, comprising: an acquisition module for acquiring, in response to an individual recommendation request for a target item, preference information of a target object regarding the target item, the preference information being used to characterize the target object's degree of preference for each of multiple quality categories of the target item; a first determination module for determining expected quality information of each of multiple individuals of the target item included in a to-be-recommended area based on their respective multimodal features; and a second determination module for determining an individual to be recommended from the multiple individuals based on the preference information and the expected quality information of the multiple individuals.

[0006] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0007] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method as described above.

[0008] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program implements the method described above when executed by a processor.

[0009] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0011] Figure 1 An exemplary system architecture to which the recommendation method and apparatus according to an embodiment of the present disclosure can be applied is schematically shown.

[0012] Figure 2 The flowchart of the recommendation method according to the embodiment of the present disclosure is schematically shown.

[0013] Figure 3 The flowchart of the method for determining preference information according to an embodiment of the present disclosure is schematically shown.

[0014] Figure 4 The figure schematically shows a method for predicting expected quality information of an individual according to an embodiment of the present disclosure.

[0015] Figure 5 The following schematically shows a workflow diagram of a shopping guide device according to an embodiment of the present disclosure.

[0016] Figure 6 The block diagram of a recommendation device according to an embodiment of the present disclosure is schematically shown.

[0017] Figure 7 A schematic block diagram of an example electronic device that can be used to implement embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0018] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art 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.

[0019] For non-standardized items, take agricultural products such as vegetables and fruits as an example. In modern life, many people have higher and higher requirements for the quality of agricultural products, but due to lack of experience or knowledge, they often have difficulty making the right choices when shopping. For example, when selecting vegetables, users may not know how to distinguish between tender and old, or when choosing fruits, they may not be able to judge their internal quality by appearance. The traditional way of shopping relies on personal experience, and the accumulation of such experience takes time, which may lead to wrong choices or waste of time for novices. Moreover, when consumers buy agricultural products in the market, they can usually only judge the quality of agricultural products by their own immature experience reserves, which is unscientific and inefficient.

[0020] In view of this, embodiments of the present disclosure provide a recommendation method, apparatus, electronic device and storage medium, the recommendation method comprising: in response to an individual recommendation request for a target item, obtaining preference information of a target object regarding a target item, the preference information being used to characterize the target object's degree of preference for each of multiple quality categories of the target item; determining expected quality information of each of the multiple individuals based on multimodal features of each of the multiple individuals of the target item included in an area to be recommended; and determining an individual to be recommended from the multiple individuals based on the preference information and the expected quality information of each of the multiple individuals.

[0021] Figure 1 An exemplary system architecture to which the recommendation method and apparatus according to an embodiment of the present disclosure can be applied is schematically shown.

[0022] It should be noted that Figure 1 The examples shown are only examples of system architectures to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but do not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios. For example, in another embodiment, the exemplary system architecture to which the recommendation method and apparatus can be applied may include a terminal device, but the terminal device may implement the recommendation method and apparatus provided by the embodiments of the present disclosure without interacting with the server.

[0023] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0024] The target object can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the terminal devices 101, 102, 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients and / or social platform software, etc. (only as examples).

[0025] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0026] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for the content browsed by the target object using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as the target object's request, and feed back the processing results (such as the web page, information, or data obtained or generated according to the target object's request) to the terminal device.

[0027] It should be noted that the recommendation method provided in the embodiment of the present disclosure can generally be executed by the terminal device 101, 102, or 103. Accordingly, the recommendation device provided in the embodiment of the present disclosure can also be set in the terminal device 101, 102, or 103.

[0028] Alternatively, the recommendation method provided in the embodiment of the present disclosure may also be generally performed by the server 105. Accordingly, the recommendation device provided in the embodiment of the present disclosure may generally be arranged in the server 105. The recommendation method provided in the embodiment of the present disclosure may also be performed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the recommendation device provided in the embodiment of the present disclosure may also be arranged in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0029] For example, when the target object reads an e-book online, the terminal devices 101, 102, and 103 can obtain the target content in the e-book pointed by the target object, and then send the obtained target content to the server 105, which analyzes the target content to determine the characteristic information of the target content; predicts the content that the target object is interested in based on the characteristic information of the target content; and extracts the content that the target object is interested in. Alternatively, a server or server cluster that can communicate with the terminal devices 101, 102, 103 and / or the server 105 analyzes the target content and finally extracts the content that the target object is interested in.

[0030] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0031] In the technical solution disclosed herein, the collection, storage, use, processing, transmission, provision, disclosure and application of the target object’s personal information are in compliance with the relevant laws and regulations, necessary confidentiality measures are taken, and do not violate public order and good morals.

[0032] In the technical solution of the present disclosure, the authorization or consent of the target object is obtained before obtaining or collecting the personal information of the target object.

[0033] Figure 2 The flowchart of the recommendation method according to the embodiment of the present disclosure is schematically shown.

[0034] like Figure 2 As shown, the method includes operations S210 to S230.

[0035] In operation S210 , in response to an individual recommendation request for a target item, preference information of a target object regarding the target item is acquired.

[0036] In operation S220 , expected quality information of each of the plurality of target items included in the to-be-recommended area is determined based on the multimodal features of each of the plurality of target items.

[0037] In operation S230, an individual to be recommended is determined from among the plurality of individuals based on the preference information and the expected quality information of each of the plurality of individuals.

