Information processing method and device

Abstract

The embodiment of the invention provides an information processing method and device, and the method comprises the steps: obtaining the attribute information and pricing information of an object in a target domain, and training a pricing model as a target pricing model according to the attribute information and the pricing information; pricing structured information meeting loading conditions is screened and loaded to a pricing factor library, and the pricing structured information is constructed according to a prediction result of the pricing model in a training stage; extracting target pricing information of a target object in a target category by using the target pricing model, and constructing target pricing structured information according to the target pricing information and target attribute information of the target object; the target pricing structured information is loaded to the pricing factor library, and the pricing structured information loaded in the pricing factor library is used for constructing object pricing rule information corresponding to the target field.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of information processing technology, and more particularly, to information processing methods and devices. Background Art

[0002] With the development of computers and the internet, more and more users are choosing to purchase furniture and appliances online. Furniture and appliances, as highly customized products, are often bundled with value-added services such as installation, maintenance, and insurance. Value-added services are typically calculated based on the object's attributes. For example, in the appliance industry, installation, maintenance, and custom cabinets all require a combination of these attributes to calculate the price of these services. For example, appliance installation typically offers different quotes based on attributes such as brand, power, and weight, while custom cabinets typically offer different quotes based on attributes such as size and material. However, because physical merchants often fail to consider the sales attributes of these services when publishing these objects on the platform, a large number of unmaintained or incorrectly maintained object attributes exist, leading to inaccurate pricing for these services and causing significant user frustration. Furthermore, different pricing standards vary for different object types, requiring manual setup not only for specialized personnel but also requiring significant manpower and time. Therefore, an effective solution is urgently needed to address these issues. Summary of the Invention

[0003] In view of this, embodiments of this specification provide an information processing method. One or more embodiments of this specification also relate to an information processing apparatus, a computing device, a computer-readable storage medium, and a computer program product to address technical deficiencies in the prior art.

[0004] According to a first aspect of the embodiments of this specification, there is provided an information processing method, including: Acquire attribute information and pricing information of objects in a target domain, and train a pricing model into a target pricing model based on the attribute information and the pricing information; Screening pricing structured information that meets the loading conditions and loading it into the pricing factor library, wherein the pricing structured information is constructed based on the prediction results of the pricing model in the training phase; Extracting target pricing information of target objects in a target category using the target pricing model, and constructing target pricing structured information based on the target pricing information and target attribute information of the target objects; The target pricing structured information is loaded into the pricing factor library, wherein the pricing structured information loaded into the pricing factor library is used to construct object pricing rule information corresponding to the target domain.

[0005] According to a second aspect of the embodiments of this specification, another information processing method is provided, which is applied to a server, including: Receive object attribute information of the object to be priced submitted by the client, and input the object attribute information into the target pricing model for processing to obtain object pricing information of the object to be priced; Generating a value-added service price corresponding to the object to be priced based on the object pricing rule information corresponding to the object to be priced and the object pricing information, wherein the object pricing rule information is constructed based on the structured information loaded in the pricing factor library in the above method; The value-added service price is sent to the client for display.

[0006] According to a third aspect of the embodiments of this specification, there is provided an information processing device, including: an acquisition module configured to acquire attribute information and pricing information of objects in a target domain, and train a pricing model into a target pricing model based on the attribute information and the pricing information; a screening module configured to screen pricing structured information that meets the loading conditions and load it into the pricing factor library, wherein the pricing structured information is constructed based on the prediction results of the pricing model in the training phase; a construction module configured to extract target pricing information of target objects in a target category using the target pricing model, and construct target pricing structured information based on the target pricing information and target attribute information of the target objects; The loading module is configured to load the target pricing structured information into the pricing factor library, wherein the pricing structured information loaded into the pricing factor library is used to construct object pricing rule information corresponding to the target domain.

[0007] According to a fourth aspect of the embodiments of this specification, another information processing device is provided, which is applied to a server, including: An information receiving module is configured to receive object attribute information of an object to be priced submitted by a client, and input the object attribute information into a target pricing model for processing to obtain object pricing information of the object to be priced; a price generation module configured to generate a value-added service price corresponding to the object to be priced based on object pricing rule information corresponding to the object to be priced and the object pricing information, wherein the object pricing rule information is constructed based on the structured information loaded in the pricing factor library in the above method; The price sending module is configured to send the value-added service price to the client for display.

[0008] According to a fifth aspect of the embodiments of this specification, there is provided a computing device, including: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the above-mentioned information processing method are implemented.

[0009] According to a sixth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above-mentioned information processing method are implemented.

[0010] According to a seventh aspect of the embodiments of this specification, a computer program product is provided, comprising a computer program or instructions, which implement the steps of the above-mentioned information processing method when executed by a processor.

[0011] The information processing method provided in this embodiment is to accurately extract pricing information for objects in the e-commerce platform, and to build structured information storage in combination with the attribute information and pricing information of the objects, so as to facilitate the construction of object pricing rule information for various objects. After obtaining the attribute information and pricing information of the objects in the target field, the pricing model can be trained as a target pricing model based on the attribute information and pricing information, so that the target pricing model can predict its corresponding pricing information in combination with the object attributes. In order to enable downstream services to use high-quality structured information to build object pricing rule information, pricing structured information that meets the loading conditions can be screened and loaded into the pricing factor library, wherein the pricing structured information is constructed according to the prediction results of the pricing model in the training stage; on this basis, in order to enrich the structured information loaded in the pricing factor library, the target pricing model can be used to extract the target object. The target pricing information of the target object in the standard category, and the target pricing structured information is constructed according to the target pricing information and the target attribute information of the target object; after the target pricing structured information is loaded into the pricing factor library, it can support the downstream to construct the object pricing rule information corresponding to the target field according to the pricing structured information loaded in the pricing factor library, thereby realizing the automatic completion of the object pricing information extraction, and ensuring the extraction accuracy, and constructing and storing structured information based on this, which can reasonably and standardizedly construct pricing rule information for the value-added services bound in the field, thereby quickly solving the problem of inconsistent pricing of objects in different categories. At the same time, the above processing does not require additional human resources to participate. The extraction of the corresponding pricing information of the object and the construction of structured information can be completed only through the pricing model, which can effectively save development costs and time, thereby making the pricing setting more standardized. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a flow chart of an information processing method provided by one embodiment of this specification; Figure 2is a flowchart of another information processing method provided by one embodiment of this specification; Figure 3 This is a flowchart of a processing process of an information processing method provided by one embodiment of this specification; Figure 4 This is a schematic diagram of the structure of an information processing device provided by one embodiment of this specification; Figure 5 is a structural diagram of another information processing device provided by an embodiment of this specification; Figure 6 This is a structural block diagram of a computing device provided by one embodiment of this specification. DETAILED DESCRIPTION

[0013] The following description sets forth many specific details to facilitate a thorough understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0014] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a," "the," and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0015] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0016] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0017] The pricing model in the technical solution provided in the embodiments of this application can use a deep learning model with a relatively large model parameter scale, such as a large model containing billions or more model parameters. A large model is only an example of a model, and the embodiments of this application do not limit the number of model parameters supported by the deep learning model used, with the goal of meeting actual needs.

