A data recommendation method and device based on model
By cleaning and feature extraction of user historical data, combining recursive feature elimination method and Bayesian network, predictive feature vectors are generated and inputted into prediction models, the problem of low accuracy of data recommendation in the existing technology is solved, and users are significantly improved Experience.
Patent Information
- Application Number
- CN202411141762.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-08-20
AI Technical Summary
The accuracy of data recommendation in the prior art is not high, resulting in poor user experience.
By obtaining the user's historical data, performing data cleaning, extracting feature sets and iteratively training the preset model, using recursive feature elimination method and Bayesian network to generate predicted feature vectors, and finally input the features into the prediction model to improve the recommendation accuracy .
The accuracy of data recommendation and user experience are improved. Through the combination of recursive feature elimination and Bayesian network, feature extraction and model training are optimized, and the accuracy of recommended results is enhanced.
Smart Images

Figure CN118656546B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of computer technology, and in particular relates to a model-based data recommendation method and device. Background Art
[0002] With the development of the Internet, the amount of data on the Internet is increasing. Users have different browsing habits when browsing the Internet. Internet service providers need to recommend data content that may be of interest to users based on their habits and preferences. In order to solve this type of problem, the prior art provides a method for recommending data based on a mathematical model. However, the mathematical model in the prior art can only use the data in the data set, and the accuracy of the recommendation for users is not high.
[0003] Therefore, how to improve the accuracy of data recommendations and enhance the user experience is a technical problem to be solved by those skilled in the art. Summary of the invention
[0004] The purpose of the present invention is to solve the technical problems in the prior art of low accuracy of data recommendation and low user experience.
[0005] To achieve the above technical objectives, on the one hand, the present invention provides a model-based data recommendation method, the method comprising:
[0006] Acquire historical data of user browsing, and generate first recommendation data according to the historical data through a preset model;
[0007] Extracting a feature set based on the historical data and the first recommended data;
[0008] Iteratively training a preset model based on the feature set to obtain a prediction model;
[0009] A first feature is extracted based on the historical data, and the first feature is input into the prediction model to obtain a recommendation result.
[0010] Furthermore, before extracting the feature set based on the historical data and the first recommended data, the method also includes performing data cleaning processing on the historical data and the first recommended data, and the data cleaning processing specifically includes removing interference data and abnormal data.
[0011] Furthermore, the prediction model is obtained by iteratively training the preset model based on the feature set, specifically using a recursive feature elimination method to iteratively train the preset model. Specifically, at the end of each generation of training, the features that do not satisfy the objective function are removed from the feature set by the recursive feature elimination method, and new features are formed and incorporated into the feature set to train the preset model in the next generation of training.
[0012] Furthermore, the forming of new features specifically includes:
[0013] Characterize the user behavior according to the feature set trained in the contemporary era to obtain a characterization feature vector;
[0014] Inputting the characterization feature vector into a Bayesian network to obtain a prediction feature vector;
[0015] The predicted feature vector is used as a new feature.
[0016] Furthermore, the first feature specifically includes item numerical features, user numerical features and numerical association features.
[0017] Furthermore, the step of inputting the first feature into the prediction model to obtain a recommendation result specifically includes:
[0018] Inputting the first feature into a prediction model to obtain a plurality of initial recommendation results;
[0019] All initial recommendation results are compared pairwise, and the initial recommendation result with the highest number of recommendations is output as the recommendation result.
[0020] On the other hand, the present invention also provides a model-based data recommendation device, the device comprising:
[0021] An acquisition module, used to acquire historical data of user browsing, and generate first recommendation data according to the historical data through a preset model;
[0022] An extraction module, configured to extract a feature set based on the historical data and the first recommended data;
[0023] A training module, used for iteratively training a preset model based on the feature set to obtain a prediction model;
[0024] The recommendation module is used to extract a first feature based on the historical data and input the first feature into the prediction model to obtain a recommendation result.
[0025] The present invention provides a model-based data recommendation method and device. Compared with the prior art, the method first obtains the historical data browsed by the user, and generates first recommendation data according to the historical data through a preset model; then extracts a feature set based on the historical data and the first recommendation data; then iteratively trains the preset model based on the feature set to obtain a prediction model; finally, extracts the first feature based on the historical data, and inputs the first feature into the prediction model to obtain a recommendation result, which can improve the accuracy of data recommendation and improve the user experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0027] Figure 1 The figure is a flow chart of a model-based data recommendation method provided in an embodiment of this specification;
[0028] Figure 2 The figure is a schematic diagram of the structure of a model-based data recommendation device provided in an embodiment of this specification. DETAILED DESCRIPTION
[0029] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.
