An animation material asset intelligent classification retrieval reuse management system

Abstract

The application discloses an animation material asset intelligent classification retrieval reuse management system, and belongs to the technical field of animation material management. The system comprises a material library, a label management module, a classification module and a retrieval module. A material association module marks the explicit labels of animation materials in the material library, and forms an explicit label set of the animation materials. The classification module classifies each animation material with the same explicit label set into a category, and obtains each unit classification. The animation materials in the unit classification are prioritized, and a priority sequence corresponding to the unit classification is obtained. Each unit classification is analyzed based on the highest retrieval efficiency, and each material classification is obtained. The priority sequence of each unit classification is used to adjust the corresponding animation materials of the corresponding material classification in layers. The retrieval module identifies and responds to a retrieval request initiated by a user, and performs retrieval matching according to the layered priority order.

Description

Technical Field

[0001] This invention belongs to the field of animation material management technology, specifically an intelligent classification, retrieval, and reuse management system for animation material assets. Background Technology

[0002] With the rapid development of various animation content industries, the number of animation material resources has increased significantly. Currently, the animation production field generally adopts a centralized and unified material library, relying on a supporting database to complete the unified storage, tag archiving, and shared retrieval of materials, effectively solving the problems of scattered material management and difficulty in sharing. However, the existing unified material library model has obvious technical shortcomings and cannot adapt to the personalized usage needs of different production personnel and different production scenarios. The current material library uses a unified and fixed tag system and search rules. The tag dimensions and filtering logic are globally uniform, which cannot be personalized according to user habits, search preferences, and business scenarios. The search methods are highly homogenized and lack accuracy.

[0003] Meanwhile, existing material library database retrieval mostly adopts a global traversal matching mode, which results in long retrieval times and high response latency when faced with massive amounts of inbound materials. In addition, the existing material library database's tag classification structure is static and fixed, without dynamic optimization based on user retrieval logs, material call frequency, and other data within the database. There is no hierarchical retrieval scheduling mechanism, high-frequency reused materials cannot be retrieved quickly, and a large number of inefficient materials consume database retrieval resources. Overall, the database material management and retrieval performance is poor, failing to meet the needs of efficient operation and rapid reuse of system material libraries in modern animation production.

[0004] To address the aforementioned shortcomings, this invention proposes an intelligent classification, retrieval, and reuse management system for animation material assets. Summary of the Invention

[0005] To address the problems of the above solutions, this invention provides an intelligent classification, retrieval, and reuse management system for animation material assets.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] An intelligent classification, retrieval, and reuse management system for animation assets includes an asset library, a tag management module, a classification module, and a retrieval module;

[0008] The resource library is used to store various animation resources.

[0009] The material association module connects to the material library, identifies the material tags corresponding to each animation material in the material library, obtains the user's material search records in real time, identifies each material tag of the search application in real time based on the material search records, marks the material tags as explicit tags, marks the explicit tags of the animation material in the material library, and forms the explicit tag set of the animation material.

[0010] Furthermore, filtering prerequisites are set, namely, the number of displayed tags is greater than threshold X1 and the number of animation materials with explicit tags is greater than threshold X2.

[0011] Real-time evaluation is conducted based on the pre-screening criteria to determine whether the pre-screening criteria are met.

[0012] If the evaluation does not meet the pre-screening criteria, no screening process will be performed.

[0013] When the evaluation meets the preconditions for screening, feature extraction is performed on each explicit label based on the preset screening conditions to obtain screening features, and explicit labels whose screening features do not meet the screening conditions are removed.

[0014] The classification module groups animation materials with the same explicit tag set into one category to obtain unit categories; it prioritizes the animation materials within each unit category to obtain the priority sequence corresponding to the unit category; it analyzes each unit category based on the highest retrieval efficiency to obtain material categories, with each material category including at least one unit category; it adjusts the classification of animation materials in the material library based on each material category, and adjusts the corresponding animation materials of each material category in a hierarchical manner based on the priority sequence of each unit category. The retrieval priority of animation materials at different levels is different, and retrieval is prioritized from the material categories corresponding to the higher priority levels.