[0038] The target item can be represented as an item of a variety. Taking the target item as an agricultural product as an example, the variety of the target item can be represented as a specific variety under a category of agricultural products, such as a specific variety of Chinese cabbage among vegetables, a specific variety of apples among fruits, etc. Each individual of the item of this variety can have different qualities. The quality of the individual can be represented as the edible quality, appearance quality, etc. of the individual. The edible quality can include, for example, the coarseness, crispness, fiber content of the meat, the sourness, sweetness, fragrance, etc. of the taste, and the degree of maturity. The appearance quality can include, for example, the integrity, freshness, color, smoothness, etc. of the individual. The meaning represented by the individual of the target item can be determined based on the purchase criteria of the target item during the item transaction process. For example, for items that exist in the form of monomers, they can be purchased in the form of "number" during the item transaction process, and the individual of this type of item can be represented as a monomer of an item, such as a watermelon, a cabbage, etc. For another example, for items that exist in the form of multiple units, during the transaction process, they can be purchased in the form of "number of strings" or "number of bundles". Then the individuals of this type of items can be represented as multiple units of one item, such as a bunch of grapes, a bundle of enoki mushrooms, etc.

[0039] The target object's preference information about the target item can be used to characterize the target object's preference for each of the multiple quality categories of the target item. For example, for edible quality in quality, the preference information can be expressed as the target object prefers a target item with a sourer taste among multiple items, or the preference information can be expressed as the target object prefers a target item with a crispier texture among multiple items.

[0040] The individual recommendation request may be triggered and generated when the target object performs a shopping behavior for the target item. The target object may be represented as a consumer who shops offline or online.

[0041] Optionally, in an offline shopping scenario, when it is detected that a consumer faces an exhibition area of ​​a target item and the consumer stays within a specific range of the exhibition area for a specific time, the consumer can be identified as a potential consumer of the target item, and an individual recommendation request for the target item can be generated. The specific range can be, for example, an annular range or a rectangular range centered on the exhibition area, and the size of the range can be set according to the specific application scenario, which is not limited here. The specific duration can be set according to the specific application scenario, for example, to 5 seconds, 10 seconds, etc., which is not limited here.

[0042] Optionally, in an online shopping scenario, when it is detected that a consumer is searching for a target item, or when the consumer stays on the product page of the target item for a specific time, the consumer can be identified as a potential consumer of the target item, and an individual recommendation request for the target item can be generated. The specific time can be set according to the specific type of the target item, for example, 5 seconds, 10 seconds, etc., which is not limited here.

[0043] Multiple individuals of the target item can be placed in the area to be recommended. In an offline shopping scenario, the area to be recommended can be represented as the exhibition area of ​​the target item in an offline shopping mall, and the multiple individuals of the target item can be represented as the multiple individuals placed in the exhibition area. In an online shopping scenario, the area to be recommended can be represented as a warehouse or storehouse where the online merchant stores the target item, and accordingly, the multiple individuals of the target item can be represented as the multiple individuals stored in the warehouse or storehouse.

[0044] The multimodal features of an individual can be extracted from the multimodal information of the individual. The multimodal information of the individual may include information reflecting the appearance of the individual, information reflecting the internal structure of the individual, etc., which is not limited here. The information reflecting the appearance of the individual can be determined by image processing and other methods, which is not limited here. The information reflecting the internal structure of the individual can be determined by physical measurement, spectral detection and other methods, which is not limited here.

[0045] The expected quality information of an individual can be represented as a vector or matrix with the same dimension as the preference information, that is, the expected quality information and preference information of an individual can be represented as a point in the vector space respectively. By means of matrix calculation, a point close to the point represented by the preference information can be determined from the points represented by the expected quality information of multiple individuals, thereby determining the individual to be recommended.

[0046] The number of individuals to be recommended can be determined based on the needs of the target object. For example, in online or offline shopping scenarios, the target object can input the number, weight, or other value information such as the transaction amount of the target items required through the front-end display device, microphone, and other interactive ports. The interactive port can be configured at the edge node of the cloud service system. After obtaining this information, the cloud service system can determine the number of individuals to be recommended in combination with the average individual weight, unit price attributes, and other information of the target items.

[0047] According to an embodiment of the present disclosure, multimodal information of individuals of the target item is collected respectively to obtain multimodal features. Based on the multimodal features, expected quality information of the individual can be obtained, and then the expected quality information of the individual is compared with the preference information of the target object to determine whether the individual is an individual that needs to be recommended. When an individual recommendation request is received, the individual to be recommended can be determined based on the recommendation result of each individual. The target object can directly complete the purchase of the target item based on the individual to be recommended, which can effectively reduce the demand for the target object's personal experience when purchasing the target item, thereby effectively improving the target object's consumption experience.

[0048] Reference below Figure 3~Figure 5 , combined with specific embodiments Figure 2 The method shown is further explained.

[0049] Optionally, in online or offline shopping scenarios, with the permission of the target object, the target object's behavior information can be recorded, and the behavior information may include the target object's shopping record, the individual quality of the purchased items, the target object's evaluation of the shopping record, etc. The target object's behavior information can be stored in a log or database as historical behavior information related to the item. Data mining based on the behavior information can obtain portrait information for the item, and the portrait information can be expressed as the target object's preference for each quality category of the item obtained through data statistical analysis and regression processing, that is, based on the behavior information, the target object's preference information about the target item can be obtained.

[0050] According to an embodiment of the present disclosure, obtaining the preference information of the target object about the target item may include the following operations:

[0051] Acquire historical behavior information related to the target item from a log related to the target object; and determine preference information based on the historical behavior information.

[0052] The behavior information related to the target item in the log may include multiple data records. By integrating the multiple data records, the historical behavior information can be obtained. The content contained in each data record is not limited here. For example, it may include the name of the item purchased in a purchase behavior, the specific type of the item, the individual quality information of the purchased item, the target object's evaluation of this purchase behavior, etc.

