Personalized recommendation method in robot intelligent aided design system

Abstract

The invention provides a personalized recommendation method in a robot intelligent aided design system, and belongs to the field of data source construction and analysis. Extracting a design function, a model type, a model parameter, a user operation score and a user search record; labeling a supervision label; constructing a training data set, and pre-training the user function recommendation model and the personalized recommendation model; calculating a function score, and calculating a personalized recommendation score; combining the function score and the personalized recommendation score to generate a final project function recommendation list; pushing an item function recommendation list to the user; verifying the correctness of the user operation score, and updating the user operation score and the user search record by adopting a reinforcement learning mode; updating the user function recommendation model and the personalized recommendation model; and the item function recommendation list is regenerated and repeated until the convergence value of the user operation scoring accuracy is higher than the preset threshold value, so that the accuracy and the flexibility of the personalized recommendation system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data source construction and analysis for personalized recommendation, and in particular to a personalized recommendation method in a robot intelligent-aided design system. Background Art

[0002] With the rapid development of artificial intelligence and machine learning, personalized recommendation systems have become an important tool in all walks of life and are widely used in e-commerce, social media, and online content platforms. Personalized recommendation systems use users' historical behavior data, preferences, and interests to analyze and mine potential patterns to provide users with customized product or service recommendations. In recent years, the introduction of deep learning technology has greatly improved the performance of recommendation systems, but they still need to be continuously optimized to meet the increasingly diverse needs of users.

[0003] Traditional personalized recommendation technologies mainly rely on collaborative filtering and content recommendation methods. Collaborative filtering makes recommendations based on the similarity of user behaviors, while content recommendation relies on matching project features with user interests. These methods often use simple scoring models combined with basic statistics and similarity metrics for recommendations. However, these traditional methods often face performance bottlenecks when dealing with high-dimensional data and complex user behaviors, and are unable to fully tap into user potential needs. At the same time, traditional technologies lack dynamic learning capabilities and cannot update user models in real time to adapt to changing preferences, resulting in inaccurate and in-time recommendation results. Summary of the invention

[0004] The technical problem to be solved by the present invention is how to improve the accuracy and flexibility of the personalized recommendation system.

[0005] The present invention solves the above technical problems through the following technical solutions: a personalized recommendation method in a robot intelligent assisted design system, comprising:

[0006] Step 1: Extract design projects from user history logs;

[0007] Step 2: Extract the design functions, model types, model parameters, user operation scores, and user search records in the design projects to construct a data set;

[0008] Step 3: Label the supervision label based on the user operation score;

[0009] Step 4: Use the data set and the supervision label to construct a training data set, and pre-train the user function recommendation model and the personalized recommendation model based on the training data set to obtain the pre-trained user function recommendation model and the personalized recommendation model;

[0010] Step 5: Use the pre-trained user function recommendation model to calculate the function score for the design projects in the user's history, and use the pre-trained personalized recommendation model to calculate the personalized recommendation score for the design projects in the user's history;

[0011] Step 6: Combine the function score and the personalized recommendation score to generate the final project function recommendation list;

[0012] Step 7: Push a list of project function recommendations to users;

[0013] Step 8: Verify the correctness of the user operation score, and use reinforcement learning to update the user operation score and user search history;

[0014] Step 9: Use the updated user operation scores and user search records to update the user function recommendation model and the personalized recommendation model respectively;

[0015] Step 10: Use the updated user function recommendation model and personalized recommendation model to regenerate the project function recommendation list, and repeat steps 7 to 10 until the convergence value of the user operation score accuracy rate is higher than the preset threshold.

[0016] The present invention establishes a more comprehensive user portrait by comprehensively analyzing the user's historical logs, ratings and search records, which can improve the accuracy of recommendations. It labels supervisory tags based on user operation ratings, and trains user function recommendation models and personalized recommendation models based on data extracted from historical logs and supervisory tags, respectively. It can effectively capture the relationship between complex user preferences and designed functions, and solve the problem that existing methods often cannot effectively explore users' potential preferences when processing user historical data, resulting in inaccurate and timely recommendation results. It adopts reinforcement learning to dynamically update user operation ratings and user search records, and continuously adjusts recommendation strategies based on user feedback, so that the system can adapt to users' changing needs in real time, improve the accuracy and flexibility of personalized recommendation systems, and significantly improve user experience and satisfaction.

[0017] Preferably, in step 2, the extraction of design functions is based on the basic modules involved in the design project, and the functions implemented by the basic modules are determined by analyzing the basic modules included in the design project; the determination of model parameters is based on the model parameter types established for each design project and the design functions corresponding thereto; the user operation scores are obtained by analyzing the model types and model parameters in the design projects; and the user search records are established by tracking the user's operational behaviors on the design process, recording the user's search keywords and frequency during use.

[0018] Preferably, the supervision label in step 3 is a set of design functions in the design project, expressed as:

[0019] D(i)={f1 ,f 2 ,...,f k}

[0020] Where D(i) is the set of design functions for the i-th design project, and f j is a specific design function, j = 1, 2, ..., k.