[0018] First, the terms involved in one or more embodiments of this specification are explained.

[0019] Prompt Tuning is a technique for efficiently fine-tuning pre-trained models. By adding prompts to the model input, it guides it to adapt to specific tasks without retraining all parameters. The prompts can be text templates or trainable embedding vectors.

[0020] LoRA is a method for optimizing model fine-tuning through low-rank matrix decomposition technology. It decomposes the parameter matrix of the pre-trained model into the product of two low-rank matrices and only updates the parameters of the decomposed low-rank matrix, thereby reducing computational complexity and storage requirements.

[0021] In this specification, an information processing method is provided. This specification also relates to an information processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, which are described in detail one by one in the following embodiments.

[0022] See also Figure 1 , Figure 1 A flowchart of an information processing method provided according to an embodiment of this specification is shown, which specifically includes the following steps.

[0023] Step S102: Acquire attribute information and pricing information of objects in the target domain, and train a pricing model into a target pricing model based on the attribute information and the pricing information.

[0024] The information processing method provided in this embodiment can be applied to the construction of a pricing factor library for any object that needs to be configured with a value-added service, such as air conditioners, televisions, bookcases, bathroom cabinets, basins, cabinets, floor tiles, floors, etc. If such objects are purchased through an e-commerce platform, the platform or merchant will generally provide value-added services for such objects to provide object installation. When providing value-added services for different types of objects, different methods will be used to calculate the value-added service price, affected by the object type; for example, for air conditioners, the value-added service price needs to be calculated according to the floor, the length of the outdoor unit and the indoor unit pipe, while the value-added service price for cabinets needs to be calculated according to the length, material, etc. However, the calculation of such value-added service prices is often performed by merchants or professionals and then entered into the platform for use. Before this, the user needs to understand the material, type, size and other attributes of the object, and then complete the value-added service price calculation in combination with the industry standards. This not only consumes a lot of human resources but also requires users to have a clear understanding of the pricing standards for value-added services for objects in the industry, resulting in a significant increase in cost and time.

[0025] The information processing method provided in this embodiment can train a target pricing model to address the above-mentioned problem. The target pricing model automatically extracts pricing information of different types of objects, and loads the pricing information into a pricing factor library in combination with object attributes, so that high-quality pricing information can be directly read from the pricing factor library for use when applied. The construction of pricing rule information for different types of objects can be completed without the participation of user professionals, thereby effectively improving processing efficiency and at the same time promoting more standardized price calculations for value-added services in the industry.

[0026] Specifically, the object refers to a commodity object, which may include a physical commodity object, a digital collection, or other virtual commodity objects, and this embodiment does not impose any limitations thereon. The target domain specifically refers to the domain to which the object for which pricing information extraction is required belongs, such as the furniture domain, the home appliance domain, the customized furniture domain, etc.; accordingly, the attribute information specifically refers to the multimodal information that characterizes the attributes of the object, including but not limited to the text description information of the object, the image of the object, the name of the object, the recommended information of the object, etc. Accordingly, the pricing information specifically refers to the information required to participate in the calculation of the value-added price of the object, including but not limited to the length, height, width, material, etc. of the object. Accordingly, the pricing model specifically refers to the model to be trained that inputs the attribute information of the object and outputs the pricing information of the object. It can be understood that the pricing model is a model for extracting the pricing information corresponding to the object, and the target pricing model specifically refers to the model obtained after it is fully trained with samples.

[0027] Based on this, in order to accurately extract pricing information for objects in the e-commerce platform, and build structured information storage based on the object's attribute information and pricing information, so as to facilitate the construction of object pricing rule information for various objects, after obtaining the attribute information and pricing information of objects in the target field, the pricing model can be trained as a target pricing model based on the attribute information and pricing information, so that the target pricing model can predict its corresponding pricing information based on the object attributes. In order to enable downstream services to use high-quality structured information to build object pricing rule information, the pricing structured information that meets the loading conditions can be screened and loaded into the pricing factor library, wherein the pricing structured information is constructed based on the prediction results of the pricing model in the training phase; on this basis, in order to enrich the structured information loaded in the pricing factor library, the target pricing model can be used to extract the objects in the target category. The target pricing information of the target object is obtained, and the target pricing structured information is constructed according to the target pricing information and the target attribute information of the target object; thus, after the target pricing structured information is loaded into the pricing factor library, the downstream can be supported to construct the object pricing rule information corresponding to the target field according to the pricing structured information loaded in the pricing factor library, thereby realizing the automatic completion of the object pricing information extraction and ensuring the extraction accuracy. Based on this, structured information is constructed and stored, and pricing rule information can be reasonably and standardizedly constructed for the value-added services bound within the field, thereby quickly solving the problem of inconsistent pricing of objects of different categories. At the same time, the above processing does not require additional human resources to participate. The extraction of the corresponding pricing information of the object and the construction of structured information can be completed only through the pricing model, which can effectively save development costs and time, thereby making the pricing setting more standardized.

[0028] Furthermore, during model training, in order to achieve higher prediction accuracy and extract pricing information of any object, a multimodal pricing model can be combined to complete the training of the pricing model. In this embodiment, the specific implementation is as follows: Obtain image information and description information of objects in a target domain, as well as size information and material information of the objects, wherein the image information and the description information constitute attribute information, and the size information and the material information constitute pricing information; input the attribute information into a pricing model for processing to obtain predicted pricing information; adjust the parameters of the pricing model according to the predicted pricing information and the pricing information until a target pricing model that meets the training stop conditions is obtained.

[0029] Specifically, image information refers to the image corresponding to the object, and description information refers to the text description corresponding to the object, such as evaluation and introduction. Size information refers to the length, width and height dimensions corresponding to the object, and material information refers to the material, color and other information of the raw materials corresponding to the object. Prediction and pricing information refers to the prediction results output by the pricing model. Training stop conditions refer to the conditions for stopping the training model, including but not limited to the number of iterations, loss value comparison conditions, validation set verification conditions, etc. During specific implementation, they can be selected according to actual needs, and this embodiment does not impose any restrictions here.

[0030] Based on this, in order to fully train the model, the image information (such as appearance, structural pictures) and description information (such as text description, parameter text) of the object can be obtained from the target domain object, and the object's size (such as length, width, height, volume) and material (such as metal, plastic) and other pricing information can be collected simultaneously; then the image and description information are integrated into multimodal "attribute information" as model input, and the size and material information are combined into "pricing information" as training labels; then the attribute information is input into the pre-trained pricing model (such as a vision-language model), and the predicted pricing result is output through feature fusion and prediction layer; finally, by comparing the error between the predicted pricing information and the actual pricing information, the model parameters are iteratively adjusted using optimization algorithms such as gradient descent, and training stop conditions are set (such as error rate threshold or upper limit of iteration number) until the model performance meets the requirements, and finally a high-precision target pricing model is output for subsequent use.