[0015] Furthermore, the animation assets within each unit category are prioritized, including:

[0016] Identify associated animation materials whose similarity to animation materials is greater than a threshold of X3;

[0017] Estimate the probability of each animation asset being called; identify the probability of each associated animation asset being called;

[0018] The priority value of the animation material is calculated based on the calling probability of the animation material and the calling probability of related animation materials;

[0019] Sort by priority from highest to lowest priority value.

[0020] Furthermore, the priority value calculation formula is as follows:

[0021]

[0022] In the formula: QA is the priority value; DP0 represents the probability of calling the corresponding animation material within the unit category; i represents the associated animation material of the corresponding animation material, i=1, 2, ..., n, where n is the number of associated animation materials; s iThis indicates the similarity between related animation assets and other animation assets; DP i This indicates the probability of calling the corresponding associated animation material.

[0023] Furthermore, the analysis is conducted based on the highest retrieval efficiency among each unit classification, including:

[0024] Based on the classification of each unit, a combination simulation is performed to obtain various simulated classification methods;

[0025] Obtain the user's historical search records, and statistically analyze the historical search needs and the proportion of those needs based on the historical search records;

[0026] Based on various historical search needs, search simulations were performed for each simulated classification method to obtain the search time for each historical search need using the simulated classification method.

[0027] The efficiency evaluation value is calculated based on the evaluation formula, which is:

[0028]

[0029] In the formula: PG is the efficiency evaluation value; j represents the corresponding historical search demand, j=1, 2, ..., n, where n is the number of historical search demands; β j This indicates the percentage of historical search requests; T j This indicates the search duration corresponding to historical search requests;

[0030] The simulated classification method with the lowest efficiency evaluation value is selected as the target classification method. Classification is performed according to the target classification method to obtain the classification of each material.

[0031] Furthermore, based on the priority sequence of each unit category, the corresponding animation materials for each material category are layered and adjusted, including:

[0032] Step SA1: Obtain the priority sequence of each unit category within the material category, and set the minimum call probability of each layer; mark a terminal category for each priority sequence according to the minimum call probability of each layer, and the terminal category is the combination of animation materials determined based on the preset iterative combination method; normalize the priority values ​​in the priority sequence corresponding to each unit category.

[0033] Step SA2: Based on the iterative combination method, mark the animation material combinations in each priority sequence according to priority order;

[0034] When the combination of animation materials intersects with the final category, the remaining animation materials that do not belong to each layer are grouped into one layer. The priority of each layer is determined according to the order in which the layers are determined, and the analysis ends.

[0035] When the animation material combination does not intersect with the final category, mark the first animation material in the priority sequence after the animation material combination as the combination analysis material; when the combination analysis material belongs to the final category, set the corresponding layers for the animation material combination and the final category respectively, determine the priority of each layer according to the layer determination order, and end the analysis; when the combination analysis material does not belong to the final category, proceed to step SA3.

[0036] Step SA3: Identify the priority value of the combined analysis material and the combination priority value of the animation material combination. The combination priority value is the average of the priority values ​​of each animation material in the animation material combination or the priority value of the animation material that is ranked last in the animation material combination; substitute the priority value and the combination priority value into the preset distance formula to calculate the combination distance between the combined analysis material and the animation material combination.

[0037] Step SA4: When the combination distance is not less than the threshold X4, the animation material combination is layered, and the animation material corresponding to the layer in the priority sequence is removed, and the process returns to step SA2;

[0038] When the combined distance is less than the threshold X4, the combined analysis material is merged into the animation material combination to obtain a new animation material combination. The combined analysis material is then identified based on the new animation material combination, and the process returns to step SA3.

[0039] Furthermore, the iterative layering method includes:

[0040] Identify the order of combination, which is either the priority order or the reverse of the priority order; the order of combination is the reverse of the priority order only when determining the end classification.