[0021] Preferably, the training data set in step 4 is represented as:

[0022] T={F(i),MT(i),MP(i),U(i),R(i),D(i)}

[0023] Where T represents the training dataset, F(i) represents the design function of the i-th design project, MT(i) represents the model type of the i-th design project, MT(i) represents the model parameter of the i-th design project, U(i) represents the user operation score of the i-th design project, R(i) is the user search record of the i-th design project, and D(i) is the supervision label of the i-th design project.

[0024] Preferably, in step 4, a deep neural network is used to pre-train the user function recommendation model, and the pre-training process includes:

[0025] 4.1.1. Extract design features, model types, model parameters, user operation scores, and user search records, and input them into the deep neural network N to predict the supervision label The calculation formula is:

[0026]

[0027] in, is the supervised label predicted by the deep neural network, F(i) represents the design function of the i-th design project, MT(i) represents the model type of the i-th design project, MP(i) represents the model parameter of the i-th design project, U(i) represents the user operation score of the i-th design project, R(i) is the user search record of the i-th design project, and N is the deep neural network;

[0028] 4.1.2. Calculate the loss function. When the loss function is the smallest, stop training and use the back propagation algorithm to update all parameters of the deep neural network N to obtain the pre-trained user function recommendation model. The calculation formula of the loss function is:

[0029]

[0030] Where L represents the loss function of the pre-training of the user feature recommendation model, and D j (i) is the actual supervision label, is the supervised label predicted by the deep neural network, and n is the number of supervised labels.

[0031] Preferably, in step 4, a deep neural network is used to pre-train the personalized recommendation model, and the pre-training process includes:

[0032] 4.2.1. Input the model type MT(i) into the personalized recommendation model P and predict the personalized score S(i):

[0033] S(i)=P(MT(i))

[0034] Among them, S(i) represents the predicted personalized score, and P is the personalized recommendation model;

[0035] 4.2.2. Input the user search record R(i) and user behavior characteristics B(i) into the personalized recommendation model P to obtain the user item preference sequence P(i):

[0036] P(i)=P(R(i),B(i))

[0037] 4.2.3. Calculate the loss function. When the loss function is the smallest, stop training and use the back propagation algorithm to update all parameters of the deep neural network N to obtain the pre-trained personalized recommendation model. The calculation formula of the loss function is:

[0038]

[0039] Among them, L p represents the loss function of the personalized recommendation model pre-training, U k (i) is the actual user operation score, S k (i) is the predicted personalized score, w k is the weight calculated based on the user-item preference sequence, and m is the number of ratings.

[0040] Preferably, the output of the deep neural network is expressed as N(X), where X represents input data, and the input data X includes design functions, model types, model parameters, user operation scores, and user search records. Feature extraction is performed through multiple hidden layers, and the output of the tth hidden layer can be expressed as:

[0041] N(X)=H t (W t ·H t-1 (W t-1 …H 1 (W 1 ·X+b 1 )+b 2 )+…+b t )

[0042] Among them, Ht is the output of the tth hidden layer, W t is the weight of the tth hidden layer, b t is the corresponding bias;

[0043] The calculation formula for defining the loss function L is:

[0044]

[0045] Among them, D j is the actual supervision label, N(X j ) is the network output, n is the number of samples;

[0046] After calculating the loss function, we get the gradient of the model parameters. We use the chain rule to calculate the derivative of the loss function for each weight and bias. The update rule is:

[0047]

[0048] Among them, η is the learning rate, which controls the update step size.

[0049] Preferably, the specific process of calculating the functional score in step 5 includes:

[0050] 5.1. Evaluate the recommended score of each design function in the design project. Add up the recommended scores of all design functions and take the average to get the function score. The calculation formula for the recommended score of each design function is:

[0051]

[0052] Among them, S i,n represents the recommendation score of the nth design feature of the ith design project, w j is the design weight of the design item for each training sample, F j (i) is the design function of the jth training sample in the i-th design project;

[0053] 5.2. Input the user search records R(i) and user behavior features B(i) into the trained personalized recommendation model P, obtain the user project preference sequence P(i) for each design function, and calculate the preference weighted score P of all design projects. The calculation formula is:

[0054]

[0055] Among them, O(i) is the output of the deep neural network, W i is the preference selection weight of the i-th design project;

[0056] 5.3. Update the preference selection weights. The update formula is: W new =Wold +α·(RW old ), where R represents the user’s actual preference for the model and α is the preset update rate.

[0057] Preferably, the process of updating the user operation score and the user search record by using the reinforcement learning method in step eight includes:

[0058] 8.1. Integrate user search history, behavior, and user operation scores into reinforcement learning states, select recommended items as actions, calculate user preference weighted scores, and define reward functions and state value functions;

[0059] 8.2. Compare the results output by the recommendation model with actual user feedback to determine whether the recommended design function is effective;

[0060] 8.3. Calculate user feedback and update the model;

[0061] 8.4. Dynamically adjust user behavior and preferences.

[0062] Preferably, the process of updating the user function recommendation model and the personalized recommendation model in step nine includes:

[0063] 9.1. Input user search history, behavior data and rating information into the personalized recommendation model;

[0064] 9.2. The personalized recommendation model calculates the user preference weighted set, compares the output of the personalized recommendation model with the actual score, and calculates the loss function. The calculation formula of the loss function L is:

[0065]

[0066] Among them, Y k Score actual user actions, is the score predicted by the personalized recommendation model, and n is the number of samples;

[0067] 9.3. Use the gradient descent method to adjust the parameters of the user function recommendation model. The update method of the model parameters is:

[0068]

[0069] Among them, θ represents the model parameters to be updated, It is the derivative of the loss function with respect to the parameters, reflecting the impact of the current parameters on the loss;

[0070] 9.4. Update the score of each design feature.