[0030] like Figure 1 The flowchart of the model-based data recommendation method provided by the embodiment of this specification is shown. Although this specification provides the method operation steps or device structures shown in the following embodiments or drawings, the method or device may include more or fewer operation steps or module units after partial merger based on routine or no creative labor. In the steps or structures where there is no necessary causal relationship logically, the execution order of these steps or the module structure of the device is not limited to the execution order or module structure shown in the embodiments or drawings of this specification. When the method or module structure is applied in actual devices, servers or terminal products, it can be executed sequentially or in parallel according to the method or module structure shown in the embodiments or drawings (for example, in a parallel processor or multi-threaded processing environment, or even in a distributed processing, server cluster implementation environment).
[0031] The model-based data recommendation method provided in the embodiments of this description can be applied to terminal devices such as clients and servers. Figure 1 As shown, the method specifically comprises the following steps:
[0032] Step S101: Acquire historical data of user browsing, and generate first recommendation data according to the historical data through a preset model.
[0033] Specifically, keywords are extracted from historical data such as news, videos, and articles that users have browsed and then input into a preset model. The preset model may be a large language model (LLM). The preset model is used to generate first recommended data, which are data that users have not seen but may be interested in. These data are also used as part of the training data.
[0034] Step S102: extracting a feature set based on the historical data and the first recommended data.
[0035] Specifically, the historical data and the first recommended data are first subjected to data cleaning processing, and the data cleaning processing specifically includes removing interference data and abnormal data. Specifically, it can be to first add regularization filtering to normalize the text, remove hypertext markup language and tags, filter stop words and noise words, correct spelling errors and clear sensitive information, that is, remove interference data, and then use an artificial intelligence model to clean and remove abnormal data, and finally eliminate bad data based on the Bayesian algorithm.
[0036] The historical data and the first recommended data after the above processing are subjected to feature transformation, feature selection and feature preprocessing in sequence to extract a feature set. The feature preprocessing specifically includes normalization, standardization, robustness and feature combination.
[0037] Step S103: iteratively train the preset model based on the feature set to obtain a prediction model.
[0038] Specifically, the prediction model is obtained by iteratively training the preset model based on the feature set, specifically using a recursive feature elimination method to iteratively train the preset model, specifically at the end of each generation of training, using the recursive feature elimination method to remove features that do not meet the objective function in the feature set, and forming new features and incorporating them into the feature set to train the preset model in the next generation of training.
[0039] The new features include:
[0040] Characterize the user behavior according to the feature set trained in the contemporary era to obtain a characterization feature vector;
[0041] Inputting the characterization feature vector into a Bayesian network to obtain a prediction feature vector;
[0042] The predicted feature vector is used as a new feature.
[0043] Select or exclude some features each time. Use the recursive feature elimination method based on LLM to perform multiple rounds of training on the preset model, eliminating some features each time and then training again based on the new features.
[0044] In a specific application scenario, the following steps can be used:
[0045] 1. Based on the historical data or feature sets browsed by users, the Large Language Model Reasoning Graphs (LLMRG) is designed. The core module of the LLM-based Adaptive Reasoning Architecture (LLM-based ARA) is the LLM-based adaptive reasoning architecture. The model is used to reason out the user historical feature data with logical correlation, that is, the user behavior. After processing it, the feature vector describing the user behavior is obtained. The user behavior can also be processed to obtain the behavioral reasoning feature. There is an intrinsic logical association between the key-value vector and the available vector between the behavioral reasoning features.
[0046] 2. Input the characterization feature vector into the Bayesian network to obtain the prediction feature vector, that is, the new feature. The behavior reasoning feature mentioned above can also be further reasoned through a feasible neural network model to obtain the prediction reasoning feature, and then through the scoring of the vector database, the prediction reasoning features with higher scores, such as the top 100, can be used as the new feature.
[0047] The elimination of some features mentioned above means that at the end of each generation of training, the features that are not satisfactory to the objective function are removed from the feature set to complete the elimination of old features.
[0048] Step S104: extract a first feature based on the historical data, and input the first feature into the prediction model to obtain a recommendation result.