[0053] Determining preference information based on historical behavior information can be achieved using classification models based on various model architectures. For example, the classification model can be a time series classification model based on a long short-term memory network, and the behavior data sequence represented by the historical behavior information can be sequentially input into the classification model. The classification model can output the scores of each quality category, and the high or low score can indicate the target object's preference for the quality category of the target item. By analyzing the time series data sequence using a time series model, the evolution of the target object's preference for the target item can be clarified, thereby obtaining more accurate preference information.

[0054] Optionally, in an online shopping or offline shopping scenario, the target object may be a new user, or the target object does not wish to retain behavioral information, that is, the target object's behavioral information is not retained in the log or database. At this time, you can interact with the target object through the interaction port, and obtain the target object's preference information about the target item based on the interaction result.

[0055] According to an embodiment of the present disclosure, obtaining the preference information of the target object about the target item may include the following operations:

[0056] The interaction information is sent through the interaction port with the target object so that the target object inputs feedback information based on the interaction information; and, in response to receiving the feedback information, preference information is determined based on the feedback information.

[0057] The interactive port may include a front-end display device, a microphone, etc., and its specific implementation form is related to the carrier device of the interactive port, or may be related to the personalized settings of the carrier device by the target object, which is not limited here. Taking the interactive port as a front-end display device as an example, the interactive information may be a questionnaire containing multiple multiple-choice questions, which may be generated by the system itself, or may be randomly extracted from a question bank by the system, or may be pre-set by a technician, which is not limited here. The target object may answer each multiple-choice question through an information input device on the carrier device, such as a mouse, keyboard, physical button, virtual button, touch screen, etc., and provide feedback on the interactive information to input feedback information through the interactive port. For another example, taking the interactive port as a microphone, the carrier device may output voice information through the equipped audio, and the language information may represent the audio of the question that the target object needs to answer, and the target object may reply to the audio of the question within the voice response range of the carrier device, and the audio of the reply result may be captured by the microphone of the carrier device to form feedback information, so as to input feedback information through the interactive port.

[0058] Based on the obtained feedback information, the conversion of feedback information into preference information can be completed through data statistics, model reasoning, etc., and the specific implementation method is not limited here.

[0059] According to the embodiments of the present disclosure, based on the human-computer interaction method, feedback information of the target object is obtained through the interactive port, and preference information is generated based on the feedback information, so as to realize personalized recommendation for the target item and improve the shopping experience of the target object.

[0060] Optionally, when there is behavior information in a database or log, feedback information may be obtained through human-computer interaction with the target object, and preference information may be determined in combination with historical behavior information and feedback information, which is not limited here.

[0061] Figure 3 The flowchart of the method for determining preference information according to an embodiment of the present disclosure is schematically shown.

[0062] like Figure 3 As shown, the carrier device 301 may be configured with interactive ports such as a display panel and / or a microphone. The carrier device 301 may send interactive information 303 to the target object 302. The target object 302 may form feedback information 304 based on the interactive information 303 and return the feedback information 304 to the carrier device 301.

[0063] The carrier device 301 may be connected to a cloud service system 305 in the cloud. The carrier device 301 may return feedback information 304 to the cloud service system 305, and the cloud service system 305 may process the feedback information 304 to obtain preference information 306.

[0064] Alternatively, the cloud may also be configured with a database 307 , which may record the target object's behavior records regarding the target item, based on which historical behavior information 308 may be obtained, and the cloud service system 305 may process the historical behavior information 308 to obtain preference information 306 .

[0065] Alternatively, the cloud service system 305 may perform information processing in combination with the feedback information 304 and the historical behavior information 308 to obtain the preference information 306 , which is not limited herein.

[0066] Alternatively, in the case where the target object's behavior information regarding the target item does not exist in the database or log, if the target object's behavior information regarding another item exists in the database or log, the preference information may be determined based on the target object's behavior information regarding the other item. The other item may be another item of the same category or having similar characteristics as the target item.

[0067] When determining the target object's preference information about the target item, the expected quality information of each of the multiple individuals of the target item may be determined synchronously or asynchronously based on the multimodal features of each of the multiple individuals of the target item.

[0068] According to an embodiment of the present disclosure, the multimodal features may include appearance features, physical features, and internal features. The appearance features, physical features, and internal features of an individual may be collected separately to obtain the multimodal features of the individual.

[0069] Appearance features are features used to represent the shape, size, color, surface defects, etc. of an individual. Optionally, the appearance features of an individual can be determined based on an image of the individual. The method for acquiring the image of the individual is not limited here. For example, the image of the individual can be obtained by controlling a camera device to shoot the individual. The camera device can be, for example, a camera, an industrial camera, etc., which are not limited here. Optionally, in order to more accurately determine the appearance features of an individual, images of the individual can be collected based on multiple angles to obtain multiple images of the individual, and the appearance features of the individual can be determined based on the multiple images of the individual.

[0070] According to an embodiment of the present disclosure, after acquiring an image of an individual, determining the appearance features of the individual based on the image of the individual may include the following operations:

[0071] Perform contour extraction on the image to obtain shape information of the individual; obtain color information of the individual based on the pixel values ​​of each of the multiple pixels included in the target area of ​​the image, wherein the target area is the area in the image corresponding to the individual; determine the abnormal sub-area contained in the target area based on the pixel values ​​of each of the multiple pixels included in the target area; obtain surface defect information of the individual based on the position and number of the abnormal sub-areas; and perform feature extraction based on the shape information, color information and surface defect information to obtain appearance features.