[0071] The advantages provided by the present invention are:

[0072] (1) The present invention establishes a more comprehensive user portrait by comprehensively analyzing the user's historical logs, ratings and search records, which can improve the accuracy of recommendations. It labels supervisory tags based on user operation ratings, and trains user function recommendation models and personalized recommendation models based on data extracted from historical logs and supervisory tags, respectively. It can effectively capture the relationship between complex user preferences and designed functions, and solve the problem that existing methods often fail to effectively explore users' potential preferences when processing user historical data, resulting in inaccurate and timely recommendation results. It uses reinforcement learning to dynamically update user operation ratings and user search records, and continuously adjusts recommendation strategies based on user feedback, so that the system can adapt to users' changing needs in real time, improve the accuracy and flexibility of personalized recommendation systems, and significantly improve user experience and satisfaction.

[0073] (2) The present invention uses the set of design functions in the design project as the supervision label. The user operation score reflects the user's overall satisfaction with the design project. The supervision label characterizes the specific performance of the design project through the functional characteristics of the design project. In this way, the relationship between the user operation score and the design function is established to form a training data set in supervised learning. The user operation score provides a target value for supervised learning, and the supervision label is the set of design functions related to the user operation score. During the training process, the user operation score and the supervision label together constitute the input and output, helping the model learn how to predict user satisfaction based on the different performances of the design function, thereby providing personalized design function suggestions in subsequent recommendations.

[0074] (3) In the process of training the personalized recommendation model using a deep neural network, the loss function includes weights calculated based on the user's item preference sequence. Through back propagation and optimization, the model will automatically learn the preference information related to the user's item preference sequence.

[0075] (4) The present invention uses a deep neural network to train the user function recommendation model and the personalized recommendation model respectively, so that the recommendation system has the ability of self-learning and realizes continuous learning and self-optimization. Reinforcement learning enables the system to learn in each user interaction. By continuously collecting user feedback and behavior data, the system can continuously optimize model parameters and recommendation strategies to adapt to changes in user preferences. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 A flowchart of a personalized recommendation method in a robot intelligent assisted design system provided by an embodiment of the present invention;

[0077] Figure 2 A flowchart of constructing a training data set in a personalized recommendation method in a robot intelligent assisted design system provided by an embodiment of the present invention;

[0078] Figure 3 A sub-flow chart of a personalized recommendation method in a robot intelligent assisted design system provided by an embodiment of the present invention;

[0079] Figure 4 A sub-flow chart of a personalized recommendation method in a robot intelligent assisted design system provided by an embodiment of the present invention;

[0080] Figure 5 A sub-flow chart of a personalized recommendation method in a robot intelligent assisted design system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0081] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the technical solution of the present invention is clearly and completely described below in combination with specific embodiments and with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0082] like Figure 1 As shown, this embodiment provides a personalized recommendation method in a robot intelligent assisted design system, comprising the following steps:

[0083] Step 1: Extract design projects from user history logs.

[0084] Step 2: Extract the design functions, model types, model parameters, user operation scores, and user search records in the design projects and build a data set.

[0085] The extraction of design functions is based on the basic modules involved in the design project. By analyzing the basic modules contained in the design project, the functions implemented by the basic modules are determined. For example, the design function of the i-th design project can be expressed as F(i) = {p1, p2, ..., pn}, where F(i) is the design function of the i-th design project, pj is the corresponding model parameter type, and j = 1, 2, ..., n.

[0086] The determination of model parameters depends on the type of model parameters established for each design project and its corresponding design function.

[0087] By analyzing the model type and model parameters in the design project to obtain the user operation score, the user operation score of the i-th design project can be expressed as:

[0088] U(i)=g(MT(i),MP(i))

[0089] Among them, U(i) is the user operation score of the i-th design project, MT(i) represents the model type of the i-th design project, MP(i) represents the model parameters of the i-th design project, and g is a scoring function used to map the model type and model parameters to the user's satisfaction score.

[0090] The user search record is established by tracking the user's operational behavior on the design process, recording the user's search keywords and frequency during use. The user search record R(i) of the i-th design project can be expressed as R(i) = {k1, k2, ..., km}, where R(i) is the user search record of the i-th design project, kj is the keyword searched by the user, and j = 1, 2, ..., m.

[0091] For example, if the basic modules involved in a design project are sensors and actuators, the corresponding design functions may include data acquisition and motion control, the model parameters may be the sensitivity of the sensor and the response time of the actuator, the user operation score may be 5 points, indicating the user's satisfaction with the design project, and the user search history may show that the user frequently searches for sensor types and actuator options. By integrating this information, a data set can be constructed to facilitate subsequent model training and optimization.

[0092] Step 3: Label the supervision labels based on user operation scores.

[0093] The supervision label is the set of design features in a design project and can be expressed as:

[0094] D(i)={f 1 ,f 2 ,...,f k}

[0095] Where D(i) is the set of design functions for the i-th design project, and f j is a specific design function, j = 1, 2, ..., k.