[0031] In summary, through multimodal data fusion (image and text) and dynamic parameter tuning, the accuracy of object pricing prediction is significantly improved, the reliance on manual labeling is reduced, and accurate modeling of complex object properties (such as material and size) is achieved. This reduces the error rate, enhances the model's generalization ability for new scenarios, and reduces training resource consumption.

[0032] Furthermore, in order to enable the model to complete training automatically, Prompt Tuning technology and LoRA technology can be used to complete the model setting. In this embodiment, the specific implementation method is as follows: At least one parameter matrix in a multimodal deep learning model is decomposed by a low-rank matrix decomposition algorithm to obtain low-rank matrix parameters; the multimodal deep learning model is updated using the low-rank matrix parameters to obtain an initial pricing model; prompt information to be trained is added to the input layer of the initial pricing model, and the pricing model is generated according to the adding result; wherein the prompt information to be trained includes discrete prompt information and continuous prompt information.

[0033] Specifically, the multimodal deep learning model refers to a model capable of processing multimodal object attribute information, the low-rank matrix decomposition algorithm refers to LoRA technology, which is used to fine-tune the model; the prompt information to be trained refers to the preset prompt information added to the input layer of the initial pricing model to inform the model of the processing tasks to be performed, and will be continuously optimized during the model training phase to enable the model to have a strong ability to extract pricing information. Among them, discrete prompt information is a predefined special marker used to guide the model to predict the key attributes of pricing information, and continuous prompt information is a trainable virtual token vector, the number of which can be 8-32, and the parameters are optimized through backpropagation.

[0034] Based on this, in order to enable the pricing model to be automatically trained and able to process multimodal information, a multimodal deep learning model can be used to construct the pricing model. During this process, at least one parameter matrix in the multimodal deep learning model can be decomposed using a low-rank matrix decomposition algorithm to obtain low-rank matrix parameters. The low-rank matrix parameters can then be used to update the multimodal deep learning model to obtain an initial pricing model. This eliminates the need for extensive parameter adjustments during the training phase, and only requires adjusting the low-rank matrix parameters to give the model new processing capabilities. On this basis, prompt information to be trained can be added to the input layer of the initial pricing model, and a pricing model can be generated based on the added results. The prompt information to be trained includes discrete prompt information and continuous prompt information, so that the model has the ability to extract pricing information.

[0035] In other words, to enable the pricing model to accurately extract pricing information from object attributes for use in building pricing rules, LoRA (Low Rank Adaptation) fine-tuning and Prompt Tuning techniques can be used to automate model training. Furthermore, to ensure model robustness, the model can be validated using small-scale data, guiding the model to extract key information from object text descriptions and images.

[0036] For example, in an e-commerce scenario, to quickly extract pricing information for furniture products, a pricing model can be trained. Initially, the product attributes and pricing information can be obtained. For example, a bathroom cabinet description ("This bathroom cabinet is made of environmentally friendly wood...") and a product image can be included. The model's dimensions ("80cm*50cm*85cm"), material ("Solid Wood"), and color ("Cream White") can also be determined. Based on this, the model can be automatically trained using LoRA (Low Rank Adaptation) fine-tuning and PromptTuning techniques. During training, the product description and image can be fed into the pricing model for processing. The model outputs pricing information, including the cabinet's dimensions, material, and color. Parameters can then be adjusted based on the actual dimensions, material, and color of the cabinet, combined with the model output. Training can then continue with new samples. After a set number of model iterations, the target pricing model is generated. This model can input any furniture product image and description and output the product's dimensions, color, and material.

[0037] In summary, by achieving efficient parameter updates through low-rank matrix decomposition and optimizing input guidance by combining discrete and continuous prompt information, the parameter efficiency and task adaptability of the multimodal pricing model are significantly improved. While reducing computing resource consumption, the pricing accuracy is improved and the model's generalization ability for complex multimodal inputs is enhanced.

[0038] Step S104 , screening pricing structured information that meets the loading conditions and loading it into the pricing factor library, wherein the pricing structured information is constructed according to the prediction results of the pricing model in the training phase.

[0039] Specifically, after the target pricing model that can extract pricing information is trained as described above, in order to build a high-quality pricing factor library, samples that are accurately predicted by the model during the training phase can be selected to construct pricing structured information and loaded into the pricing factor library, ensuring that the pricing structured information loaded in the pricing factor library is structured information of pricing information accurately extracted based on attribute information, while achieving the purpose of data backup, and can be used for the subsequent construction of pricing rule information for the target field.

[0040] Specifically, the loading conditions refer to the conditions for selecting pricing information and its corresponding attribute information that are correctly predicted by the model. Pricing structured information specifically refers to structured data constructed by combining pricing information and attribute information, which is sufficient for loading into the pricing factor library. Furthermore, pricing structured information is structured data constructed from pricing information accurately predicted by the model and its associated attribute information. Accordingly, the pricing factor library specifically refers to a database that stores high-quality pricing structured information from which pricing information is accurately extracted.

[0041] Furthermore, when screening pricing structured information that meets the conditions, in order to ensure that the pricing structured information in the pricing factor library is accurate and high-quality pricing structured information, the correct prediction result can be selected for construction. In this embodiment, the specific implementation method is as follows: Acquire multiple training sample pairs used by the pricing model in the training phase, wherein the training sample pairs include attribute information and pricing information; determine the prediction result information corresponding to each of the multiple training sample pairs, and filter target training sample pairs that meet the loading conditions from the multiple training sample pairs based on the prediction result information; construct pricing structured information based on the attribute information and pricing information included in the target training sample pairs, and load it into the pricing factor library.

[0042] Specifically, the multiple training sample pairs refer to sample pairs constructed by the pricing model using attribute information and pricing information corresponding to objects in the target domain. The prediction result information refers to whether the model accurately extracts pricing information after processing the attribute information corresponding to each object during the training phase. The target training sample pair refers to the training sample pair for which the prediction result is correct.

[0043] Based on this, in order to ensure that the pricing structured information loaded into the pricing factor library is of higher quality and accuracy, the correct prediction results during the model training phase can be selected and constructed. During this process, multiple training sample pairs used by the pricing model during the training phase can be obtained, where the training sample pairs contain attribute information and pricing information. Afterwards, the prediction result information corresponding to each of the multiple training sample pairs can be determined, and based on the prediction result information, target training sample pairs that meet the loading conditions can be selected from the multiple training sample pairs. At this point, pricing structured information can be constructed based on the attribute information and pricing information contained in the target training sample pairs and loaded into the pricing factor library for subsequent use.

[0044] Continuing with the above example, during the training phase of the pricing model, each time model training is completed, it is necessary to compare the size, material, and color extracted by the model with the actual size, material, and color of the furniture product. If the two are consistent, the description information and image of the furniture product can be obtained, and the pricing structured information can be constructed in combination with its corresponding size, material, and color. It can then be loaded into the pricing factor library. Similarly, when the model training is completed, the pricing factor library will store a large amount of information to extract accurate pricing structured information, which is convenient for downstream services to use.

