Creative design learning resource intelligent recommendation method and system

Abstract

The invention provides a creative design learning resource intelligent recommendation method and system, and relates to the technical field of learning resource recommendation. The method comprises the following steps: collecting historical resource use data and historical work data of a target user about creative design learning, and extracting multiple groups of local resource use features and local design behavior features; performing style matching on the local resource use features and the local design behavior features to generate multiple groups of local resource style matching results and local behavior style matching results; constructing a resource use offset path and a design style evolution path of the target user, and extracting state trend disturbance characteristics of the target user in different time periods; and obtaining preference feature data and outlier feature data of the target user based on the state trend disturbance features in different time periods, generating a target resource recommendation strategy of the target user, and performing learning resource recommendation about creative design to the target user. According to the method and the device, learning resources related to creative design are accurately and stably recommended to the user.

Description

Technical Field

[0001] The present invention relates to the technical field of learning resource recommendation, and in particular to a method and system for intelligently recommending creative design learning resources. Background Art

[0002] With the increasing popularity of online education platforms, personalized recommendations for learning resources have become a key means of improving user learning efficiency. Conventional recommendation methods are often based on information such as users' historical clickthrough rates, ratings, or resource tags, and are suitable for learning scenarios focused on structured knowledge. However, in creative design learning, where learning objectives emphasize personalized expression and stylized creation, user behavior often exhibits strong nonlinearity and stage-specific characteristics. Recommendations based solely on static behavior or content similarity fail to accurately capture users' true interests and abilities.

[0003] During the creative design learning process, users' resource selection reflects their interest in specific styles or skills, and their creative works directly reflect their current design capabilities and stylistic tendencies. If the recommendation process ignores the stylistic preferences and behavioral differences expressed by users in these two dimensions, the recommendations can easily deviate from the user's development direction or creative goals. Furthermore, users may exhibit short-term fluctuations in interests or engage in style exploration. Failure to identify and address these outliers can also affect the stability and personalization of recommendations. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes an intelligent recommendation method and system for creative design learning resources. By comprehensively analyzing the information on users' learning resource usage and creative style evolution, the style preferences and evolution trends contained in users' learning resource selection behavior and actual design work style behavior are explored, thereby achieving accurate and stable recommendation of learning resources related to creative design to users.

[0005] In a first aspect, the present invention provides a method for intelligently recommending creative design learning resources, comprising: Collect historical resource usage data and historical work data of target users related to creative design learning, segment the historical resource usage data and historical work data into time periods, and extract the local resource usage characteristics and local design behavior characteristics of target users in different time periods; Obtain resource creation association data about learning resources and target creation styles, perform style matching on local resource usage characteristics and local design behavior characteristics based on the resource creation association data, and generate local resource style matching results and local behavior style matching results for target users in different time periods; Based on multiple sets of local resource style matching results and local behavior style matching results, the target user's resource usage offset path and design style evolution path are generated. State trend disturbance analysis is performed on the resource usage offset path and design style evolution path to extract the state trend disturbance characteristics of the target user in different time periods. Based on the state trend disturbance characteristics in different time periods, the preference feature data and outlier feature data of the target user are obtained, and a target resource recommendation strategy for the target user is generated according to the preference feature data and outlier feature data. Based on the target resource recommendation strategy, learning resources about creative design are recommended to the target user.

[0006] Preferably, style matching is performed on local resource usage characteristics and local design behavior characteristics respectively according to resource creation associated data, including: Resource creation-related data includes multiple learning resources associated with multiple target creation styles. The element application reference sequence for each target creation style is determined, and the local element application sequence of each local design behavior feature is extracted. Each local element application sequence is matched with multiple element application reference sequences to generate the local behavior style matching results of the target user in different time periods. Determine the local resource usage sequence corresponding to each local resource usage feature, generate a resource usage reference sequence for each target creative style based on the multiple learning resources associated with the target creative style, match each local resource usage sequence with multiple resource usage reference sequences respectively, and generate local resource style matching results for the target user in different time periods.

[0007] Preferably, a state trend disturbance analysis is performed on the resource usage offset path and the design style evolution path to extract the state trend disturbance features of the target user in different time periods, including: Sliding window detection is performed on the resource usage offset path and design style evolution path respectively, and the resource usage offset state sequence and design style evolution state sequence of each time period are extracted. The linear direction parameters and trend deviation parameters corresponding to the offset state sequence and design style evolution state sequence in each time period are calculated respectively. The trend disturbance parameters in each time period are calculated based on the linear direction parameters and trend deviation parameters, and the resource usage trend disturbance sequence and design style trend disturbance sequence of the target user are generated.