[0096] For example, if the model type of a design project is "linear regression", the model parameters are "learning rate 0.01, iteration number 1000", and the user operation score may be 4.5, indicating that the user is highly satisfied with the design project. At this time, the supervisory label D(i) may include "prediction accuracy", "model stability" and "computational efficiency", and these three design functions constitute the functional characteristics of the design project. The user operation score U(i) reflects the user's overall satisfaction with the design project. The supervisory label D(i) characterizes the specific performance of the design project through the functional characteristics of the design project. In this way, the relationship between the user operation score U(i) and the design function is established to form a training data set in supervised learning. The user operation score U(i) provides a target value for supervised learning, and the supervisory label D(i) is a set of design functions related to the user operation score U(i). During the training process, the user operation score U(i) and the supervisory label D(i) together constitute the input and output, helping the model learn how to predict user satisfaction based on the different performances of the design function, thereby providing personalized design function suggestions in subsequent recommendations. User operation scores and their corresponding supervisory labels will be used for subsequent model training and optimization to ensure that the recommendation system can provide personalized design function recommendations based on the user's historical interaction behavior.

[0097] Step 4: Use the data set constructed in step 2 and the supervision labels obtained in step 3 to construct a training data set, and pre-train the user function recommendation model and the personalized recommendation model based on the training data set to obtain the pre-trained user function recommendation model and the personalized recommendation model.

[0098] like Figure 2 As shown in the figure, the data set is constructed based on the design functions, model types, model parameters, user operation scores, and user search records extracted from the design project in step 2. The training data set can be expressed as:

[0099] T={F(i),MT(i),MP(i),U(i),R(i),D(i)}

[0100] Where T represents the training dataset, F(i) represents the design function of the i-th design project, MT(i) represents the model type of the i-th design project, MT(i) represents the model parameter of the i-th design project, U(i) represents the user operation score of the i-th design project, R(i) is the user search record of the i-th design project, and D(i) is the supervision label of the i-th design project.

[0101] like Figure 3 As shown, in step 4, a deep neural network is used to pre-train the user function recommendation model and the personalized recommendation model respectively, wherein the process of pre-training the user function recommendation model based on the training data set T includes:

[0102] 4.1.1. Extract design features, model types, model parameters, user operation scores, and user search records, and input them into the deep neural network N to predict the supervision label The calculation formula is:

[0103]

[0104] in, is the supervised label predicted by the deep neural network, F(i) represents the design function of the ith design project, MT(i) represents the model type of the ith design project, MP(i) represents the model parameter of the ith design project, U(i) represents the user operation score of the ith design project, R(i) is the user search record of the ith design project, and N is the deep neural network.

[0105] 4.1.2. Calculate the loss function. When the loss function is the smallest, stop training and use the back propagation algorithm to update all parameters of the deep neural network N to obtain the pre-trained user function recommendation model. The loss function is expressed by mean square error, and the calculation formula is:

[0106]

[0107] Where L represents the loss function of the pre-training of the user feature recommendation model, and D j (i) is the actual supervision label, is the supervised label predicted by the deep neural network, and n is the number of supervised labels.

[0108] The process of pre-training the personalized recommendation model based on the training data set T includes:

[0109] 4.2.1. Input the model type MT(i) into the personalized recommendation model P and predict the personalized score S(i):

[0110] S(i)=P(MT(i))

[0111] Among them, S(i) represents the predicted personalized score, and P is the personalized recommendation model.

[0112] 4.2.2. Input the user search record R(i) and user behavior characteristics B(i) into the personalized recommendation model P to obtain the user item preference sequence P(i):

[0113] P(i)=P(R(i),B(i))

[0114] 4.2.3. Calculate the loss function. When the loss function is the smallest, stop training and use the back propagation algorithm to update all parameters of the deep neural network N to obtain the pre-trained personalized recommendation model. The calculation formula of the loss function is:

[0115]

[0116] Among them, L p represents the loss function of the personalized recommendation model pre-training, U k (i) is the actual user operation score, S k (i) is the predicted personalized score, w k is the weight calculated based on the user-item preference sequence, and m is the number of ratings. In the recommendation system, the user's preference is represented by the input of the model (for example, the user's embedding vector). These vectors are learned through the training process, so the loss function itself does not directly contain the user-item preference sequence P(i), but through backpropagation and optimization, the model automatically learns the preference information related to the user-item preference sequence P(i).

[0117] By minimizing the loss function and performing back propagation, the optimized parameters of the personalized recommendation model are finally obtained, and the pre-trained personalized recommendation model is obtained based on the optimized parameters.

[0118] Through pre-training, both the user function recommendation model and the personalized recommendation model have been effectively trained, providing a basis for subsequent personalized recommendations. For example, if the user gives a high score to the design function "prediction accuracy" in multiple design projects, the model will adjust the weight of this function accordingly during training, thereby improving the accuracy of the recommendation and user satisfaction.

[0119] In the process of pre-training the user function recommendation model and the personalized recommendation model using deep neural networks, all parameters in the deep neural network are updated using the gradient method to minimize the loss function.