[0072] The contour extraction of the image may include operations such as binarization, dilation, and erosion of the image, and the specific implementation method thereof is not limited here. The shape information of the individual may be used to represent the appearance, size, and the like of the individual.

[0073] The color information of an individual may include the color and gloss of the individual. The image collected for the individual may be a color image, that is, based on the pixel value of each pixel in the target area of ​​the image, the overall color of the individual can be determined, and based on the size of the area representing the reflective area in the target area of ​​the image, the gloss level of the individual can be determined. In the area representing the reflective area, the pixel values ​​of all pixels may be pixel values ​​representing white. Alternatively, the image collected for the individual may be a grayscale image, and the overall color and gloss level of the individual can be determined based on the grayscale value of each pixel in the target area of ​​the image. Specifically, the larger the grayscale value of each pixel in the target area, the brighter the color of the object, and correspondingly, the smaller the grayscale value of each pixel in the target area, the darker the color of the object.

[0074] The surface defect information of an individual can indicate the size, location and number of defective areas on the surface of the individual due to factors such as insect damage, impact, corruption, etc. The defective area can be represented as an abnormal sub-area on the image. The abnormal sub-area can be represented as an area where the pixel values ​​of the pixels inside it are obviously different from the pixel values ​​of the pixels in the surrounding area. For example, if there are black spots on an apple, the pixel values ​​of the pixels in the area where the black spots are located in the collected apple image are close to the pixel values ​​representing black, and there will be obvious differences from the pixel values ​​of the pixels in the area representing the normal surface of the apple. In this case, the area where the black spots are located can be regarded as an abnormal sub-area.

[0075] The collected color information, shape information and surface defect information can all be represented by vectors or matrices. The color information, shape information and surface defect information can be spliced, and feature extraction can be performed based on the information obtained after splicing to obtain appearance features. The splicing method is not limited here. When the color information, shape information and surface defect information can all be represented by vectors, the color information, shape information and surface defect information can be spliced ​​into a vector in a wide dimension, or the color information, shape information and surface defect information can be spliced ​​into a matrix in a high dimension.

[0076] Physical characteristics are characteristics used to indicate the hardness, softness, elasticity, etc. of an individual. Optionally, the physical characteristics of an individual can be determined based on the tactile feedback information of the individual. The method for obtaining the tactile feedback information of the individual is not limited here. For example, a manipulator can be controlled to grasp the individual, and the manipulator can be configured with multiple pressure sensors, and the manipulator can be controlled to slightly squeeze the individual during the process of grasping the individual, so that the surface of the individual produces a fixed slight deformation. The degree of the slight deformation can be less than a specific deformation threshold, and the specific deformation threshold can be expressed as a critical value of the degree of deformation that will affect the quality of the individual.

[0077] According to an embodiment of the present disclosure, after acquiring the tactile feedback information of an individual, the physical characteristics of the individual may be determined based on the tactile feedback information of the individual, which may include the following operations:

[0078] When controlling the plurality of pressure sensors to move a preset distance, obtaining pressure values ​​recorded by each of the plurality of pressure sensors; and determining the physical characteristics based on the pressure values ​​recorded by each of the plurality of pressure sensors.

[0079] The preset distance may be expressed as the distance that the pressure sensor moves when the individual is squeezed to produce a fixed slight deformation.

[0080] Different individuals have different degrees of hardness and elasticity. When the same deformation occurs, the pressure values ​​collected by the sensor at each collection may be different. That is, based on the difference in the pressure values, the difference in hardness, elasticity, etc. of different individuals can be determined. Based on the pressure values ​​recorded by the multiple pressure sensors when squeezing the individual, the physical characteristics of the individual can be determined.

[0081] Internal features may be features used to represent the structure, defects, etc. inside an individual. Optionally, the internal features of an individual may be determined based on the feedback audio of the individual. The method for obtaining the feedback audio of the individual is not limited herein. For example, the feedback audio of the individual may be collected when the target object makes a preset action to the individual. The preset action may include actions such as knocking and shaking. The setting of the specific preset action may be related to the type of the target object. For example, if the target object may be a coconut, the feedback audio of the individual may be collected when the target object shakes the individual of the target object, which is not limited herein. For another example, the manipulator may be controlled to make a preset action to the individual and synchronously collect the feedback audio of the individual, which is not limited herein.

[0082] According to an embodiment of the present disclosure, after obtaining the feedback audio of an individual, determining the internal features of the individual based on the feedback audio of the individual may include the following operations:

[0083] Performing spectrum analysis on the feedback audio to obtain the frequency and amplitude of the feedback audio; and determining the internal characteristics based on the frequency and amplitude of the feedback audio.

[0084] The feedback audio can be analyzed by using sound spectrum analysis technology, combined with short-time Fourier transform and Mel spectrum coefficient feature extraction methods, to extract the frequency and amplitude of the feedback audio, i.e., the sound features. The relationship between the sound features and the internal qualities of an individual can be identified by training a sound model, thereby determining the internal qualities of the individual, i.e., the internal features.

[0085] According to an embodiment of the present disclosure, determining expected quality information of each of the plurality of target items included in the to-be-recommended area based on the multimodal features of each of the plurality of target items may include the following operations:

[0086] The appearance features, physical features and internal features of the individual are fused to obtain the fused features of the individual; and the fused features of the individual are input into a quality recognition model corresponding to the target item to obtain the expected quality information of the individual.

[0087] Optionally, multimodal data fusion technology can be used to fuse appearance features determined based on image data, physical features determined based on tactile data, and internal features determined based on audio data through feature splicing or attention mechanism to obtain fused features.