[0045] In summary, by screening high-quality training samples to build a structured pricing factor library, the efficiency of model verification and optimization can be significantly improved, the automated knowledge accumulation of pricing rules can be achieved, the reliance on manual labeling can be reduced, the model output can be ensured to dynamically match industry standards, and the risk of pricing conflicts can be reduced.

[0046] In addition, the model can be verified by using high-quality structured information in the pricing factor library, and if the verification fails, it can be optimized using the high-quality pricing structured information. In this embodiment, the specific implementation is as follows: The target pricing model is verified using the pricing structured information loaded in the pricing factor library; if the verification fails, the target pricing model is optimized using the pricing structured information loaded in the pricing factor library.

[0047] In addition, considering that if the model is deployed and used directly after training, the prediction results may be inaccurate due to insufficient model training, in order to avoid this problem and make the data reusable multiple times, the pricing structured information loaded in the pricing factor library can be selected to optimize the model. In other words, the pricing structured information loaded in the pricing factor library can be used to verify the target pricing model first; if the verification fails, it means that the model training is insufficient. The structured information loaded in the pricing factor library is constructed from attribute information and pricing information. Therefore, the pricing structured information loaded in the pricing factor library can be used to optimize the target pricing model, thereby ensuring that the target pricing model has more accurate prediction capabilities.

[0048] In summary, the structured information of the pricing factor library can be used to achieve automated verification and closed-loop optimization of model outputs, significantly improving the compliance and accuracy of pricing results, reducing manual intervention, and enhancing the model's adaptability to complex pricing rules, ensuring that the output is dynamically consistent with industry standards and improving system robustness.

[0049] Step S106 , extracting target pricing information of target objects in a target category using the target pricing model, and constructing target pricing structured information according to the target pricing information and target attribute information of the target objects.

[0050] Specifically, after loading the pricing structured information constructed in the above-mentioned model training phase into the pricing factor library, in order to improve the richness of the pricing structured information in the pricing factor library, the trained target pricing model can also be used to extract the target pricing information of the target object in the target category. On this basis, the target pricing structured information is constructed according to the target pricing information and the target attribute information of the target object, so as to continue to load high-quality pricing structured information into the pricing factor library.

[0051] The target category specifically refers to the object category selected based on demand after deploying the target pricing model. This could include, for example, selecting an object category with high sales volume or an object category with high-quality attribute information (e.g., clear object images and accurate text descriptions). In specific implementations, the target category can be selected based on demand, and this embodiment does not impose any limitations. Accordingly, target pricing information specifically refers to the pricing information corresponding to the target objects within the target category, extracted from the target attribute information corresponding to the target objects. Accordingly, target pricing structured information is structured information constructed by combining the target object's attribute information and pricing information.

[0052] Furthermore, in order to achieve the goal of self-learning by optimizing the model while it is being applied, it is possible to select popular object categories for processing when predicting target pricing information, so that after the model has strong predictive capabilities, it can be deployed for all object categories. In this embodiment, the specific implementation method is as follows: Determine multiple candidate object categories and obtain transaction information of candidate objects in each candidate object category; determine category popularity information of each candidate object category based on the transaction information; select a target category from the multiple candidate object categories according to the category popularity information, and execute the step of extracting target pricing information of the target object in the target category using the target pricing model.

[0053] Specifically, candidate object categories refer to categories for which object pricing information can be extracted using the target pricing model, such as air conditioners, refrigerators, bookcases, cabinets, wardrobes, and lamps. Correspondingly, transaction information refers to sales figures for each candidate object. Category popularity information refers to the popularity information obtained by integrating transaction information for objects within the candidate category, representing the transaction volume of objects in that category. Subsequently, the category with the highest sales volume can be selected as the target category.

[0054] Based on this, we consider that if the target pricing model is deployed immediately after training, it may not provide accurate predictions for some categories. For example, for products with low-quality images, the target pricing model may not accurately extract pricing information. Therefore, to ensure that the target pricing model does not affect prediction accuracy after deployment and has strong processing capabilities after subsequent learning-by-application, it can be fully used on e-commerce platforms. Target categories can be filtered to construct structured information.

[0055] Therefore, multiple candidate object categories can be determined first, and then the transaction information of the candidate objects in each candidate object category can be obtained; then the category popularity information of each candidate object category can be determined based on the transaction information; and the target category can be selected from the multiple candidate object categories according to the category popularity information. On this basis, the step of extracting the target pricing information of the target object in the target category using the target pricing model can be executed.

[0056] In summary, by dynamically analyzing the popularity of object categories, accurately locating high-value target categories and prioritizing the application of pricing models, we can significantly improve resource utilization, ensure the pricing accuracy of popular categories, reduce redundant calculations, achieve intelligent matching of pricing models with market demand, and enhance the system's responsiveness to hot objects.

[0057] Step S108: Load the target pricing structured information into the pricing factor library, wherein the pricing structured information loaded into the pricing factor library is used to construct object pricing rule information corresponding to the target domain.

[0058] Specifically, after extracting the pricing information of the target object of the target category through the target pricing model mentioned above, the target pricing structured information will be constructed based on the target pricing information and target attribute information of the target object. Thereafter, the target pricing structured information can be loaded into the pricing factor library. Similarly, after the pricing information loaded in the pricing factor library meets the usage requirements, the downstream service can use the pricing structured information loaded in the pricing factor library to construct the object pricing rule information corresponding to the target field.

[0059] Furthermore, when constructing object pricing rule information, considering that the pricing factor library contains a lot of pricing structured information, which may involve different object categories, clustering can be used. In this embodiment, the specific implementation is as follows: Cluster the pricing structured information contained in the pricing factor library according to the object category to obtain an information set corresponding to the object category; and construct object pricing rule information corresponding to the object category based on the pricing information contained in the pricing structured information in the information set and the price information corresponding to the object category.

[0060] Specifically, an information set refers to a collection of structured pricing information corresponding to objects within the same object category. Correspondingly, price information refers to the unit price set for that product type within the industry, such as 1 yuan per square meter for bathroom cabinets, 3 yuan for the installation fee for a 50-inch TV, and 4 yuan for the installation fee for a 55-inch TV.

[0061] Based on this, in order to facilitate the setting of pricing rule information, the pricing structured information contained in the pricing factor library can be clustered according to the object category, so that the pricing structured information of the same object category constitutes an information set; thereafter, the object pricing rule information corresponding to the object category can be constructed based on the pricing information contained in the pricing structured information in the information set and the price information corresponding to the object category.

[0062] For example, most bathroom cabinets are 80cm*50cm*85cm in size, made of environmentally friendly wood, and creamy white in color. According to industry installation standards, the value-added service price can be set at S1 yuan, and an additional S2 yuan is charged for every 10cm exceeding the limit. This is how the pricing rules for bathroom cabinets are set. When users purchase any bathroom cabinet product through an e-commerce platform, the value-added service price of the bathroom cabinet can be calculated according to the pricing rules, so that users can understand the installation costs and improve pricing compliance.