[0120] The output of the deep neural network is expressed as N(X), where X represents the input data, which includes design functions, model types, model parameters, user operation scores, and user search records. Feature extraction is performed through multiple hidden layers, and the output of the tth hidden layer can be expressed as:

[0121] N(X)=H t (W t ·H t-1 (W t-1 …H 1 (W 1 ·X+b 1 )+b 2 )+…+bt )

[0122] Among them, H t is the output of the tth hidden layer, W t is the weight of the tth hidden layer, b t is the corresponding bias. Through this hierarchical structure, the model is able to capture the complex relationships in the input data.

[0123] In order to optimize the model performance, the loss function L is defined to measure the difference between the model output and the actual supervision label, which is usually expressed as mean square error:

[0124]

[0125] Among them, D j is the actual supervision label, N(X j ) is the network output, and n is the number of samples. After calculating the loss function, we can get the gradient of the model parameters. We can calculate the derivative of the loss function for each weight and bias by the chain rule, and update the rule:

[0126]

[0127] Among them, η is the learning rate, which controls the update step size. Using the gradient of the loss function, the model can adjust the weights to better fit the training data set. For example, suppose the model predicts a value of 6 in a certain iteration, while the actual label is 4. At this time, the loss function will calculate a large error, reflecting that the model's prediction is inaccurate. By calculating the gradient of the loss function to update the model parameters, the model will strive to reduce this error, thereby improving the accuracy of the next prediction. Through multiple iterations, the network can gradually learn the relationship between user preferences and design functions, making personalized recommendations more accurate and effective.

[0128] Step 5: Use the pre-trained user function recommendation model to calculate the function score for the design projects in the user's history, and use the pre-trained personalized recommendation model to calculate the personalized recommendation score for the design projects in the user's history.

[0129] The core of calculating the feature score is to evaluate the recommended score of each design feature in the design project. Figure 3 ,The specific process of calculating the functional score includes:

[0130] 5.1. Evaluate the recommended score of each design function in the design project. Add up the recommended scores of all design functions and take the average to get the function score. The calculation formula for the recommended score of each design function is:

[0131]

[0132] Among them, Si,n represents the recommendation score of the nth design feature of the ith design project, w j is the design weight of the design item for each training sample, F j (i) is the design function of the jth training sample in the i-th design project. This weighted summation process enables the user function recommendation model to comprehensively consider the importance of multiple design functions and thus generate a comprehensive score.

[0133] 5.2. Input the user search records R(i) and user behavior features B(i) into the trained personalized recommendation model P, obtain the user project preference sequence P(i) for each design function, and calculate the preference weighted score P of all design projects. The calculation formula is:

[0134]

[0135] Among them, O(i) is the output of the deep neural network. The model type and model parameters are processed by the deep neural network, and the output obtained is O(i). The output O(i) represents the prediction result based on the user's historical data and current input. i is the preference rating weight of the i-th design item. The formula indicates how the preference rating weight is combined with the model output to generate a final preference weighted score that reflects the user's preference for a specific design feature.

[0136] 5.3. Update the preference selection weights. In the process of updating the weights, the user's preference for the model will affect the final recommendation results. The update formula can be simplified as:

[0137] W new =W old +α·(RW old )

[0138] Among them, W new represents the updated model weight, i.e., the new recommendation strategy or preference adjusted according to user feedback, W old represents the model weight before the update, that is, the recommendation strategy or preference of the model before the current moment, W old =W i , R represents the user's actual preference for the model, and α is the preset update rate. This mechanism ensures that in each iteration, the model can continuously adjust the recommendation strategy based on user feedback and gradually optimize the weights to better reflect the user's real preferences.

[0139] For example, if a user scores a certain feature in a design project with a score of 5, and the feature has a higher weight in the previous model, the updated weight will make the feature score higher in future recommendations. This dynamic adjustment process allows the model to match user needs more accurately, and through multiple rounds of iterations, it can ultimately achieve more personalized recommendation results and enhance user experience.

[0140] In step five, the function score is calculated by the user function recommendation model, the preference weighted score of the design project is calculated by the personalized recommendation model, and the personalized score of the design project is calculated based on the function score and the preference weighted score of the design project. The function score is obtained by comprehensively calculating the recommendation score of each design function. Specifically, it is necessary to first calculate the score for each function in each design project, which reflects the degree of recommendation of the function. Then, these function scores are averaged to obtain the recommendation score of the overall project, that is, the function score. This process ensures that the model can take into account the impact of all functions on user decisions, thereby providing a comprehensive evaluation.

[0141] The user function model aims to extract the user's actual needs and function preferences, while the personalized recommendation model is optimized based on the user's historical behavior and ratings. The combination of the two can more accurately reflect the user's interests and needs. In practical applications, suppose a design project contains three design functions with recommendation scores of 4, 5, and 3, respectively. Then we add these scores and average them to get a recommendation score of 4 for the project.

[0142] This scoring mechanism provides users with an intuitive reference to help them make choices among many design projects. With the continuous iteration and optimization of the model, the final recommendation results will be more in line with the personalized needs of users, thereby improving the accuracy and effectiveness of the recommendation system.

[0143] Step 6: Combine the function score and personalized recommendation score to generate the final project function recommendation list.

[0144] Step 7: Push a list of project function recommendations to users.