[0088] The quality identification model can be a fusion model based on multiple deep learning frameworks. By inputting the fusion features into the quality identification model, the expected quality information of the individual can be predicted.

[0089] Optionally, other methods may be used to make decisions on expected quality information based on fusion features. The other methods may include random forests, support vector machines, etc., which are not limited here.

[0090] Figure 4 The figure schematically shows a method for predicting expected quality information of an individual according to an embodiment of the present disclosure.

[0091] like Figure 4 As shown, the method may include operations S401 to S408.

[0092] In operation S401 , a camera device is controlled to photograph an individual to obtain an image of the individual.

[0093] In operation S402, the image of the individual is processed using a computer vision algorithm to obtain the appearance features of the individual.

[0094] In operation S403, the hardness and elasticity of the individual are measured using a pressure sensor to obtain tactile data.

[0095] In operation S404, the tactile data is analyzed to obtain individual physical characteristics.

[0096] In operation S405 , audio data is captured through a microphone.

[0097] In operation S406, an audio analysis algorithm is used to determine the internal structure of the individual to obtain the internal characteristics of the individual.

[0098] In operation S407, feature fusion is performed based on the appearance feature, the physical feature, and the internal feature to obtain an individual fusion feature.

[0099] In operation S408, the fused features are processed based on the machine learning model to obtain expected quality information of the individual.

[0100] Operations S401 to S402 , operations S403 to S404 , and operations S405 to S406 may be performed synchronously or asynchronously, which is not limited herein.

[0101] The cloud service system can manage the entire life cycle of the collected multimodal features of the individual. After the multimodal features of the individual are collected, the life cycle of the multimodal features of the individual begins. In this stage, an effective period range can be set for the multimodal features of the individual, and the duration of the effective period range is the effective duration of the multimodal features. The effective duration of the multimodal features can be set according to the variety and current quality of the specific items. Optionally, the effective duration can be set based on the average duration of the quality change of the item in the natural environment. The effective period range of an individual can be expressed as [T-t0, T], where T can be expressed as the current moment and t0 can be expressed as the effective duration.

[0102] Optionally, when the multimodal features of an individual are within the valid period, the multimodal features recorded in the cloud service system can be directly used as the multimodal features of the individual at the current moment. Accordingly, the historical expected quality information determined when the multimodal features were collected at a historical moment can be used as the expected quality information of the individual at the current moment.

[0103] According to an embodiment of the present disclosure, determining expected quality information of each of the plurality of target items included in the to-be-recommended area based on the multimodal features of each of the plurality of target items may include the following operations:

[0104] Determine the time when the historical multimodal features of the individual are collected; when it is determined that the time when the historical multimodal features of the individual are collected is within the effective period, obtain the historical expected quality information of the individual determined based on the historical multimodal features of the individual; and determine the historical expected quality information of the individual as the expected quality information of the individual.

[0105] Optionally, in an offline shopping scenario, there may also be a situation where expected quality information of a selected individual is required. In this case, determining the expected quality information of the individual may include the following operations:

[0106] Data matching is performed in a database using the image of the individual to obtain a matching result; when the matching result indicates that the multimodal features of the individual exist in the database, the time at which the historical multimodal features of the individual are collected is determined; when it is determined that the time at which the historical multimodal features of the individual are collected is within a valid period, historical expected quality information of the individual determined based on the historical multimodal features of the individual is obtained; and, the historical expected quality information of the individual is determined to be the expected quality information of the individual.

[0107] Optionally, when the collection moment of the historical multimodal features of an individual is outside the valid period, it can be determined that the historical multimodal features of the individual are expired, which means that the life cycle of the collection moment of the historical multimodal features of the individual has ended. The historical multimodal features of the individual recorded in the cloud service system can be deleted, and the multimodal features of the individual can be re-collected, and the expected quality information of the individual can be determined based on the multimodal features. It will not be repeated here.

[0108] According to the embodiments of the present disclosure, by determining the expected quality information of an individual based on the multimodal characteristics of the individual, the multi-dimensional characteristics of the item can be comprehensively considered, so that the predicted quality information is more accurate, thereby improving the accuracy of the recommendation.

[0109] Optionally, both the preference information and the expected quality information of the individual can be represented as a matrix or a vector. The similarity between the preference information and the expected quality information of the individual can be determined through a similarity analysis method, and the individuals to be recommended can be screened based on the ranking of the similarities of multiple individuals.

[0110] According to an embodiment of the present disclosure, determining an individual to be recommended from a plurality of individuals based on preference information and expected quality information of each of the plurality of individuals may include the following operations:

[0111] A similarity analysis is performed based on the preference information and the expected quality information of the individual to obtain the similarity value of the individual; and an individual to be recommended is determined from the multiple individuals based on the similarity values ​​of the multiple individuals.

[0112] The similarity algorithm used in the similarity analysis is not limited here. For example, a cosine similarity algorithm, a similarity algorithm based on Euclidean distance, etc. may be used.

[0113] Optionally, among the individuals to be recommended determined by the similarity algorithm, there may be individuals with lower quality. Further, for the individuals obtained by similarity screening, individuals with better quality can be selected as individuals to be recommended.

[0114] According to an embodiment of the present disclosure, determining an individual to be recommended from a plurality of individuals based on preference information and expected quality information of each of the plurality of individuals may include the following operations:

[0115] A similarity analysis is performed based on preference information and expected quality information of individuals to obtain similarity values ​​of individuals; at least one candidate individual is determined from multiple individuals based on a similarity threshold and the similarity values ​​of multiple individuals; and an individual to be recommended is determined from the at least one candidate individual using a quality screening strategy.