[0063] In summary, through object category clustering and structured information modeling, we can accurately construct pricing rule information for each category, significantly improve the matching degree between pricing rules and actual market prices, reduce the cost of manual rule-making, achieve cross-category pricing standardization and dynamic updating, and enhance the model's adaptability to segmented market scenarios.

[0064] Furthermore, in order to achieve self-learning of the model and improve its prediction accuracy, error pricing information can be collected after the model is deployed to optimize it. In this embodiment, the specific implementation method is as follows: Receive erroneous pricing information submitted for the target pricing model, and determine the associated object corresponding to the erroneous pricing information and the associated attribute information of the associated object; construct an initial model optimization sample pair based on the erroneous pricing information and the associated attribute information, and adjust the initial optimization sample pair to a model optimization sample pair; use the model optimization sample pair to optimize the target pricing model.

[0065] Specifically, erroneous pricing information refers to incorrect pricing information extracted by the target pricing model when extracting pricing information for an object, which can be reported by the merchant. Accordingly, the associated object is the object from which the pricing extraction error occurred, and the associated attribute information is the attribute information corresponding to the associated object. Accordingly, the initial model optimization sample pair refers to the sample pair constructed based on the erroneous pricing information and associated attribute information of the associated object, and the model optimization sample pair refers to the sample pair obtained after correcting the erroneous pricing information.

[0066] Based on this, after the target pricing model is deployed, the problem of target pricing model prediction errors cannot be ruled out. When the prediction is wrong, the merchant can feedback an error request. At this time, the erroneous pricing information can be determined based on the error request. On this basis, in order to achieve continuous learning and evolution of the model, the associated objects corresponding to the erroneous pricing information and the associated attribute information of the associated objects can be determined; then the initial model optimization sample pair can be constructed based on the erroneous pricing information and the associated attribute information, and after adjusting the initial optimization sample pair to the model optimization sample pair, the model optimization sample pair can be used to optimize the target pricing model, so that the model can learn the missing knowledge and adapt to more pricing information extraction scenarios.

[0067] In other words, incorrect pricing information reported by merchants is fed back into the model's bad case data for fine-tuning. This closed-loop mechanism allows the pricing model to gradually converge to a more accurate prediction rate.

[0068] For example, when the target pricing model extracts pricing information for bookcase B, it determines that the size of bookcase B is 120cm*30cm*200cm, the color is gray, and the material is sawtooth wood; the installation fee is calculated according to the pricing information and then fed back to the merchant. At this time, an error reminder is received from the merchant, and the size of bookcase B is determined to be 100cm*30cm*200cm. At this time, it is determined that the pricing information extracted by the target pricing model is wrong, and it can be stored in the bad case data of the model, so that the model can be fine-tuned in combination with the bad case data in the future, so that the model can continue to learn and optimize, thereby having stronger and more accurate prediction capabilities.

[0069] In summary, by capturing erroneous pricing information in real time and constructing precise optimized sample pairs, dynamic adaptive optimization of the model can be achieved, which significantly reduces manual intervention, improves pricing robustness, and enhances the model's ability to correct abnormal inputs, ensuring that pricing results dynamically match actual attributes.

[0070] The information processing method provided in this embodiment leverages the vast amount of data and prior knowledge from the e-commerce industry, employing prompt tuning and LoRA (low-rank adaptation) fine-tuning techniques. By updating only a small number of parameters (such as low-rank matrix factorization), this method significantly reduces training costs while maintaining the model's high performance. By designing domain-specific prompt templates, the model is guided to more accurately extract key pricing features of objects. Furthermore, the model incorporates image processing capabilities to extract visual information from object images and integrates it with text features for multimodal fusion, enabling comprehensive structured storage and analysis of object attributes.

[0071] Secondly, by implementing automated training and deployment, and continuously learning and evolving strategies, we ensure that the model can dynamically optimize performance based on the latest object data. By tracking dynamic updates to object data and integrating it with an automated pipeline, we complete incremental training and testing of the model. During the continuous evolution of the model, a small-scale reflow iteration mechanism is designed to address extreme error scenarios. Specifically, when the model makes errors in extracting a very small number of special object attributes, the system automatically triggers an active learning strategy, reflowing these samples into the training set and performing rapid iterative optimization through small-batch gradient descent. This approach not only alleviates the model's misjudgment of abnormal object attributes but also significantly improves the model's robustness and generalization capabilities.

[0072] The information processing method provided in this embodiment is to accurately extract pricing information for objects in the e-commerce platform, and to build structured information storage in combination with the attribute information and pricing information of the objects, so as to facilitate the construction of object pricing rule information for various objects. After obtaining the attribute information and pricing information of the objects in the target field, the pricing model can be trained as a target pricing model based on the attribute information and pricing information, so that the target pricing model can predict its corresponding pricing information in combination with the object attributes. In order to enable downstream services to use high-quality structured information to build object pricing rule information, pricing structured information that meets the loading conditions can be screened and loaded into the pricing factor library, wherein the pricing structured information is constructed according to the prediction results of the pricing model in the training stage; on this basis, in order to enrich the structured information loaded in the pricing factor library, the target pricing model can be used to extract the target object. The target pricing information of the target object in the standard category, and the target pricing structured information is constructed according to the target pricing information and the target attribute information of the target object; after the target pricing structured information is loaded into the pricing factor library, it can support the downstream to construct the object pricing rule information corresponding to the target field according to the pricing structured information loaded in the pricing factor library, thereby realizing the automatic completion of the object pricing information extraction, and ensuring the extraction accuracy, and constructing and storing structured information based on this, which can reasonably and standardizedly construct pricing rule information for the value-added services bound in the field, thereby quickly solving the problem of inconsistent pricing of objects in different categories. At the same time, the above processing does not require additional human resources to participate. The extraction of the corresponding pricing information of the object and the construction of structured information can be completed only through the pricing model, which can effectively save development costs and time, thereby making the pricing setting more standardized.

[0073] See also Figure 2 , Figure 2 A flowchart of another information processing method provided according to an embodiment of the present specification is shown, which is applied to a server and specifically includes the following steps.

[0074] Step S202: receiving object attribute information of the object to be priced submitted by the client, and inputting the object attribute information into the target pricing model for processing to obtain object pricing information of the object to be priced.

[0075] Step S204: Generate the value-added service price corresponding to the object to be priced according to the object pricing rule information corresponding to the object to be priced and the object pricing information, wherein the object pricing rule information is constructed according to the structured information loaded in the pricing factor library in the above method.

[0076] Step S206: Send the value-added service price to the client for display.

[0077] For the contents not fully described in detail in the other information processing method provided in this embodiment, please refer to the same or corresponding description in the above embodiments, and this embodiment will not be described in detail here.