[0145] Step 8: Verify the correctness of the user operation score, and use reinforcement learning to update the user operation score and user search history, such as Figure 4 As shown in FIG. 1 , the specific process of updating the user operation score and the user search record by using reinforcement learning includes the following steps:

[0146] 8.1. Integrate user search history, behavior, and user operation scores into the state of reinforcement learning, which provides the model with contextual information about the current user. The actions of reinforcement learning are the recommended items selected based on these states, and the user preference weighted scores are derived from them, reflecting the relevance of the recommendation results.

[0147] In this framework, we can define the reward function and state value function. The reward function represents the user's feedback on the recommendation results, and the state value function represents the long-term reward that can be obtained by taking a specific action in the current state. This structure enables the model to gradually learn how to select the best recommended action in a given state, thereby improving user satisfaction.

[0148] 8.2. Compare the results output by the recommendation model with actual user feedback to determine whether the recommended design features are effective. This comparison provides the basis for subsequent feedback loops, allowing the model to accurately identify which recommendations are successful and which need improvement. For each design project, it is necessary to ensure that the recommended design features can truly meet the needs of users.

[0149] 8.3. The process of calculating user feedback involves estimating the optimal solution. The degree of match between the user's actual preferences and the recommended function will affect the calculation of the reward, thereby affecting the learning direction of the subsequent model. This feedback process ensures that the model is continuously optimized.

[0150] 8.4. The step of updating user behavior and preferences is based on the difference between the recommended results and the actual values. By introducing user update weights and update rates, user behavior and preferences can be adjusted dynamically. For example, when the recommendation output by the model matches the user's actual preferences, the weight of the item can be increased so that it appears more frequently in subsequent recommendations.

[0151] In this way, the model can adapt to the changing needs of users, making the recommendation system more personalized and accurate. Figure 5 As shown in Figure 1, it is a flowchart for optimizing the user rating and recommendation model through reinforcement learning. Specifically, if a design project has received a high score in past recommendations, the model will increase the recommendation score of the project, giving it a higher priority in future recommendations.

[0152] In summary, this optimization process enables the recommendation system to effectively learn user preferences through continuous feedback loops and dynamic adjustments, thereby improving the accuracy of recommendations and user satisfaction. Through this reinforcement learning method, the system can not only adapt to the user's immediate needs, but also extract potential patterns from historical data to further improve the quality of recommendations.

[0153] Step 9: Use the updated user operation scores and user search records to update the user function recommendation model and the personalized recommendation model respectively. The specific process includes the following:

[0154] 9.1. Input user search history, behavior data and rating information into the personalized recommendation model. These data provide valuable information for the personalized recommendation model, helping it to better understand the user's interests. For ease of processing, the user's input information is converted into a vector representation, which is initially a zero vector and then updated by the input data.

[0155] 9.2. The personalized recommendation model calculates a user preference weighted set, which quantifies the user's preference for the design features of different design projects. The accuracy of the personalized recommendation model can be evaluated by comparing the user's historical ratings with the output of the current personalized recommendation model. This process involves using a loss function of a deep neural network, the purpose of which is to minimize the difference between the model's predicted value and the actual user operation score. The loss function is calculated as:

[0156]

[0157] Among them, Y k is the actual score of the k-th design project, is the predicted score of the kth design project, and n is the number of samples. By calculating the mean square error of these scores, the personalized recommendation model can accurately understand its own predictive ability.

[0158] 9.3. Use the gradient descent method to adjust the parameters of the user function recommendation model. In the process of updating the user function recommendation model, similar processing is performed for each design project and each design function. Figure 5 The figure shows the flowchart of the updating process and parameter adjustment of the user function recommendation model. For each model type and model parameter, the user preference weighted set is first output and compared with the actual score to obtain the corresponding error. Then, the gradient descent method is used to adjust the model parameters. This process ensures that the user function recommendation model can be continuously optimized. The same scoring and updating operations are performed for each design function to ensure that all functions can be reasonably recommended.

[0159] Suppose a user gives a high score to a certain design project, the model will record this preference. In subsequent updates, the preference weighted set of the project will be improved, making it more likely that the project will be pushed to the user first in future recommendations. This dynamic adjustment mechanism not only improves the recommendation accuracy of the model, but also enhances user satisfaction, ultimately achieving a personalized recommendation experience. Through continuous iteration and updating, the recommendation system can better adapt to the changing needs of users and provide more accurate services.

[0160] 9.4. Update the score of each design function. When updating each design function, parameter adjustment through gradient descent is a key step. Specifically, first define the score as the target value of each design project, denoted as S n , and then adjust the update rate of the jth design function, denoted as n, by calculating the derivative of the loss function with respect to the score. This derivative reflects the difference between the current model output and the actual user preference. If the derivative is positive, it means that the current score is low; if it is negative, it means that the score is high.

[0161] In this process, user behavior data is input into the personalized recommendation model to obtain a weighted set of user preferences. This set integrates user search records and behavior data to form feedback information for specific design functions. Then the objective function is solved to optimize the model to reduce errors.

[0162] Through the gradient descent method, the objective function can be expressed in the form of a loss function. The derivative of the loss function will guide us how to adjust the model parameters. Specifically, assuming the loss function is L, the following formula can be used to update the parameters:

[0163]

[0164] Among them, θ represents the model parameters to be updated, It is the derivative of the loss function with respect to the parameters, reflecting the impact of the current parameters on the loss. In this way, the parameters can be effectively updated in the direction of reducing the loss, thereby improving the prediction accuracy of the model.