[0041] Identify the probability of calling the animation material that is ranked first in the combination order;

[0042] When the calling probability is not lower than the lowest calling probability of the layer, the animation material is treated as an independent animation material combination;

[0043] When the call probability is lower than the minimum call probability of the layer, the animation material ranked second in the combination order is combined with the animation material ranked first to obtain an initial combination, and the call probability of the initial combination is estimated. When the call probability of the initial combination is not lower than the minimum call probability of the layer, the initial combination is marked as an animation material combination. When the call probability of the initial combination is lower than the minimum call probability of the layer, the animation material ranked third in the combination order is combined with the initial combination to obtain a new initial combination, and the call probability of the new initial combination is estimated. When the call probability of the initial combination is not lower than the minimum call probability of the layer, the initial combination is marked as an animation material combination. When the call probability of the initial combination is lower than the minimum call probability of the layer, the next animation material in the combination order is combined with the initial combination to obtain a new initial combination. This process continues until an animation material combination is obtained.

[0044] Furthermore, the distance formula is:

[0045]

[0046] In the formula: L is the combination distance; QA is the priority value of the combination analysis material; QB is the combination priority value of the animation material combination.

[0047] The retrieval module recognizes and responds to the retrieval request initiated by the user, and performs retrieval and matching starting from the first layer according to the priority of the layers. When the retrieval results of the layer meet the user's retrieval needs, the results are returned directly and the retrieval process is terminated. When the animation materials returned by the layer do not meet the retrieval requirements, the module automatically enters the next layer to continue the retrieval until the animation materials that meet the retrieval requirements are obtained.

[0048] Furthermore, add implicit tags to the animation materials in the material library. Implicit tags are material tags that are not explicit tags.

[0049] When explicit tags alone cannot yield search results that meet the search requirements, a secondary search is performed based on the implicit tags of the animation materials.

[0050] Furthermore, a secondary search is performed based on the implicit tags of the animation materials, including:

[0051] The search is performed by adding implicit tags according to the hierarchical priority.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] This invention employs a dual-layer tagging system of explicit and implicit tags. Frequently used search criteria are uniformly grouped into explicit tags for routine searches, while deeper features of the materials are set as implicit tags to supplement search resources. This ensures fast response times for daily searches and allows for high-precision supplementary filtering using implicit tags when explicit tag results are insufficient, avoiding the problems of tag redundancy and chaotic search matching inherent in traditional single-tag systems. Furthermore, this invention innovatively groups materials with the same explicit tags into unit classes and then performs dynamic classification simulation based on user search records and usage habits, aiming to maximize search efficiency. This overcomes the shortcomings of traditional static classification models that cannot adapt to actual user needs. Subsequent hierarchical storage is based on material classification, ultimately forming a hierarchical search architecture. During the search process, there is no need for a global traversal of the entire material library; matching only needs to begin from the highest priority first layer. If the result of the current layer meets the requirements, it is returned directly, significantly reducing the traversal range of a single search. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a block diagram illustrating the principle of the present invention. Detailed Implementation

[0056] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0057] like Figure 1 As shown, an intelligent classification, retrieval, and reuse management system for animation material assets includes a material library, a tag management module, a classification module, and a retrieval module.

[0058] The resource library is used to store various animation resources.

[0059] The material association module connects to the material library and identifies the material tags corresponding to each animation material in the library, such as material type, resolution, file format, duration, size, dimensions, scene, character type, style, era, atmosphere, renderer, frame rate, skeletal binding, lighting type, etc.; it identifies the material based on its original tags; it acquires the user's material search records in real time, identifies each material tag of the search application in real time based on the material search records, marks the material tags as explicit tags, and marks each animation material in the material library with the explicit tags that the animation material has based on each explicit tag, forming the explicit tag set of the animation material, which is composed of each explicit tag corresponding to the animation material.

[0060] In one embodiment, as time goes on, the number of explicit tags gradually increases, and in extreme cases, they may cover all material tags. Therefore, in this embodiment, the explicit tags are filtered based on the call duration interval of each explicit tag, and explicit tags with a call interval greater than a preset value are canceled, that is, the corresponding material tags are no longer used as explicit tags.

[0061] In one embodiment, the filtering criteria can also include the probability of being called, the percentage of times called, and the number of times called, etc., of explicit tags.