[0116] The quality screening strategy can be expressed as screening individuals based on quality as the dominant factor. When performing quality screening, factors such as freshness and maturity can be given priority.

[0117] Optionally, the multimodal features of the candidate individuals may be input into a quality assessment model to obtain quality scores of the candidate individuals; and based on the quality score of each of the at least one candidate individuals, the individual to be recommended is determined from the at least one candidate individual.

[0118] The quality assessment model can be a model trained by considering factors such as freshness and maturity in quality. That is, in the training process of the model, the labels of the training samples used can be expressed as the freshness and maturity of individuals. Accordingly, the quality assessment model is used to process multimodal features, and the output quality score can be obtained by the weighted sum of the freshness score and the maturity score.

[0119] Optionally, the recommendation method described above may be implemented using a shopping guide device in an offline mall. The shopping guide device may be configured with a camera and a manipulator, and the target object may obtain recommendation information and shopping instructions through the shopping guide device when shopping in the offline mall.

[0120] Figure 5 The following schematically shows a workflow diagram of a shopping guide device according to an embodiment of the present disclosure.

[0121] like Figure 5 As shown, the workflow of the shopping guide device may include operations S501 to S509.

[0122] In operation S501 , in response to a selection operation, a target item selected by a target object is determined.

[0123] In operation S502, first guidance information is sent to the target object to guide the target object to aim the camera of the shopping guide device at the individual of the target item to obtain an image of the individual.

[0124] In operation S503, appearance features of the individual are obtained based on the image of the individual.

[0125] In operation S504, it is determined whether additional data needs to be collected. If it is determined that additional data needs to be collected, operation S505 is performed. If it is determined that additional data does not need to be collected, operation S508 is performed.

[0126] In operation S505, second guidance information is sent to the target object to guide the target object to grasp and squeeze the individual using the robot hand of the shopping guide device to obtain tactile feedback information of the individual.

[0127] In operation S506, it is determined whether additional data needs to be collected. If it is determined that additional data needs to be collected, operation S507 is performed. If it is determined that additional data does not need to be collected, operation S508 is performed.

[0128] In operation S507, third guidance information is sent to the target object to guide the target object to use the robot hand of the shopping guide device to perform a preset action on the individual so as to collect feedback audio.

[0129] In operation S508, quality assessment is performed based on the collected data to obtain individual expected quality information.

[0130] In operation S509 , recommendation information and purchase suggestions are provided based on the expected quality information of each of the plurality of individuals.

[0131] In operation S503 or operation S506, quality assessment can be performed based on the collected data, and confidence analysis can be performed on the assessment result. If the analysis result indicates that the assessment result is credible, it can be determined that no additional data needs to be collected. Correspondingly, if the analysis result indicates that the assessment result is not credible, it is determined that additional data needs to be collected.

[0132] Figure 6 The block diagram of a recommendation device according to an embodiment of the present disclosure is schematically shown.

[0133] like Figure 6 As shown, the recommendation device 600 may include an acquisition module 610 , a first determination module 620 , and a second determination module 630 .

[0134] The acquisition module 610 is used to obtain the target object's preference information about the target object in response to the individual recommendation request for the target object, where the preference information is used to represent the target object's preference for each of the multiple quality categories of the target object.

[0135] The first determination module 620 is configured to determine expected quality information of each of the plurality of target items included in the to-be-recommended area based on the multimodal features of each of the plurality of target items.

[0136] The second determination module 630 is used to determine the individual to be recommended from the multiple individuals based on the preference information and the expected quality information of each of the multiple individuals.

[0137] According to an embodiment of the present disclosure, the multimodal features include appearance features, physical features, and internal features.

[0138] According to an embodiment of the present disclosure, the first determining module 620 includes a first determining unit and a second determining unit.

[0139] The first determining unit is used to perform feature fusion on the appearance feature, the physical feature and the internal feature of the individual to obtain the fused feature of the individual.

[0140] The second determination unit is used to input the individual fusion features into the quality recognition model corresponding to the target item to obtain the individual expected quality information.

[0141] According to an embodiment of the present disclosure, the recommendation device 600 further includes a collection module.

[0142] The acquisition module is used to respectively acquire the appearance features, physical features and internal features of an individual to obtain multimodal features of the individual.

[0143] According to an embodiment of the present disclosure, the acquisition module includes a first acquisition unit, a second acquisition unit and a third acquisition unit.

[0144] The first acquisition unit is used to determine the appearance characteristics of the individual based on the image of the individual, wherein the image of the individual is obtained by controlling a camera device to photograph the individual.

[0145] The second acquisition unit is used to determine the physical characteristics of the individual based on the tactile feedback information of the individual, wherein the tactile feedback information of the individual is obtained by controlling a plurality of pressure sensors to squeeze the individual.

[0146] The third collection unit is used to determine the internal characteristics of the individual based on the feedback audio of the individual, wherein the feedback audio of the individual is collected when the target object performs a preset action to the individual.

[0147] According to an embodiment of the present disclosure, the first acquisition unit includes a first acquisition subunit, a second acquisition subunit, a third acquisition subunit, a fourth acquisition subunit and a fifth acquisition subunit.

[0148] The first acquisition subunit is used to extract the contour of the image to obtain the shape information of the individual.

[0149] The second acquisition subunit is used to obtain color information of the individual based on the pixel values ​​of each of a plurality of pixel points included in a target area of ​​the image, wherein the target area is an area in the image corresponding to the individual.

[0150] The third acquisition subunit is used to determine the abnormal sub-region included in the target area based on the pixel values ​​of each of the multiple pixel points included in the target area.