[0165] Furthermore, the updating process includes inputting the user behavior data for a certain number of times, set as k, and optimizing the objective function in combination with the preset initial value. For each input, we calculate the loss function and update the parameters according to the derivatives to ensure that the model gradually approaches the optimal state.

[0166] For example, if the user's score for a design feature is significantly higher than the model's predicted score, the derivative of the loss function will be a large negative value, indicating that the score of the design feature needs to be significantly increased. By applying the gradient descent method, the model will adjust the parameters so that in the next prediction, the score of the feature will be improved, thus better matching the user's actual preferences. After multiple iterations, the final model will be able to accurately reflect the user's true preferences for each design feature.

[0167] Step 10: Use the updated user function recommendation model and personalized recommendation model to regenerate the project function recommendation list, and repeat steps 7 to 10 until the convergence value of the user operation score accuracy rate is higher than the preset threshold.

[0168] By extracting the user's historical logs, user operation scores and user search records, the system can gain a deep understanding of the user's preferences and needs, thereby generating more accurate design function recommendations and improving the accuracy of personalized recommendations. This data-driven approach enables the recommendation system to identify the user's specific needs in different design projects, thereby reducing the time and energy wasted by the user in the selection process.

[0169] By continuously updating user operation scores and feedback, the recommendation system can adjust the recommended content in real time to make it more in line with the actual needs of users. The introduction of reinforcement learning algorithms enables the system to learn from each user interaction, thereby optimizing the recommendation strategy, improving the user experience, and enhancing the user's trust and satisfaction with the system.

[0170] The present invention can help users quickly find the most suitable design solution and improve design efficiency. By generating a functional scoring and recommendation list for design projects, users can more clearly see the advantages and disadvantages of different projects, thereby making better decisions, promoting the optimization of design projects, and optimizing the design process.

[0171] The use of deep learning and reinforcement learning algorithms enables the recommendation system to have the ability to self-learn, achieve continuous learning and self-optimization. By continuously collecting user feedback and behavior data, the system can continuously optimize model parameters and recommendation strategies to adapt to changes in user preferences. This dynamic adaptability ensures that the system can always provide high-quality recommendations and improve the overall intelligence level.

[0172] By analyzing the user's preferences for different design functions, the system can recommend a variety of design options, encourage users to explore new design ideas, and promote innovation. This function not only meets the user's current needs, but also provides users with more possibilities, supports them to make more extensive attempts in the design process, and improves the diversity and innovation of design functions.

[0173] The present invention can not only improve the effectiveness of personalized recommendations, but also bring significant improvements in user experience, design efficiency and system intelligence, ultimately achieving a higher level of user satisfaction and system value.

[0174] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A personalized recommendation method in a robot intelligent assisted design system, characterized by: include: Step 1: Extract design projects from user history logs; Step 2: Extract the design functions, model types, model parameters, user operation scores, and user search records in the design projects to construct a data set; Step 3: Label the supervision label based on the user operation score; Step 4: Use the data set and the supervision label to construct a training data set, and pre-train the user function recommendation model and the personalized recommendation model based on the training data set to obtain the pre-trained user function recommendation model and the personalized recommendation model; Step 5: Use the pre-trained user function recommendation model to calculate the function score for the design projects in the user's history, and use the pre-trained personalized recommendation model to calculate the personalized recommendation score for the design projects in the user's history; Step 6: Combine the function score and the personalized recommendation score to generate the final project function recommendation list; Step 7: Push a list of project function recommendations to users; Step 8: Verify the correctness of the user operation score, and use reinforcement learning to update the user operation score and user search history; Step 9: Use the updated user operation scores and user search records to update the user function recommendation model and the personalized recommendation model respectively; Step 10: Use the updated user function recommendation model and personalized recommendation model to regenerate the project function recommendation list, and repeat steps 7 to 10 until the convergence value of the user operation score accuracy rate is higher than the preset threshold.

2. A personalized recommendation method in a robot intelligent assisted design system according to claim 1, characterized in that: In the step 2, the design function is extracted based on the basic modules involved in the design project, and the functions implemented by the basic modules are determined by analyzing the basic modules included in the design project; the model parameters are determined based on the model parameter type established for each design project and its corresponding design function; Obtain user operation scores by analyzing model types and model parameters in design projects; User search records are established by tracking users' operational behaviors on the design process, recording the users' search keywords and frequency during use.

3. A personalized recommendation method in a robot intelligent assisted design system according to claim 1, characterized in that: The supervision label in step 3 is a collection of design features in the design project, expressed as: D(i)={f1,f2,...,f k } Where D(i) is the set of design functions for the i-th design project, and f j is a specific design function, j = 1, 2, ..., k.

4. A personalized recommendation method in a robot intelligent assisted design system according to claim 3, characterized in that: The training data set in step 4 is represented as: T={F(i), MT(i), MP(i), U(i), R(i), D(i)} Where T represents the training dataset, F(i) represents the design function of the i-th design project, MT(i) represents the model type of the i-th design project, MT(i) represents the model parameter of the i-th design project, U(i) represents the user operation score of the i-th design project, R(i) is the user search record of the i-th design project, and D(i) is the supervision label of the i-th design project.