[0062] In one embodiment, based on the above embodiment, a filtering prerequisite is set: the number of displayed tags is greater than threshold X1 and the number of animation materials with explicit tags is greater than threshold X2; after the filtering prerequisite is met, explicit tags are filtered according to the above embodiment.

[0063] In one embodiment, the filtering prerequisite is that the number of displayed tags is greater than threshold X1 or the number of animation materials with explicit tags is greater than threshold X2.

[0064] The classification module groups animation materials with the same explicit tag set into one category to obtain unit categories; it prioritizes the animation materials within each unit category to obtain the priority sequence corresponding to the unit category; it analyzes each unit category based on the highest retrieval efficiency to obtain material categories, with each material category including at least one unit category; it adjusts the classification of animation materials in the material library based on each material category, and adjusts the corresponding animation materials of each material category in a hierarchical manner based on the priority sequence of each unit category. The retrieval priority of animation materials at different levels is different, and retrieval is prioritized from the material categories corresponding to the higher priority levels.

[0065] In one embodiment, prioritizing the animation assets within a unit category includes:

[0066] Identify the number of times each animation asset is called and prioritize them according to the number of times they are called.

[0067] In one embodiment, prioritizing the animation assets within a unit category includes:

[0068] The system can predict the calling probability of each animation material in real time and prioritize them according to the calling probability from high to low. The calling probability can be calculated by combining the number of calls within a period; or it can be based on AI and other intelligent technologies to analyze user habits and preferences to determine a more accurate calling probability.

[0069] In one embodiment, prioritizing the animation assets within a unit category includes:

[0070] Identify associated animation materials whose similarity to animation materials is greater than a threshold X3. This includes all animation materials inside and outside the unit category. As long as the similarity is greater than the threshold X3, they are considered as associated animation materials.

[0071] Estimate the probability of each animation asset being called; identify the probability of each associated animation asset being called;

[0072] The priority value of the animation material is calculated based on the calling probability of the corresponding animation material and the calling probability of the associated animation material;

[0073] Sort by priority from highest to lowest priority value.

[0074] In one embodiment, priority value = probability of calling animation material + ∑ conversion coefficient × probability of calling associated animation material.

[0075] The conversion rate is initially set by the platform and can be adjusted by the user as needed; it can also be set based on the similarity score.

[0076] In one embodiment, the priority value is calculated using the following formula:

[0077]

[0078] In the formula: QA is the priority value; DP0 represents the probability of calling the corresponding animation material within the unit category; i represents the associated animation material of the corresponding animation material, i=1, 2, ..., n, where n is the number of associated animation materials; s i This indicates the similarity between related animation assets and other animation assets; DP i This indicates the probability of calling the corresponding associated animation material.

[0079] This priority value calculation breaks through the limitations of traditional methods that only sort by the frequency of single material calls. It incorporates the popularity of cross-unit category related materials into the weight, which not only fits the real scenario of material linkage and reuse in animation production, but also allows materials with high relevance potential to be prioritized, greatly compressing the search path and significantly improving the retrieval efficiency of high-frequency related materials without increasing search computing power.

[0080] In one embodiment, the analysis is performed based on the highest retrieval efficiency among the various unit categories. The analysis is conducted using a simulation method, that is, the various unit categories are simulated and combined to determine the various classification methods. The retrieval efficiency is estimated based on user retrieval habits, work direction, etc., and the materials are classified according to the highest retrieval efficiency to obtain the various material categories.

[0081] For example, using historical user search records, user usage habit data, and work scope as inputs, and taking maximizing search efficiency as the optimization goal, classification simulation calculations are performed. The classification simulation calculations are implemented using reinforcement learning algorithms, with search response time as the core indicator of the reward function. Through multiple rounds of iteration, the affiliation relationship of different unit classes in material classification is adjusted until the average search response time of the overall system converges to the minimum value, thus obtaining the final stable material classification result.