[0008] Preferably, obtaining the target user's preference feature data and outlier feature data based on the state trend disturbance features in different time periods includes: Based on the resource usage offset path and design style evolution path, a resource creation consistency matrix for multiple target creation styles is constructed, and the style alignment index of each target creation style is extracted based on the resource creation consistency matrix. The trend disturbance index of each target creative style is calculated based on the resource usage trend disturbance sequence and the design style trend disturbance sequence. The stability score of each target creative style is determined based on the style alignment index and the trend disturbance index. According to the stability score, the preference feature data of the target users containing multiple stable styles and the outlier feature data containing multiple outlier styles are generated.

[0009] Preferably, generating a target resource recommendation strategy for a target user based on the preference feature data and the outlier feature data includes: A global style preference evolution optimization analysis is performed on the target users' preference feature data and outlier feature data, including extracting the global preference evolution index of each target creative style from the resource usage trend perturbation sequence and the design style trend perturbation sequence, optimizing the global style preference evolution of the stability score through the global preference evolution index, generating a dynamic preference score for each target creative style, and generating a target resource recommendation strategy for target users regarding multiple learning resources based on the dynamic preference score.

[0010] Preferably, the resource creation consistency matrix includes the frequency of occurrence of any two target creation styles as dominant styles in the resource usage offset path and the design style evolution path within the same period.

[0011] A second aspect of the present invention provides a creative design learning resource intelligent recommendation system, which is used to implement the above-mentioned creative design learning resource intelligent recommendation method, comprising: The local feature extraction module is used to collect the target user's historical resource usage data and historical work data related to creative design learning, segment the historical resource usage data and historical work data into time periods, and extract the target user's local resource usage features and local design behavior features in different time periods; The style matching analysis module is used to obtain resource creation association data about learning resources and target creation styles, and perform style matching on local resource usage characteristics and local design behavior characteristics based on the resource creation association data, thereby generating local resource style matching results and local behavior style matching results for target users in different time periods. The state trend disturbance analysis module is used to generate the target user's resource usage offset path and design style evolution path based on multiple sets of local resource style matching results and local behavior style matching results. The module then performs state trend disturbance analysis on the resource usage offset path and design style evolution path to extract the state trend disturbance characteristics of the target user in different time periods. The learning resource recommendation optimization module obtains the target user's preference feature data and outlier feature data based on the state trend disturbance characteristics in different time periods, generates the target user's target resource recommendation strategy based on the preference feature data and outlier feature data, and recommends learning resources about creative design to the target user based on the target resource recommendation strategy.

[0012] The present invention has the following beneficial effects: The present invention extracts the local resource usage characteristics and design behavior characteristics of target users in creative design learning through time period segmentation, and constructs local resource style matching results and local behavior style matching results in combination with resource style association data, generating resource usage offset path and design style evolution path; adopts sliding window detection and linear direction parameter analysis to capture the state trend disturbance characteristics in time series, quantifies the matching degree fluctuation and trend deviation degree of different creative styles, and breaks through the limitations of traditional static behavior analysis; calculates the style stability score by fusing the resource creation consistency matrix and the trend disturbance index, and constructs a parameter-linked style preference evolution model combined with dynamic optimization of the global preference evolution index, accurately identifies users' stable style preferences and short-term outlier behaviors, distinguishes core needs from exploratory learning behaviors in the recommendation strategy, and effectively improves the dynamic adaptability and personalization level of creative design learning resource recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of a method for intelligently recommending creative design learning resources provided by the present invention.

[0014] Figure 2 This is a structural diagram of an intelligent recommendation system for creative design learning resources provided by the present invention. DETAILED DESCRIPTION

[0015] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0016] In the first aspect, the present invention provides a method for intelligently recommending creative design learning resources. Figure 1 , the method specifically comprises the following steps: Step S1: Collect historical resource usage data and historical work data of target users regarding creative design learning, segment the historical resource usage data and historical work data into time periods, and extract local resource usage characteristics and local design behavior characteristics of target users in different time periods.