[0151] The fourth acquisition subunit is used to obtain individual surface defect information based on the position and number of abnormal sub-regions.

[0152] The fifth acquisition subunit is used to extract features based on shape information, color information and surface defect information to obtain appearance features.

[0153] According to an embodiment of the present disclosure, the second acquisition unit includes a sixth acquisition subunit and a seventh acquisition subunit.

[0154] The sixth acquisition subunit is used to obtain the pressure values ​​recorded by each of the multiple pressure sensors when controlling the multiple pressure sensors to move a preset distance.

[0155] The seventh acquisition subunit is used to determine the physical characteristics based on the pressure values ​​recorded by each of the multiple pressure sensors.

[0156] According to an embodiment of the present disclosure, the third acquisition unit includes an eighth acquisition subunit and a ninth acquisition subunit.

[0157] The eighth acquisition subunit is used to perform spectrum analysis on the feedback audio to obtain the frequency and amplitude of the feedback audio.

[0158] The ninth acquisition subunit is used to determine the internal characteristics based on the frequency and amplitude of the feedback audio.

[0159] According to an embodiment of the present disclosure, the first determining module 620 includes a third determining unit, a fourth determining unit, a fifth determining unit, and a sixth determining unit.

[0160] The third determining unit is used to perform data matching in a database using the image of the individual to obtain a matching result.

[0161] The fourth determining unit is used to determine the time of collecting the historical multimodal features of the individual when the matching result indicates that the multimodal features of the individual exist in the database.

[0162] The fifth determining unit is used to obtain the individual's historical expected quality information determined based on the individual's historical multimodal features when it is determined that the collection time of the individual's historical multimodal features is within the effective period.

[0163] The sixth determining unit is used to determine the individual's historical expected quality information as the individual's expected quality information.

[0164] According to an embodiment of the present disclosure, the second determining module 630 includes a seventh determining unit and an eighth determining unit.

[0165] The seventh determining unit is used to perform similarity analysis based on the preference information and the expected quality information of the individual to obtain the similarity value of the individual.

[0166] The eighth determining unit is used to determine the individual to be recommended from the multiple individuals based on the similarity values ​​of the multiple individuals.

[0167] According to an embodiment of the present disclosure, the second determining module 630 includes a ninth determining unit, a tenth determining unit, and an eleventh determining unit.

[0168] The ninth determining unit is used to perform similarity analysis based on the preference information and the expected quality information of the individual to obtain the similarity value of the individual.

[0169] The tenth determining unit is configured to determine at least one candidate individual from the multiple individuals based on the similarity threshold and the respective similarity values ​​of the multiple individuals.

[0170] The eleventh determining unit is used to determine the individual to be recommended from at least one candidate individual by using a quality screening strategy.

[0171] According to an embodiment of the present disclosure, the eleventh determining unit includes a first determining subunit and a second determining subunit.

[0172] The first determination subunit is used to input the multimodal features of the candidate individuals into the quality evaluation model to obtain the quality scores of the candidate individuals.

[0173] The second determining subunit is used to determine the individual to be recommended from at least one candidate individual based on the quality score of each of the at least one candidate individuals.

[0174] According to an embodiment of the present disclosure, the acquisition module 610 includes a first acquisition unit and a second acquisition unit.

[0175] The first acquisition unit is used to acquire historical behavior information related to the target object from the log related to the target object.

[0176] The second acquisition unit is used to determine the preference information based on the historical behavior information.

[0177] According to an embodiment of the present disclosure, the acquisition module 610 includes a third acquisition unit and a fourth acquisition unit.

[0178] The third acquisition unit is used to send interaction information to the target object through the interaction port, so that the target object inputs feedback information based on the interaction information.

[0179] The fourth acquisition unit is configured to determine preference information based on the feedback information in response to receiving the feedback information.

[0180] 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.

[0181] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0182] According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method as described above.

[0183] According to an embodiment of the present disclosure, a computer program product includes a computer program, and when the computer program is executed by a processor, the computer program implements the method as described above.

[0184] Figure 7 A schematic block diagram of an example electronic device 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, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, 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 required herein.

[0185] like Figure 7 As shown, the device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0186] A number of components in the device 700 are connected to an input / output (I / O) interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a disk, an optical disk, etc.; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 709 allows the device 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0187] The computing unit 701 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 701 performs the various methods and processes described above, such as the recommendation method. For example, in some embodiments, the recommendation method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the recommendation method described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform the recommendation method in any other appropriate manner (e.g., by means of firmware).

[0188] Various implementations 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 products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations 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.

[0189] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code 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.

[0190] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. 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 (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0191] To provide interaction with a target object, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the target object; and a keyboard and pointing device (e.g., a mouse or trackball) through which the target object can provide input to the computer. Other types of devices can also be used to provide interaction with the target object; for example, the feedback provided to the target object can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the target object can be received in any form (including acoustic input, voice input, or tactile input).

[0192] The systems and techniques described herein may be implemented in a computing system that includes a backend component (e.g., as a data server), or a computing system that includes a middleware component (e.g., an application server), or a computing system that includes a frontend component (e.g., a target object computer with a graphical target object interface or a web browser through which the target object can interact with an implementation 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 digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0193] A computer system may include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises through computer programs running on respective computers and having a client-server relationship to each other. The server may be a cloud server, a server in a distributed system, or a server combined with a blockchain.