[0082] For example, mathematical methods such as enumeration are used to combine and simulate according to the classification of each unit to obtain various simulated classification methods;

[0083] Obtain the user's historical search records, and statistically analyze the historical search needs and their proportions based on these records. The proportion of each historical search need is calculated based on the number of times it was searched.

[0084] Based on various historical search needs, search simulations are performed for each simulated classification method to obtain the search duration for each historical search need using the simulated classification method. The search duration is determined based on the animation materials selected by the user in the corresponding historical search records. The specific simulation process involves classifying according to the simulated classification method and subsequent hierarchical classification, and then searching according to the search needs to estimate the search duration.

[0085] The efficiency evaluation value is calculated based on the evaluation formula, which is:

[0086]

[0087] In the formula: PG is the efficiency evaluation value; j represents the corresponding historical search demand, j=1, 2, ..., n, where n is the number of historical search demands; β j This indicates the percentage of historical search requests; T j This indicates the search duration corresponding to historical search requests;

[0088] The simulated classification method with the lowest efficiency evaluation value is selected as the target classification method. Classification is performed according to the target classification method to obtain the classification of each material.

[0089] In one embodiment, the animation materials corresponding to the corresponding material categories are layered and adjusted based on the priority sequence of each unit category. The layers are divided according to storage levels, such as high-speed storage layer (e.g., NVMe SSD), medium-speed storage layer (e.g., SAS / SATA HDD), and low-speed archiving layer (e.g., object storage / tape). Each level is pre-divided, and resources are allocated according to the calling probability or data volume of the material category. Then, resources are allocated to each unit category within the category. Finally, the level of each animation resource is determined according to the priority sequence.

[0090] In one embodiment, this embodiment does not perform tiering based on actual storage resources. To further improve retrieval efficiency, the tiering adjustment is performed in the following manner:

[0091] Step SA1: Obtain the priority sequence of each unit category within the material category, and set the minimum call probability of each layer, that is, the minimum call probability of each layer after layering cannot be lower than the minimum call probability of that layer; mark an end category for each priority sequence according to the minimum call probability of the layer. The end category is the combination of animation materials determined by starting from the last animation material in the priority sequence sorting based on a preset iterative combination method; normalize the priority values ​​in the priority sequence corresponding to each unit category, such as converting the maximum priority value to 1 and 0 to 0, and perform normalization processing according to the conversion relationship.

[0092] Step SA2: Based on the iterative combination method, mark the animation material combinations in each priority sequence according to priority order;

[0093] When the combination of animation materials intersects with the final category, the remaining animation materials that do not belong to each layer are grouped into one layer. The priority of each layer is determined according to the order in which the layers are determined, and the analysis ends.

[0094] When the animation material combination does not intersect with the final category, mark the first animation material in the priority sequence after the animation material combination as the combination analysis material; when the combination analysis material belongs to the final category, set the corresponding layers for the animation material combination and the final category respectively, determine the priority of each layer according to the layer determination order, and end the analysis; when the combination analysis material does not belong to the final category, proceed to step SA3.

[0095] Step SA3: Identify the priority value of the combined analysis material and the combination priority value of the animation material combination. The combination priority value is the average of the priority values ​​of each animation material in the animation material combination or the priority value corresponding to the animation material ranked last in the animation material combination. In other embodiments, it can also be a representative value (combination priority value) determined based on mathematical statistics, such as the mode, maximum value, or minimum value. Substitute the priority value and the combination priority value into a preset distance formula to calculate the combination distance between the combined analysis material and the animation material combination.

[0096] Step SA4: When the combination distance is not less than the threshold X4, perform layering based on the combination of animation materials, obtain the corresponding layers, remove the animation materials corresponding to the layers in the priority sequence, and return to step SA2;

[0097] When the combined distance is less than the threshold X4, the combined analysis material is merged into the animation material combination to obtain a new animation material combination. The combined analysis material is then identified based on the new animation material combination, and the process returns to step SA3.

[0098] In one embodiment, the distance formula is:

[0099] Combined distance = |priority value - combined priority value|.

[0100] In one embodiment, the distance formula is:

[0101]

[0102] In the formula: L is the combination distance; QA is the priority value of the combination analysis material; QB is the combination priority value of the animation material combination.