[0017] In this step, the target user's relevant historical data on creative design learning is obtained. The historical resource usage data records the courses, videos, case libraries and other learning resources that the user has visited, studied, or interacted with, as well as specific information such as usage duration, completion rate, click frequency, and user evaluation. The design work data includes record information related to the works uploaded or submitted by the user, such as the submission time of the work, the task type, the use of different elements in the design process, and other information.

[0018] In order to capture the temporal evolution characteristics of users' learning and creation behaviors, these historical data are divided into multiple time periods according to time, such as every 5 days as a stage, and local resource usage characteristics and local design behavior characteristics are extracted in each time period. The local resource usage characteristics are used to indicate the aggregated characteristics of the learning resources used by users in this time period, including multiple usage-related data of each learning resource used. The local design behavior characteristics are used to express the style characteristics of the user's design works in this time period, such as the main color type, font structure, layout complexity, white space ratio, image and text integration, visual movement and other related design elements.

[0019] Step S2: Obtain resource creation association data about learning resources and target creation styles, perform style matching on local resource usage characteristics and local design behavior characteristics based on the resource creation association data, and generate local resource style matching results and local behavior style matching results for target users in different time periods.

[0020] In this step, a pre-established resource-style association dataset—that is, resource creation association data about learning resources and target creative styles—is used to perform style matching based on the previously extracted resource usage characteristics and design behavior characteristics. This determines the target creative style actually associated with learning resource usage and design behavior at different time periods. Ultimately, multiple local resource style matching results and local behavior style matching results for the user are determined, including the degree of match between the user's learning resources and different target creative styles, as well as the degree of match between the user's designed works and different target creative styles, at each time period.

[0021] In a specific example, resource creation-related data includes multiple learning resources associated with multiple target creative styles, where the multiple target creative styles can be obtained by comprehensively dividing the element information used in the work design process from multiple dimensions, such as layout features, graphic element features, font features, texture and noise information in the design process of a large number of representative works, comprehensive clustering or determination through expert evaluation, such as minimalist style, retro style, futuristic style, etc. Different learning resources can be obtained based on feedback data provided by a large number of users after use, or determined by experts in the process of dividing different design styles. Each learning resource is suitable for the relevant annotation data of which target creative style needs to be learned, thereby determining the resource creation-related data containing multiple learning resources associated with the target creative style.

[0022] Taking comprehensive clustering to obtain multiple target creative styles as an example, we can construct an element application reference sequence for each target creative style based on the cluster center of each target creative style determined in the clustering process, and extract the local element application sequence of each local design behavior feature. Then, each local element application sequence is matched with multiple element application reference sequences respectively, and the matching degree between the style of the user's design works in different time periods and different target creative styles is calculated. For example, it can be quantified through the Euclidean distance between the sequences to generate the local behavior style matching results of the target user in different time periods.

[0023] For the processing of local resource usage features, the local resource usage sequence corresponding to each local resource usage feature is determined. Specifically, representative learning resources can be selected, such as learning resources whose learning time or completion rate reaches a pre-set threshold, to construct a local resource usage sequence for multiple learning resources. For the multiple learning resources associated with the target creative style, a resource usage reference sequence for each target creative style is also generated. For example, if a certain learning resource exists, it is recorded as 1, otherwise it is recorded as 0. Each local resource usage sequence is then matched with multiple resource usage reference sequences. In this process, the proportion of common learning resources between the local resource usage sequence and the resource usage reference sequence can be calculated. For example, if a user reaches a preset completion rate for 3 video courses in a certain period of time, and 5 of the video courses are about teaching the creative style of minimalism, then the matching degree is 0.6, thereby generating the local resource style matching results of the target user in different time periods.

[0024] Step S3: Generate the target user's resource usage offset path and design style evolution path based on multiple sets of local resource style matching results and local behavior style matching results, perform state trend disturbance analysis on the resource usage offset path and design style evolution path, and extract the state trend disturbance characteristics of the target user in different time periods.

[0025] In this step, based on the style matching results across different time periods, we construct resource usage shift paths and design style evolution paths. For example, we record the style matching of learning resources and the stylistic expression of design works in a time series, forming a ternary sequence of time, style, and matching degree. Then, for each time period, we further analyze the trend changes of the dominant style within the path, extracting state trend disturbance features to measure the state stability of the dominant style within that local time period.