[0194] 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 recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0195] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. 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 disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A recommendation method, comprising: In response to an individual recommendation request for a target item, obtaining preference information of the target object regarding the target item, wherein the preference information is used to represent the degree of preference of the target object for each of multiple quality categories of the target item; Determining expected quality information of each of the plurality of individuals of the target items included in the to-be-recommended area based on the multimodal features of each of the plurality of individuals; as well as Based on the preference information and the expected quality information of each of the plurality of individuals, an individual to be recommended is determined from among the plurality of individuals.

2. The method according to claim 1, wherein: The multimodal features include appearance features, physical features and internal features; The determining of the expected quality information of each of the plurality of individuals of the target items included in the to-be-recommended area based on the multimodal features of each of the plurality of individuals includes: Performing feature fusion on the appearance feature of the individual, the physical feature of the individual, and the internal feature of the individual to obtain a fused feature of the individual; and The individual fusion features are input into a quality recognition model corresponding to the target item to obtain the expected quality information of the individual.

3. The method according to claim 2, further comprising: The appearance features of the individual, the physical features of the individual and the internal features of the individual are collected respectively to obtain the multimodal features of the individual.

4. The method according to claim 3, wherein: The collecting of the appearance characteristics, the physical characteristics and the internal characteristics of the individual respectively includes: Determining appearance features of the individual based on an image of the individual, wherein the image of the individual is obtained by controlling a camera device to photograph the individual; Determining a physical characteristic of the individual based on tactile feedback information of the individual, wherein the tactile feedback information of the individual is obtained by controlling a plurality of pressure sensors to squeeze the individual; and Based on the feedback audio of the individual, internal features of the individual are determined, wherein the feedback audio of the individual is collected when the target object performs a preset action toward the individual.

5. The method according to claim 4, wherein: The determining, based on the image of the individual, the appearance feature of the individual comprises: Extracting contours of the image to obtain shape information of the individual; Obtaining color information of the individual based on respective pixel values ​​of a plurality of pixel points included in a target area of ​​the image, wherein the target area is an area in the image corresponding to the individual; Determine an abnormal sub-region included in the target region based on the pixel values ​​of each of the plurality of pixel points included in the target region; Based on the position and number of the abnormal sub-regions, obtaining surface defect information of the individual; and Feature extraction is performed based on the shape information, the color information and the surface defect information to obtain the appearance feature.

6. The method according to claim 4, wherein: The determining of the physical characteristics of the individual based on the tactile feedback information of the individual includes: When controlling the multiple pressure sensors to move a preset distance, obtaining pressure values ​​recorded by each of the multiple pressure sensors; and The physical characteristic is determined based on pressure values ​​recorded by each of the plurality of pressure sensors.

7. The method according to claim 4, wherein: The determining the internal characteristics of the individual based on the feedback audio of the individual includes: Performing spectrum analysis on the feedback audio to obtain the frequency and amplitude of the feedback audio; and The internal characteristics are determined based on the frequency and amplitude of the feedback audio.

8. The method according to claim 1, wherein: The determining of expected quality information of each of the plurality of target items included in the to-be-recommended area based on the respective multimodal features of the plurality of target items includes: Using the image of the individual to perform data matching in a database to obtain a matching result; If the matching result indicates that the multimodal features of the individual exist in the database, determining the time of collection of the historical multimodal features of the individual; When it is determined that the acquisition time of the historical multimodal features of the individual is within the effective period, obtaining the historical expected quality information of the individual determined based on the historical multimodal features of the individual; and The historical expected quality information of the individual is determined as the expected quality information of the individual.

9. The method according to claim 1, wherein: The step of determining the individual to be recommended from the multiple individuals based on the preference information and the expected quality information of each of the multiple individuals comprises: Performing a similarity analysis based on the preference information and the expected quality information of the individual to obtain a similarity value of the individual; and Based on the similarity values ​​of the multiple individuals, the individual to be recommended is determined from the multiple individuals.

10. The method according to claim 1, wherein: The step of determining the individual to be recommended from the multiple individuals based on the preference information and the expected quality information of each of the multiple individuals comprises: Performing a similarity analysis based on the preference information and the expected quality information of the individual to obtain a similarity value of the individual; Determining at least one candidate individual from the plurality of individuals based on a similarity threshold and the respective similarity values ​​of the plurality of individuals; and The individual to be recommended is determined from the at least one candidate individual by using a quality screening strategy.

11. The method according to claim 10, wherein: The step of using a quality screening strategy to determine the individual to be recommended from the at least one candidate individual includes: Inputting the multimodal features of the candidate individuals into a quality assessment model to obtain a quality score of the candidate individuals; and The individual to be recommended is determined from the at least one candidate individual based on the quality score of each of the at least one candidate individual.

12. The method according to claim 1, wherein: The obtaining of the target object's preference information about the target item includes: Acquiring historical behavior information related to the target object from a log related to the target object; and Based on the historical behavior information, the preference information is determined.

13. The method according to claim 1, wherein: The obtaining of the target object's preference information about the target item includes: sending interaction information through an interaction port with the target object so that the target object inputs feedback information based on the interaction information; and In response to receiving the feedback information, the preference information is determined based on the feedback information.

14. A recommendation device, comprising: an acquisition module, configured to acquire, in response to an individual recommendation request for a target item, preference information of a target object regarding the target item, wherein the preference information is used to represent the degree of preference of the target object for each of multiple quality categories of the target item; A first determination module, configured to determine expected quality information of each of the plurality of individuals of the target items included in the to-be-recommended area based on the multimodal features of each of the plurality of individuals; as well as The second determination module is used to determine the individual to be recommended from the multiple individuals based on the preference information and the expected quality information of each of the multiple individuals.

15. 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 13.

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

17. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 13.

Images