[0049] Specifically, the first feature includes item numerical features, user numerical features and numerical association features. Item numerical features include the number of words in an article, the number of words in a news article, the time when an article is published and the time when a news article is published. Item numerical features are mixed with item text features. Item text features include historical articles, historical news, historical article categories and historical news categories. User numerical features include the number of user comments, the number of user likes and the daily browsing time of a user. User numerical features are mixed with user text features. User text features include browsing titles, click operations and download operations. Numerical association features include geographic location and weather temperature.
[0050] Inputting the first feature into the prediction model to obtain a recommendation result specifically includes:
[0051] Inputting the first feature into a prediction model to obtain a plurality of initial recommendation results;
[0052] All initial recommendation results are compared pairwise, and the initial recommendation result with the highest number of recommendations is output as the recommendation result.
[0053] Specifically, LLM as ranker is adopted, and the implementation strategy adopts pair-wise ranking. Each time two initial recommendation results are selected and input into the prediction model, the prediction model will score the initial recommendation results input each time, that is, one initial recommendation result will be recommended once, and then after all the initial recommendation results are compared with each other, the initial recommendation result with the highest number of recommendations will be output as the recommendation result; after all the initial recommendation results are compared with each other, the top n ranked initial recommendation results can be retained, and then the top n initial recommendation results can be expanded through hyperlinks to obtain the corresponding extended documents, and then all the extended documents are scored through the prediction model, and the initial recommendation result corresponding to the extended document with the highest score will be output as the recommendation result, which can more truly meet the user's preferences.
[0054] Based on the above-mentioned model-based data recommendation method, one or more embodiments of this specification also provide a model-based data recommendation platform and terminal. The platform or terminal may include devices, software, modules, plug-ins, servers, clients, etc. that use the methods described in the embodiments of this specification and are combined with necessary implementation hardware devices. Based on the same innovative concept, the system in one or more embodiments provided in the embodiments of this specification is as described in the following embodiments. Since the implementation scheme and method for solving the problem of the system are similar, the implementation of the specific system in the embodiments of this specification can refer to the implementation of the aforementioned method, and the repetitions will not be repeated. The terms "unit" or "module" used below can be a combination of software and / or hardware that implements predetermined functions. Although the system described in the following embodiments is preferably implemented in software, hardware and a combination of software and hardware are also possible and conceived.
[0055] Specifically, Figure 2 is a schematic diagram of the module structure of an embodiment of a model-based data recommendation device provided in this specification, such as Figure 2 As shown, the model-based data recommendation device provided in this specification includes:
[0056] An acquisition module 201 is used to acquire historical data of user browsing and generate first recommendation data according to the historical data through a preset model;
[0057] An extraction module 202, configured to extract a feature set based on the historical data and the first recommendation data;
[0058] A training module 203 is used to iteratively train a preset model based on the feature set to obtain a prediction model;
[0059] The recommendation module 204 is used to extract a first feature based on the historical data, and input the first feature into the prediction model to obtain a recommendation result.
[0060] It should be noted that the above-mentioned system may also include other implementation modes according to the description of the corresponding method embodiment. The specific implementation modes may refer to the description of the above-mentioned corresponding method embodiment, which will not be described one by one here.
[0061] The present application also provides an electronic device, including:
[0062] processor;
[0063] a memory for storing instructions executable by the processor;
[0064] The processor is configured to execute the method provided in the above embodiment.
[0065] The electronic device provided in the embodiment of the present application stores executable instructions of the processor through a memory. When the processor executes the executable instructions, it can first obtain the historical data browsed by the user, and generate first recommendation data according to the historical data through a preset model; then extract a feature set based on the historical data and the first recommendation data; then iteratively train the preset model based on the feature set to obtain a prediction model; finally, extract the first feature based on the historical data, and input the first feature into the prediction model to obtain a recommendation result, which can improve the accuracy of data recommendation and improve the user experience.
[0066] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0067] The methods or devices described in the above embodiments provided in this specification can implement business logic through computer programs and record them on storage media, and the storage media can be read and executed by computers to achieve the effects of the solutions described in the embodiments of this specification, such as:
[0068] Acquire historical data of user browsing, and generate first recommendation data according to the historical data through a preset model;
[0069] Extracting a feature set based on the historical data and the first recommended data;
[0070] Iteratively training a preset model based on the feature set to obtain a prediction model;
[0071] A first feature is extracted based on the historical data, and the first feature is input into the prediction model to obtain a recommendation result.