5. A personalized recommendation method in a robot intelligent assisted design system according to claim 1, characterized in that: In step 4, a deep neural network is used to pre-train the user function recommendation model. The pre-training process includes: 4.1.

1. Extract design features, model types, model parameters, user operation scores, and user search records, and input them into the deep neural network N to predict the supervision label The calculation formula is: in, is the supervised label predicted by the deep neural network, F(i) represents the design function of the i-th design project, MT(i) represents the model type of the i-th design project, MP(i) represents the model parameter of the i-th design project, U(i) represents the user operation score of the i-th design project, R(i) is the user search record of the i-th design project, and N is the deep neural network; 4.1.

2. Calculate the loss function. When the loss function is the smallest, stop training and use the back propagation algorithm to update all parameters of the deep neural network N to obtain the pre-trained user function recommendation model. The calculation formula of the loss function is: Where L represents the loss function of the pre-training of the user feature recommendation model, and D j (i) is the actual supervision label, is the supervised label predicted by the deep neural network, and n is the number of supervised labels.

6. A personalized recommendation method in a robot intelligent assisted design system according to claim 1, characterized in that: In step 4, a deep neural network is used to pre-train the personalized recommendation model. The pre-training process includes: 4.2.

1. Input the model type MT(i) into the personalized recommendation model P and predict the personalized score S(i): S(i)=P(MT(i)) Among them, S(i) represents the predicted personalized score, and P is the personalized recommendation model; 4.2.

2. Input the user search record R(i) and user behavior characteristics B(i) into the personalized recommendation model P to obtain the user item preference sequence P(i): P(i)=P(R(i),B(i)) 4.2.

3. Calculate the loss function. When the loss function is the smallest, stop training and use the back propagation algorithm to update all parameters of the deep neural network N to obtain the pre-trained personalized recommendation model. The calculation formula of the loss function is: Among them, L p represents the loss function of the personalized recommendation model pre-training, U k (i) is the actual user operation score, S k (i) is the predicted personalized score, w k is the weight calculated based on the user-item preference sequence, and m is the number of ratings.

7. A personalized recommendation method in a robot intelligent assisted design system according to claim 5 or 6, characterized in that: The output of the deep neural network is expressed as N(X), where X represents input data. The input data X includes design functions, model types, model parameters, user operation scores, and user search records. Feature extraction is performed through multiple hidden layers. The output of the tth hidden layer can be expressed as: N(X)=H t (W t ·H t-1 (W t-1 ...H1(W1·X+b1)+b2)+...+b t ) Among them, H t is the output of the tth hidden layer, W t is the weight of the tth hidden layer, b t is the corresponding bias; The calculation formula for defining the loss function L is: Among them, D j is the actual supervision label, N(X j ) is the network output, n is the number of samples; After calculating the loss function, we get the gradient of the model parameters. We use the chain rule to calculate the derivative of the loss function for each weight and bias. The update rule is: Among them, η is the learning rate, which controls the update step size.

8. A personalized recommendation method in a robot intelligent assisted design system according to claim 1, characterized in that: The specific process of calculating the functional score in step 5 includes: 5.

1. Evaluate the recommended score of each design function in the design project. Add up the recommended scores of all design functions and take the average to get the function score. The calculation formula for the recommended score of each design function is: Among them, S i，n represents the recommendation score of the nth design feature of the ith design project, w j is the design weight of the design item for each training sample, F j (i) is the design function of the jth training sample in the i-th design project; 5.

2. Input the user search records R(i) and user behavior features B(i) into the trained personalized recommendation model P, obtain the user project preference sequence P(i) for each design function, and calculate the preference weighted score P of all design projects. The calculation formula is: Among them, O(i) is the output of the deep neural network, W i is the preference selection weight of the i-th design project; 5.

3. Update the preference selection weights. The update formula is: W new =W old +α·(RW old ), where W new represents the updated model weight, W old represents the model weight before updating, R represents the user’s actual preference for the model, and α is the preset update rate.

9. A personalized recommendation method in a robot intelligent assisted design system according to claim 1, characterized in that: The process of updating the user operation score and the user search record by using the reinforcement learning method in step eight includes: 8.

1. Integrate user search history, behavior, and user operation scores into reinforcement learning states, select recommended items as actions, calculate user preference weighted scores, and define reward functions and state value functions; 8.

2. Compare the results output by the recommendation model with actual user feedback to determine whether the recommended design function is effective; 8.

3. Calculate user feedback and update the model; 8.

4. Dynamically adjust user behavior and preferences.

10. A personalized recommendation method in a robot intelligent assisted design system according to claim 1, characterized in that: The process of updating the user function recommendation model and the personalized recommendation model in step nine includes: 9.

1. Input user search history, behavior data and rating information into the personalized recommendation model; 9.

2. The personalized recommendation model calculates the user preference weighted set, compares the output of the personalized recommendation model with the actual score, and calculates the loss function. The calculation formula of the loss function L is: Among them, Y k Score actual user actions, is the score predicted by the personalized recommendation model, and n is the number of samples; 9.

3. Use the gradient descent method to adjust the parameters of the user function recommendation model. The update method of the model parameters is: Among them, θ represents the model parameters to be updated, It is the derivative of the loss function with respect to the parameters, reflecting the impact of the current parameters on the loss; 9.

4. Update the score of each design feature.

Images