[0026] In a specific example, the process of performing state trend disturbance analysis on the resource usage offset path and the design style evolution path specifically includes: A sliding window test is performed on each resource usage offset path and design style evolution path to extract the resource usage offset state sequence and design style evolution state sequence for each time period. In this process, the offset state sequence and design style evolution state sequence are specifically the matching degrees corresponding to the dominant target creative style of each time period within a specific range before and after each time period. For example, taking the adjacent time periods as an example, for the resource usage offset path, the dominant target creative style in each time period is determined, that is, the target creative style with the highest matching degree. Then, the matching degrees corresponding to the target creative style in the previous and next time periods are extracted, thereby forming the resource usage offset state sequence for the time period.

[0027] Then, the linear direction parameter and trend deviation parameter corresponding to the offset state sequence and the design style evolution state sequence in each time period are calculated. In this embodiment, the linear direction parameter is used to describe whether the matching degree of the current dominant style is in an upward, downward, or stable trend. It is defined as the average of the differences before and after, that is, the average of the differences in matching degrees between two adjacent time periods, where:

[0028] Where, Indicates the The linear direction parameter of the time period, 、 Respectively The target creative style dominated by the period is Hedi The matching degree corresponding to each time period.

[0029] The trend deviation parameter is used to measure whether the dominant style matching degree of the current period deviates significantly from that of the adjacent periods, reflecting whether there are nonlinear disturbances such as mutations and fluctuations.

[0030] Where, Indicates the The matching degree of the target creative style dominated by each period.

[0031] After the linear direction parameters and trend deviation parameters corresponding to the offset state sequence and the design style evolution state sequence in each period are calculated by the above formula, the trend disturbance parameters in each period are calculated according to the linear direction parameters and trend deviation parameters. The calculation includes:

[0032] Therefore, based on multiple trend perturbation parameters, the resource usage trend perturbation sequence and design style trend perturbation sequence of the target user are generated, which represent the state trend perturbation characteristics of the target user in different time periods and are used to reflect the preference formation process and style evolution law behind the user behavior.

[0033] Step S4: Obtain the target user's preference feature data and outlier feature data based on the state trend disturbance characteristics in different time periods, generate a target resource recommendation strategy for the target user based on the preference feature data and outlier feature data, and recommend learning resources about creative design to the target user based on the target resource recommendation strategy.

[0034] In this step, based on the state trend disturbance characteristics and combined with the target user's resource usage offset path and design style evolution path, the global performance characteristics of different target creative styles in user behavior are further analyzed. By constructing the target user's preference feature data and outlier feature data, it is used to evaluate the user's actual preference relationship for different target creative styles, thereby formulating a personalized target resource recommendation strategy for the target user. Based on the target resource recommendation strategy, learning resources about creative design are recommended to the target user, realizing the coordinated matching of learning behavior and creative style growth in resource recommendation, and accurately and stably recommending learning resources related to creative design to users.

[0035] In a specific example, obtaining the target user's preference feature data and outlier feature data based on the state trend disturbance features in different time periods specifically includes: A resource creation consistency matrix for multiple target creation styles is constructed based on the resource usage offset path and the design style evolution path, and the style alignment index of each target creation style is extracted based on the resource creation consistency matrix.

[0036] Specifically, a resource creation consistency matrix is constructed based on the dominant styles in the resource usage offset path and the design style evolution path, respectively. The elements in the matrix specifically represent the frequency of occurrence of the dominant style in the resource usage offset path and the design style evolution path during the same period between the two target creation styles. For each target creation style, the style alignment index is used to measure the degree of consistency between user learning and creation. That is, the frequency of occurrence of the same target creation style as the dominant style in the resource usage offset path and the design style evolution path during the same period. By normalizing the corresponding frequencies of occurrence of multiple target creation styles, a style alignment index is generated for each target creation style. The larger the value, the more consistent the style is in user learning and creation, and the more likely it is to be the true preferred style of internalized expression.

[0037] At the same time, the trend disturbance index of each target creative style is calculated based on the resource usage trend disturbance sequence and the design style trend disturbance sequence. Specifically, the mean of multiple trend disturbance parameters of each target creative style in the resource usage trend disturbance sequence and the design style trend disturbance sequence is determined, and the trend disturbance index of each target creative style is generated after normalization. The larger the value, the more continuous positive or negative deviation the trend of the style in learning and creation is, and the more evolutionary it is. Otherwise, it means that the style only appears for a short period of time or there are unstable phenomena such as fluctuations.