[0078] Specifically, the client refers to the client owned by the merchant or user who submitted the object to be priced and needs to retrieve its corresponding pricing information. Accordingly, the object to be priced refers to the object for which pricing information is to be retrieved. Object pricing information and object attribute information refer to the attribute information and pricing information corresponding to the object to be priced. Object pricing rule information refers to the pricing rule information for the object category corresponding to the object to be priced. The value-added service price refers to the installation fee or value-added service fee corresponding to the object to be priced.

[0079] Based on this, after receiving the object attribute information of the object to be priced submitted by the client, the object attribute information can be input into the target pricing model for processing to obtain the object pricing information of the object to be priced. Based on this, the value-added service price corresponding to the object to be priced can be generated based on the object pricing rule information and object pricing information corresponding to the object to be priced. The value-added service price can then be sent to the client for display, allowing the user to decide whether to purchase the object to be priced or whether to select the value-added service corresponding to the object.

[0080] For example, when a user buys a TV on an e-commerce platform, he or she chooses an 85-inch TV from Brand A. At this time, the product description and image of the 85-inch TV from Brand A can be input into the pricing model for processing, and the model outputs the pricing information {85 inches, black, TV}. Then, based on the cost standard of the TV installation service, the value-added service price of the TV can be calculated as S4 yuan, and the cost of purchasing an additional mounting bracket is S5 yuan. The value-added service price can then be displayed to the user so that the user can understand the installation service fee of the TV.

[0081] In summary, when users purchase any type of object with a value-added service price, the pricing information of the object can be automatically extracted through the target pricing model, and the calculation of the value-added service can be completed in combination with the pricing rule information corresponding to the object. This can make it easier for users to understand the value-added service price and object pricing information, thereby improving the user's object purchasing experience.

[0082] The following combined Figure 3 , taking the application of the information processing method provided in this specification in the customized furniture scenario as an example, the information processing method is further explained. Figure 3 A flowchart of a processing process of an information processing method provided by an embodiment of this specification is shown, which specifically includes the following steps.

[0083] Step S302: Acquire image information and description information of the product in the target area, as well as size information and material information of the product.

[0084] Step S304: compose attribute information based on the image information and the description information, and compose pricing information based on the size information and the material information.

[0085] Step S306: Obtain a multimodal deep learning model, and decompose at least one parameter matrix in the multimodal deep learning model using a low-rank matrix decomposition algorithm to obtain low-rank matrix parameters.

[0086] Step S308: Use the low-rank matrix parameters to update the multimodal deep learning model to obtain an initial pricing model.

[0087] Step S310 , adding prompt information to be trained to the input layer of the initial pricing model, and generating a pricing model according to the adding result; wherein the prompt information to be trained includes discrete prompt information and continuous prompt information.

[0088] Step S312: input the attribute information into the pricing model for processing to obtain predicted pricing information.

[0089] Step S314: Adjust the parameters of the pricing model according to the predicted pricing information and the pricing information until a target pricing model that meets the training stop conditions is obtained.

[0090] Step S316: Acquire multiple training sample pairs used by the pricing model in the training phase, wherein the training sample pairs include attribute information and pricing information.

[0091] Step S318: determining prediction result information corresponding to the plurality of training sample pairs, and screening target training sample pairs that meet the loading conditions from the plurality of training sample pairs according to the prediction result information.

[0092] Step S320: construct pricing structured information based on the attribute information and pricing information contained in the target training sample pair, and load it into the pricing factor library.

[0093] Step S322 , extracting target pricing information of target products in the target category using the target pricing model, and constructing target pricing structured information based on the target pricing information and target attribute information of the target products.

[0094] Step S324: Load the target pricing structured information into the pricing factor library, wherein the pricing structured information loaded into the pricing factor library is used to construct commodity pricing rule information corresponding to the target domain.

[0095] Step S326: upon receiving the commodity attribute information of the commodity to be priced submitted by the client, the commodity attribute information is input into the target pricing model for processing to obtain the commodity pricing information of the commodity to be priced.

[0096] Step S328: Generate the value-added service price corresponding to the commodity to be priced based on the commodity pricing rule information and commodity pricing information corresponding to the commodity to be priced, and send the value-added service price to the client for display.

[0097] In summary, in order to accurately extract pricing information for goods on the e-commerce platform, and build structured information storage based on the attribute information and pricing information of the goods, so as to facilitate the construction of commodity pricing rule information for various commodities, after obtaining the attribute information and pricing information of the goods in the target field, the pricing model can be trained as a target pricing model based on the attribute information and pricing information, so that the target pricing model can predict the corresponding pricing information based on the commodity attributes. In order to enable downstream services to use high-quality structured information to build commodity pricing rule information, the pricing structured information that meets the loading conditions can be screened and loaded into the pricing factor library, wherein the pricing structured information is constructed based on the prediction results of the pricing model in the training phase; on this basis, in order to enrich the structured information loaded in the pricing factor library, the target pricing model can be used to extract the target category. The target pricing information of the target commodity, and the target pricing structured information is constructed based on the target pricing information and the target attribute information of the target commodity; after the target pricing structured information is loaded into the pricing factor library, the downstream can be supported to construct the commodity pricing rule information corresponding to the target field according to the pricing structured information loaded in the pricing factor library, thereby realizing the automatic completion of commodity pricing information extraction and ensuring the extraction accuracy, and constructing and storing structured information based on this, which can reasonably and standardizedly construct pricing rule information for the value-added services bound within the field, thereby quickly solving the problem of inconsistent pricing of commodities in different categories. At the same time, the above processing does not require additional human resources to participate, and the extraction of corresponding pricing information of commodities and the construction of structured information can be completed only through the pricing model, which can effectively save development costs and time, thereby making the pricing setting more standardized.

[0098] Corresponding to the above method embodiment, this specification also provides an information processing device embodiment, Figure 4 FIG1 shows a schematic diagram of the structure of an information processing device provided by an embodiment of this specification. Figure 4 As shown, the device includes: An acquisition module 402 is configured to acquire attribute information and pricing information of objects in a target domain, and train a pricing model into a target pricing model based on the attribute information and the pricing information; A screening module 404 is configured to screen pricing structured information that meets the loading conditions and load it into the pricing factor library, wherein the pricing structured information is constructed based on the prediction results of the pricing model in the training phase; A construction module 406 is configured to extract target pricing information of target objects in a target category using the target pricing model, and construct target pricing structured information based on the target pricing information and target attribute information of the target objects; The loading module 408 is configured to load the target pricing structured information into the pricing factor library, wherein the pricing structured information loaded into the pricing factor library is used to construct object pricing rule information corresponding to the target domain.

[0099] In an optional embodiment, the acquiring of attribute information and pricing information of objects in the target domain, and training a pricing model into a target pricing model based on the attribute information and the pricing information, includes: Obtain image information and description information of objects in a target domain, as well as size information and material information of the objects, wherein the image information and the description information constitute attribute information, and the size information and the material information constitute pricing information; input the attribute information into a pricing model for processing to obtain predicted pricing information; adjust the parameters of the pricing model according to the predicted pricing information and the pricing information until a target pricing model that meets the training stop conditions is obtained.