[0072] The storage medium may include a physical device for storing information, which is usually a medium that digitizes information and then stores it in an electrical, magnetic or optical manner. The storage medium may include: a device that stores information in an electrical energy manner, such as various memories, such as RAM, ROM, etc.; a device that stores information in a magnetic energy manner, such as a hard disk, a floppy disk, a magnetic tape, a magnetic core memory, a magnetic bubble memory, a USB flash drive; a device that stores information in an optical manner, such as a CD or a DVD. Of course, there are other readable storage media, such as quantum memory, graphene memory, etc.
[0073] The embodiments of this specification are not limited to complying with industry communication standards, standard computer resource data update and data storage rules, or the situations described in one or more embodiments of this specification. Certain industry standards or slightly modified implementation plans based on the implementation described in the custom method or embodiment can also achieve the same, equivalent or similar, or predictable implementation effects after deformation of the above-mentioned embodiments. The embodiments obtained by using these modified or deformed data acquisition, storage, judgment, processing methods, etc. can still fall within the scope of the optional implementation plans of the embodiments of this specification.
[0074] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a computer-readable medium such as a microprocessor or processor and a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, ATMEL AT91SAM, MICROCHIP PIC18F26K20, and SILICONE LABS C8051F320. The memory controller can also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software modules for implementing the method and structures within the hardware component.
[0075] The device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. There may be other divisions in actual implementation, such as multiple units or plug-ins may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some interface, indirect coupling or communication connection of the device or unit, which may be electrical, mechanical or other forms.
[0076] These computer program instructions may also be loaded onto a computer or other programmable resource data updating device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions executed on the computer or other programmable device for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0077] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. In the description of this specification, the description of the reference terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of this specification. In addition, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.
[0078] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.
Claims
1. A model-based data recommendation method, It is characterized in that The method comprises: Acquire historical data of user browsing, and generate first recommendation data according to the historical data through a preset model; Extracting a feature set based on the historical data and the first recommended data; Iteratively training a preset model based on the feature set to obtain a prediction model; Extracting a first feature based on the historical data, and inputting the first feature into the prediction model to obtain a recommendation result, wherein the first feature specifically includes an item value feature, a user value feature, and a value association feature; The step of inputting the first feature into the prediction model to obtain a recommendation result specifically includes: Inputting the first feature into a prediction model to obtain a plurality of initial recommendation results; All initial recommendation results are compared pairwise, and the initial recommendation result with the highest number of recommendations is output as the recommendation result.
2. The model-based data recommendation method according to claim 1, It is characterized in that Before extracting the feature set based on the historical data and the first recommended data, the method further includes performing data cleaning processing on the historical data and the first recommended data, wherein the data cleaning processing specifically includes removing interference data and abnormal data.
3. The model-based data recommendation method according to claim 1, It is characterized in that The prediction model is obtained by iteratively training the preset model based on the feature set, specifically using a recursive feature elimination method to iteratively train the preset model. Specifically, at the end of each generation of training, the features that do not meet the objective function are removed from the feature set by the recursive feature elimination method, and new features are formed and incorporated into the feature set to train the preset model in the next generation of training.
4. The model-based data recommendation method according to claim 3, It is characterized in that The forming of new features specifically includes: Characterize the user behavior according to the feature set trained in the contemporary era to obtain a characterization feature vector; Inputting the characterization feature vector into a Bayesian network to obtain a prediction feature vector; The predicted feature vector is used as a new feature.
5. A data recommendation device based on a model, It is characterized in that The device comprises: An acquisition module, used to acquire historical data of user browsing, and generate first recommendation data according to the historical data through a preset model; An extraction module, configured to extract a feature set based on the historical data and the first recommended data; A training module, used for iteratively training a preset model based on the feature set to obtain a prediction model; A recommendation module, configured to extract a first feature based on the historical data, and input the first feature into the prediction model to obtain a recommendation result, wherein the first feature specifically includes an item numerical feature, a user numerical feature, and a numerical association feature; The step of inputting the first feature into the prediction model to obtain a recommendation result specifically includes: Inputting the first feature into a prediction model to obtain a plurality of initial recommendation results; All initial recommendation results are compared pairwise, and the initial recommendation result with the highest number of recommendations is output as the recommendation result.