[0038] Finally, the style alignment index and the trend perturbation index are combined to calculate the style stability score of each target creative style. For example, the style alignment index is weighted and corrected by the trend perturbation index. The higher the style stability score of the target creative style, the more likely it is that the user has consistently displayed this style in both learning and creation, and the trend evolution is stable. Conversely, a lower style stability score indicates that the style only appears for a short period of time or is inconsistent in performance, which may be a temporary behavior or an outlier. The various target creative styles are divided by a preset stability threshold. Those above the preset stability threshold are classified as stable styles, and the rest are classified as outlier segments. In this way, preference feature data of the target user containing multiple stable styles and outlier feature data containing multiple outlier styles are generated according to the stability score to obtain structured user portrait information.

[0039] In a specific example, generating a target resource recommendation strategy for a target user based on preference feature data and outlier feature data specifically includes: A global style preference evolution optimization analysis is performed on the target user's preference feature data and outlier feature data to generate a dynamic preference score for each target creative style. Based on the dynamic preference score, a target resource recommendation strategy for the target user regarding multiple learning resources is generated.

[0040] Specifically, the global preference evolution index of each target creative style is extracted from the resource usage trend disturbance sequence and the design style trend disturbance sequence. For example, the trend disturbance parameters of multiple time-series changes of the target creative style are fitted to generate a linear trend line about the trend disturbance parameters. In this embodiment, the trend disturbance sequence and the design style trend disturbance can be fitted separately, and the mean of the linear regression slopes of the linear trend lines is taken as the global preference evolution index of the target creative style, which is used to describe whether the target creative style is growing rapidly in the overall user behavior sequence, such as frequent appearance and an upward trend, or tending to degenerate, such as a downward trend or being replaced by other styles. Finally, the global preference evolution index and the stability score are merged, and the sum of the original stability score and the global preference evolution index weighted and corrected by the global preference evolution index is used as the dynamic preference score of the target creative style, which more accurately reflects the user's style orientation in real development.

[0041] Ultimately, a target resource recommendation strategy for target users regarding multiple learning resources can be formulated based on dynamic preference scores. For example, different learning resources can be sorted according to the dynamic preference scores of the target creative style, and personalized learning resource recommendation plans for creative design can be generated for users. In addition, by regularly collecting users' recent data, such as data within the past three months, the dynamic preference scores can be updated in stages to enhance the time sensitivity and adaptability of the recommendation strategy.

[0042] In a second aspect, the present invention provides a creative design learning resource intelligent recommendation system for implementing the above-mentioned creative design learning resource intelligent recommendation method. Figure 2 , the system comprises: The local feature extraction module is used to collect the target user's historical resource usage data and historical work data related to creative design learning, segment the historical resource usage data and historical work data into time periods, and extract the target user's local resource usage features and local design behavior features in different time periods; The style matching analysis module is used to obtain resource creation association data about learning resources and target creation styles, and perform style matching on local resource usage characteristics and local design behavior characteristics based on the resource creation association data, thereby generating local resource style matching results and local behavior style matching results for target users in different time periods. The state trend disturbance analysis module is used to generate the target user's resource usage offset path and design style evolution path based on multiple sets of local resource style matching results and local behavior style matching results. The module then performs state trend disturbance analysis on the resource usage offset path and design style evolution path to extract the state trend disturbance characteristics of the target user in different time periods. The learning resource recommendation optimization module obtains the target user's preference feature data and outlier feature data based on the state trend disturbance characteristics in different time periods, generates the target user's target resource recommendation strategy based on the preference feature data and outlier feature data, and recommends learning resources about creative design to the target user based on the target resource recommendation strategy.

[0043] The foregoing description is merely a detailed description of the present invention, which is intended to enable those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art. Portions not described in detail in this specification are well known to those skilled in the art.

Claims

1. A method for intelligently recommending creative design learning resources, characterized in that: include: Collect historical resource usage data and historical work data of target users related to creative design learning, segment the historical resource usage data and historical work data into time periods, and extract the local resource usage characteristics and local design behavior characteristics of target users in different time periods; Obtain resource creation association data about learning resources and target creation styles, perform style matching on local resource usage characteristics and local design behavior characteristics based on the resource creation association data, and generate local resource style matching results and local behavior style matching results for target users in different time periods; Based on multiple sets of local resource style matching results and local behavior style matching results, the target user's resource usage offset path and design style evolution path are generated. State trend disturbance analysis is performed on the resource usage offset path and design style evolution path to extract the state trend disturbance characteristics of the target user in different time periods. Based on the state trend disturbance characteristics in different time periods, the preference feature data and outlier feature data of the target user are obtained, and a target resource recommendation strategy for the target user is generated according to the preference feature data and outlier feature data. Based on the target resource recommendation strategy, learning resources about creative design are recommended to the target user.