[0100] In an optional embodiment, before the step of adjusting the parameters of the pricing model according to the predicted pricing information and the pricing information is performed, the method further includes: At least one parameter matrix in a multimodal deep learning model is decomposed by a low-rank matrix decomposition algorithm to obtain low-rank matrix parameters; the multimodal deep learning model is updated using the low-rank matrix parameters to obtain an initial pricing model; prompt information to be trained is added to the input layer of the initial pricing model, and the pricing model is generated according to the adding result; wherein the prompt information to be trained includes discrete prompt information and continuous prompt information.

[0101] In an optional embodiment, screening pricing structured information that meets the loading conditions and loading it into the pricing factor library includes: Acquire multiple training sample pairs used by the pricing model in the training phase, wherein the training sample pairs include attribute information and pricing information; determine the prediction result information corresponding to each of the multiple training sample pairs, and filter target training sample pairs that meet the loading conditions from the multiple training sample pairs based on the prediction result information; construct pricing structured information based on the attribute information and pricing information included in the target training sample pairs, and load it into the pricing factor library.

[0102] In an optional embodiment, after the step of loading the target pricing structured information into the pricing factor library is executed, the method further includes: Receive erroneous pricing information submitted for the target pricing model, and determine the associated object corresponding to the erroneous pricing information and the associated attribute information of the associated object; construct an initial model optimization sample pair based on the erroneous pricing information and the associated attribute information, and adjust the initial optimization sample pair to a model optimization sample pair; use the model optimization sample pair to optimize the target pricing model.

[0103] In an optional embodiment, after the step of loading the target pricing structured information into the pricing factor library is executed, the method further includes: Cluster the pricing structured information contained in the pricing factor library according to the object category to obtain an information set corresponding to the object category; and construct object pricing rule information corresponding to the object category based on the pricing information contained in the pricing structured information in the information set and the price information corresponding to the object category.

[0104] In an optional embodiment, before the step of extracting target pricing information of target products in a target category using the target pricing model is performed, the method further includes: Determine multiple candidate object categories and obtain transaction information of candidate objects in each candidate object category; determine category popularity information of each candidate object category based on the transaction information; select a target category from the multiple candidate object categories according to the category popularity information, and execute the step of extracting target pricing information of the target object in the target category using the target pricing model.

[0105] In an optional embodiment, after the step of training the pricing model into a target pricing model according to the attribute information and the pricing information is executed, the method further includes: The target pricing model is verified using the pricing structured information loaded in the pricing factor library; if the verification fails, the target pricing model is optimized using the pricing structured information loaded in the pricing factor library.

[0106] The information processing device provided in this embodiment can accurately extract pricing information for objects in the e-commerce platform, and build structured information storage in combination with the attribute information and pricing information of the objects, so as to facilitate the construction of object pricing rule information for various objects. After obtaining the attribute information and pricing information of the objects in the target field, the pricing model can be trained as a target pricing model based on the attribute information and pricing information, so that the target pricing model can predict its corresponding pricing information in combination with the object attributes. In order to enable downstream services to use high-quality structured information to build object pricing rule information, the pricing structured information that meets the loading conditions can be screened and added to the pricing factor library, wherein the pricing structured information is constructed according to the prediction results of the pricing model in the training stage; on this basis, in order to enrich the structured information loaded in the pricing factor library, the target pricing model can be used to extract the target pricing information. The target pricing information of the target object in the standard category, and the target pricing structured information is constructed according to the target pricing information and the target attribute information of the target object; after the target pricing structured information is loaded into the pricing factor library, it can support the downstream to construct the object pricing rule information corresponding to the target field according to the pricing structured information loaded in the pricing factor library, thereby realizing the automatic completion of the object pricing information extraction, and ensuring the extraction accuracy, and constructing and storing structured information based on this, which can reasonably and standardizedly construct pricing rule information for the value-added services bound in the field, thereby quickly solving the problem of inconsistent pricing of objects in different categories. At the same time, the above processing does not require additional human resources to participate. The extraction of the corresponding pricing information of the object and the construction of structured information can be completed only through the pricing model, which can effectively save development costs and time, thereby making the pricing setting more standardized.

[0107] The above is a schematic diagram of an information processing device according to this embodiment. It should be noted that the technical solution of the information processing device and the technical solution of the above-mentioned information processing method are based on the same concept. For details not described in detail in the technical solution of the information processing device, please refer to the description of the technical solution of the above-mentioned information processing method.

[0108] Corresponding to the above method embodiment, this specification also provides another information processing device embodiment, Figure 5 FIG1 shows a schematic diagram of the structure of another information processing device provided by an embodiment of this specification. Figure 5 As shown, the device is applied to the server and includes: The information receiving module 502 is configured to receive object attribute information of the object to be priced submitted by the client, and input the object attribute information into the target pricing model for processing to obtain object pricing information of the object to be priced; A price generation module 504 is configured to generate a value-added service price corresponding to the object to be priced based on the object pricing rule information corresponding to the object to be priced and the object pricing information, wherein the object pricing rule information is constructed based on the structured information loaded in the pricing factor library in the above method; The price sending module 506 is configured to send the value-added service price to the client for display.

[0109] In summary, when users purchase any type of object with a value-added service price, the pricing information of the object can be automatically extracted through the target pricing model, and the calculation of the value-added service can be completed in combination with the pricing rule information corresponding to the object. This can make it easier for users to understand the value-added service price and object pricing information, thereby improving the user's object purchasing experience.

[0110] The above is a schematic diagram of an information processing device according to this embodiment. It should be noted that the technical solution of the information processing device and the technical solution of the above-mentioned information processing method are based on the same concept. For details not described in detail in the technical solution of the information processing device, please refer to the description of the technical solution of the above-mentioned information processing method.

[0111] Figure 6 6 shows a block diagram of a computing device 600 according to one embodiment of the present disclosure. Components of the computing device 600 include, but are not limited to, a memory 610 and a processor 620. The processor 620 is connected to the memory 610 via a bus 630, and a database 650 is used to store data.

[0112] Computing device 600 also includes an access device 640 that enables computing device 600 to communicate via one or more networks 660. Examples of such networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. Access device 640 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, or a near field communication (NFC) interface.

[0113] In one embodiment of the present specification, the above components of the computing device 600 and Figure 6 Other components not shown in the figure may also be connected to each other, for example, via a bus. Figure 6 The computing device structure block diagram shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art may add or replace other components as needed.

[0114] Computing device 600 can be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, personal digital assistant, laptop computer, notebook computer, netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable computing device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or personal computer (PC). Computing device 600 can also be a mobile or stationary server.

[0115] The processor 620 is configured to execute the following computer-executable instructions, which implement the steps of the above-mentioned information processing method when executed by the processor.