2. The method for intelligently recommending creative design learning resources according to claim 1, characterized in that: Based on the resource creation related data, style matching is performed on local resource usage characteristics and local design behavior characteristics, including: Resource creation-related data includes multiple learning resources associated with multiple target creation styles. The element application reference sequence for each target creation style is determined, and the local element application sequence of each local design behavior feature is extracted. Each local element application sequence is matched with multiple element application reference sequences to generate the local behavior style matching results of the target user in different time periods. Determine the local resource usage sequence corresponding to each local resource usage feature, generate a resource usage reference sequence for each target creative style based on the multiple learning resources associated with the target creative style, match each local resource usage sequence with multiple resource usage reference sequences respectively, and generate local resource style matching results for the target user in different time periods.

3. The method for intelligently recommending creative design learning resources according to claim 2, characterized in that: Perform state trend disturbance analysis on resource usage offset paths and design style evolution paths to extract the state trend disturbance features of target users in different time periods, including: Sliding window detection is performed on the resource usage offset path and design style evolution path respectively, and the resource usage offset state sequence and design style evolution state sequence of each time period are extracted. The linear direction parameters and trend deviation parameters corresponding to the offset state sequence and design style evolution state sequence in each time period are calculated respectively. The trend disturbance parameters in each time period are calculated based on the linear direction parameters and trend deviation parameters, and the resource usage trend disturbance sequence and design style trend disturbance sequence of the target user are generated.

4. The method for intelligently recommending creative design learning resources according to claim 3, characterized in that: Obtain target users' preference feature data and outlier feature data based on the state trend disturbance characteristics in different time periods, including: Based on the resource usage offset path and design style evolution path, a resource creation consistency matrix for multiple target creation styles is constructed, and the style alignment index of each target creation style is extracted based on the resource creation consistency matrix. The trend disturbance index of each target creative style is calculated based on the resource usage trend disturbance sequence and the design style trend disturbance sequence. The stability score of each target creative style is determined based on the style alignment index and the trend disturbance index. According to the stability score, the preference feature data of the target users containing multiple stable styles and the outlier feature data containing multiple outlier styles are generated.

5. The method for intelligently recommending creative design learning resources according to claim 4, characterized in that: Generate a target resource recommendation strategy for target users based on preference feature data and outlier feature data, including: A global style preference evolution optimization analysis is performed on the target users' preference feature data and outlier feature data, including extracting the global preference evolution index of each target creative style from the resource usage trend perturbation sequence and the design style trend perturbation sequence, optimizing the global style preference evolution of the stability score through the global preference evolution index, generating a dynamic preference score for each target creative style, and generating a target resource recommendation strategy for target users regarding multiple learning resources based on the dynamic preference score.

6. The method for intelligently recommending creative design learning resources according to claim 4, characterized in that: The resource creation consistency matrix includes the frequency of occurrence of any two target creation styles as dominant styles in the resource usage offset path and the design style evolution path within the same period.

7. An intelligent recommendation system for creative design learning resources, characterized by: The system is used to implement the method for intelligently recommending creative design learning resources according to any one of claims 1 to 6, comprising: The local feature extraction module is used to collect the target user's historical resource usage data and historical work data related to creative design learning, segment the historical resource usage data and historical work data into time periods, and extract the target user's local resource usage features and local design behavior features in different time periods; The style matching analysis module is used to obtain resource creation association data about learning resources and target creation styles, and perform style matching on local resource usage characteristics and local design behavior characteristics based on the resource creation association data, thereby generating local resource style matching results and local behavior style matching results for target users in different time periods. The state trend disturbance analysis module is used to generate the target user's resource usage offset path and design style evolution path based on multiple sets of local resource style matching results and local behavior style matching results. The module then performs state trend disturbance analysis on the resource usage offset path and design style evolution path to extract the state trend disturbance characteristics of the target user in different time periods. The learning resource recommendation optimization module obtains the target user's preference feature data and outlier feature data based on the state trend disturbance characteristics in different time periods, generates the target user's target resource recommendation strategy based on the preference feature data and outlier feature data, and recommends learning resources about creative design to the target user based on the target resource recommendation strategy.

Images