[0116] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above-mentioned information processing method are based on the same concept. For details not described in detail in the technical scheme of the computing device, please refer to the description of the technical scheme of the above-mentioned information processing method.

[0117] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which implement the steps of the above-mentioned information processing method when executed by a processor.

[0118] The above is a schematic diagram of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of the storage medium and the technical solution of the information processing method described above are based on the same concept. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the information processing method described above.

[0119] An embodiment of the present specification further provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned information processing method.

[0120] The above is an illustrative solution of a computer program of this embodiment. It should be noted that the technical solution of the computer program and the technical solution of the above-mentioned information processing method are based on the same concept. For details not described in detail in the technical solution of the computer program, please refer to the description of the technical solution of the above-mentioned information processing method.

[0121] An embodiment of the present specification further provides a computer program product, including a computer program or instructions, which implement the steps of the above-mentioned information processing method when executed by a processor.

[0122] The above is an illustrative solution of a computer program product of this embodiment. It should be noted that the technical solution of this computer program product and the technical solution of the above-mentioned information processing method are based on the same concept. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the above-mentioned information processing method.

[0123] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0124] The computer instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium may be appropriately increased or decreased based on the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0125] 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 be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0126] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0127] The preferred embodiments disclosed above are intended only to help illustrate this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made based on the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. An information processing method, comprising: Acquire attribute information and pricing information of objects in a target domain, and train a pricing model into a target pricing model based on the attribute information and the pricing information; Screening pricing structured information that meets the loading conditions and loading it into the pricing factor library, wherein the pricing structured information is constructed according to the prediction results of the pricing model in the training phase; Extracting target pricing information of target objects in a target category using the target pricing model, and constructing target pricing structured information based on the target pricing information and target attribute information of the target objects; The target pricing structured information is loaded into the pricing factor library, wherein the pricing structured information loaded into the pricing factor library is used to construct object pricing rule information corresponding to the target domain.

2. The information processing method according to claim 1, wherein the step of acquiring attribute information and pricing information of objects in the target domain and training a pricing model into a target pricing model based on the attribute information and the pricing information comprises: Acquire image information and description information of an object in a target area, as well as size information and material information of the object, wherein the image information and description information constitute attribute information, and the size information and material information constitute pricing information; Inputting the attribute information into a pricing model for processing to obtain predicted pricing information; The pricing model is adjusted according to the predicted pricing information and the pricing information until a target pricing model that meets the training stop condition is obtained.

3. The information processing method according to claim 2, before performing the step of adjusting the parameters of the pricing model according to the predicted pricing information and the pricing information, further comprising: Decomposing at least one parameter matrix in the multimodal deep learning model by a low-rank matrix decomposition algorithm to obtain low-rank matrix parameters; Using the low-rank matrix parameters to update the multimodal deep learning model to obtain an initial pricing model; Prompt information to be trained is added to the input layer of the initial pricing model, and the pricing model is generated according to the adding result; wherein the prompt information to be trained includes discrete prompt information and continuous prompt information.

4. The information processing method according to claim 1, wherein the step of screening the pricing structured information that meets the loading conditions and loading it into the pricing factor library comprises: Acquire a plurality of training sample pairs used by the pricing model in a training phase, wherein the training sample pairs include attribute information and pricing information; Determining prediction result information corresponding to each of the plurality of training sample pairs, and selecting a target training sample pair that meets a loading condition from the plurality of training sample pairs according to the prediction result information; Pricing structured information is constructed based on the attribute information and pricing information contained in the target training sample pair, and loaded into a pricing factor library.

5. The information processing method according to any one of claims 1 to 4, further comprising: after the step of loading the target pricing structured information into the pricing factor library is executed: receiving erroneous pricing information submitted for the target pricing model, and determining an associated object corresponding to the erroneous pricing information and associated attribute information of the associated object; Constructing an initial model optimization sample pair based on the erroneous pricing information and the associated attribute information, and adjusting the initial optimization sample pair to a model optimization sample pair; The target pricing model is optimized using the model optimization sample pairs.

6. The information processing method according to any one of claims 1 to 4, further comprising: after the step of loading the target pricing structured information into the pricing factor library is executed: Clustering the pricing structured information contained in the pricing factor library according to object categories to obtain an information set corresponding to the object categories; According to the pricing information included in the pricing structured information in the information set and the price information corresponding to the object category, object pricing rule information corresponding to the object category is constructed.

7. The information processing method according to any one of claims 1 to 4, before executing the step of extracting target pricing information of target objects in the target category using the target pricing model, further comprising: Determine multiple candidate object categories and obtain transaction information of candidate objects in each candidate object category; Determine category popularity information of each candidate category based on the transaction information; A target category is selected from the multiple candidate object categories according to the category popularity information, and the target pricing information of the target object in the target category is extracted by using the target pricing model.

8. The information processing method according to any one of claims 1 to 4, further comprising: after executing the step of training the pricing model into a target pricing model based on the attribute information and the pricing information; Verifying the target pricing model using the pricing structured information loaded in the pricing factor library; In the case that the verification fails, the target pricing model is optimized using the pricing structured information loaded in the pricing factor library.

9. An information processing method, applied to a server, comprising: Receive object attribute information of the object to be priced submitted by the client, and input the object attribute information into the target pricing model for processing to obtain object pricing information of the object to be priced; generating a value-added service price corresponding to the object to be priced based on object pricing rule information corresponding to the object to be priced and the object pricing information, wherein the object pricing rule information is constructed based on the structured information loaded in the pricing factor library in the method according to any one of claims 1 to 8; The value-added service price is sent to the client for display.

10. An information processing device comprising: an acquisition module configured to acquire attribute information and pricing information of objects in a target domain, and train a pricing model into a target pricing model based on the attribute information and the pricing information; a screening module configured to screen pricing structured information that meets the loading conditions and load it into the pricing factor library, wherein the pricing structured information is constructed based on the prediction results of the pricing model in the training phase; a construction module configured to extract target pricing information of target objects in a target category using the target pricing model, and construct target pricing structured information based on the target pricing information and target attribute information of the target objects; The loading module is configured to load the target pricing structured information into the pricing factor library, wherein the pricing structured information loaded into the pricing factor library is used to construct object pricing rule information corresponding to the target domain.

11. An information processing device, applied to a server, comprising: An information receiving module is configured to receive object attribute information of an object to be priced submitted by a client, and input the object attribute information into a target pricing model for processing to obtain object pricing information of the object to be priced; a price generation module configured to generate a value-added service price corresponding to the object to be priced based on object pricing rule information corresponding to the object to be priced and the object pricing information, wherein the object pricing rule information is constructed based on the structured information loaded in the pricing factor library in the method according to any one of claims 1 to 8; The price sending module is configured to send the value-added service price to the client for display.

12. A computing device comprising: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 9 are implemented.

13. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, implement the steps of the method according to any one of claims 1 to 9.

14. A computer program product comprising a computer program or instructions, which implement the steps of the method according to any one of claims 1 to 9 when executed by a processor.

Images