[0103] In one embodiment, the iterative hierarchical approach includes:

[0104] Identify the order of combination. If the normal order of combination is the priority order, the order of combination is the reverse of the priority order when determining the end classification.

[0105] Identify the probability of calling the animation material that is ranked first in the combination order;

[0106] When the calling probability is not lower than the lowest calling probability of the layer, the animation material is treated as an independent animation material combination;

[0107] When the call probability is lower than the minimum call probability of the layer, the animation material ranked second in the combination order is combined with the animation material ranked first to obtain an initial combination. The call probability of the initial combination is estimated, that is, the probability of calling an animation material from the initial combination. When the call probability of the initial combination is not lower than the minimum call probability of the layer, the initial combination is marked as an animation material combination. When the call probability of the initial combination is lower than the minimum call probability of the layer, the animation material ranked third in the combination order is combined with the initial combination to obtain a new initial combination. The call probability of the new initial combination is estimated. When the call probability of the initial combination is not lower than the minimum call probability of the layer, the initial combination is marked as an animation material combination. When the call probability of the initial combination is lower than the minimum call probability of the layer, the next animation material in the combination order (the fourth, fifth, and so on) is combined with the initial combination to obtain a new initial combination. This process is repeated until an animation material combination is obtained.

[0108] In one embodiment, the stratification can be further divided into pre-defined call probability intervals for each stratum, and combined analysis can be performed directly based on the call probability intervals to determine each stratum.

[0109] The retrieval module recognizes and responds to user-initiated retrieval requests, and performs retrieval and matching sequentially from the first layer according to the priority of the layers. It only calls explicit tags for matching. When the retrieval results of a layer meet the user's retrieval needs, it directly returns the results and terminates the retrieval process. When the number of animation materials returned by a layer is insufficient or the matching degree is not up to standard, it automatically enters the next layer to continue the retrieval until it finds animation materials that meet the retrieval requirements.

[0110] In one embodiment, to avoid the inability to obtain search results that meet the search requirements by relying solely on explicit tags, implicit tags are added to the animation materials in the material library. Implicit tags are material tags that are not explicit tags, such as material type, resolution, file format, duration, size, and dimensions; explicit tags are material type, resolution, file format, and duration; the remaining size and dimensions are implicit tags.

[0111] When explicit tags alone cannot yield search results that meet the search requirements, a secondary search is performed based on the implicit tags of the animation material to achieve high-precision matching.

[0112] For example, you can search according to the same hierarchical priority and material classification, but then combine implicit tags to search layer by layer.

[0113] The above formulas are all numerical calculations after removing dimensions. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0114] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An intelligent classification, retrieval, and reuse management system for animation material assets, characterized in that, Includes a resource library, a tag management module, a classification module, and a search module; The resource library is used to store various animation resources; The material association module identifies the material tags corresponding to each animation material in the material library; it obtains the user's material search records in real time, identifies each material tag of the search application in real time based on the material search records, marks the material tags as explicit tags, marks the explicit tags of the animation material in the material library, and forms the explicit tag set of the animation material; The classification module groups animation materials with the same explicit tag set into one category to obtain unit categories; it prioritizes the animation materials within each unit category to obtain the priority sequence corresponding to the unit category; it analyzes each unit category based on the highest retrieval efficiency to obtain material categories, with each material category including at least one unit category; it adjusts the classification of animation materials in the material library based on each material category, and adjusts the corresponding animation materials of each material category in a hierarchical manner based on the priority sequence of each unit category. The retrieval priority of animation materials at different levels is different, and retrieval is prioritized from the material categories corresponding to the higher priority levels; The retrieval module recognizes and responds to the retrieval request initiated by the user, and performs retrieval and matching starting from the first layer according to the priority of the layers. When the retrieval results of the layer meet the user's retrieval needs, the results are returned directly and the retrieval process is terminated; when the animation materials returned by the layer do not meet the retrieval requirements, the retrieval continues to the next layer until the animation materials that meet the retrieval requirements are obtained.

2. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 1, characterized in that, Set filtering prerequisites, wherein the number of displayed tags is greater than threshold X1 and the number of animation materials with explicit tags is greater than threshold X2; Real-time evaluation is conducted based on the pre-screening criteria to determine whether the pre-screening criteria are met. If the evaluation does not meet the pre-screening criteria, no screening process will be performed. When the evaluation meets the preconditions for screening, feature extraction is performed on each explicit label based on the preset screening conditions to obtain screening features, and explicit labels whose screening features do not meet the screening conditions are removed.

3. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 1, characterized in that, Prioritize and sort the animation assets within each unit category, including: Identify associated animation materials whose similarity to animation materials is greater than a threshold of X3; Estimate the probability of each animation asset being called; identify the probability of each associated animation asset being called; The priority value of the animation material is calculated based on the calling probability of the animation material and the calling probability of related animation materials; Sort by priority from highest to lowest priority value.

4. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 3, characterized in that, The formula for calculating priority value is: In the formula: QA is the priority value; DP0 represents the probability of calling the corresponding animation material within the unit category; i represents the associated animation material of the corresponding animation material, i=1, 2, ..., n, where n is the number of associated animation materials; s i This indicates the similarity between related animation assets and other animation assets; DP i This indicates the probability of calling the corresponding associated animation material.

5. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 1, characterized in that, The analysis is based on the highest retrieval efficiency among the various unit classifications, including: Based on the classification of each unit, a combination simulation is performed to obtain various simulated classification methods; Obtain the user's historical search records, and statistically analyze the historical search needs and the proportion of those needs based on the historical search records; Based on various historical search needs, search simulations were performed for each simulated classification method to obtain the search time for each historical search need using the simulated classification method. The efficiency evaluation value is calculated based on the evaluation formula, which is: In the formula: PG is the efficiency evaluation value; j represents the corresponding historical search demand, j=1, 2, ..., n, where n is the number of historical search demands; β j This indicates the percentage of historical search requests; T j This indicates the search duration corresponding to historical search requests; The simulated classification method with the lowest efficiency evaluation value is selected as the target classification method. Classification is performed according to the target classification method to obtain the classification of each material.

6. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 1, characterized in that, Based on the priority sequence of each unit category, the corresponding animation materials for each material category are layered and adjusted, including: Step SA1: Obtain the priority sequence of each unit category within the material category, and set the minimum call probability of each layer; mark a terminal category for each priority sequence according to the minimum call probability of each layer, and the terminal category is the combination of animation materials determined based on the preset iterative combination method; normalize the priority values ​​in the priority sequence corresponding to each unit category. Step SA2: Based on the iterative combination method, mark the animation material combinations in each priority sequence according to priority order; When the combination of animation materials intersects with the final category, the remaining animation materials that do not belong to each layer are grouped into one layer. The priority of each layer is determined according to the order in which the layers are determined, and the analysis ends. When the animation material combination does not intersect with the final category, mark the first animation material in the priority sequence after the animation material combination as the combination analysis material; when the combination analysis material belongs to the final category, set the corresponding layers for the animation material combination and the final category respectively, determine the priority of each layer according to the layer determination order, and end the analysis; when the combination analysis material does not belong to the final category, proceed to step SA3. Step SA3: Identify the priority value of the combined analysis material and the combination priority value of the animation material combination. The combination priority value is the average of the priority values ​​of each animation material in the animation material combination or the priority value of the animation material that is ranked last in the animation material combination; substitute the priority value and the combination priority value into the preset distance formula to calculate the combination distance between the combined analysis material and the animation material combination. Step SA4: When the combination distance is not less than the threshold X4, the animation material combination is layered, and the animation material corresponding to the layer in the priority sequence is removed, and the process returns to step SA2; When the combined distance is less than the threshold X4, the combined analysis material is merged into the animation material combination to obtain a new animation material combination. The combined analysis material is then identified based on the new animation material combination, and the process returns to step SA3.

7. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 6, characterized in that, Iterative layering methods include: Identify the order of combination, which is either the priority order or the reverse of the priority order; the order of combination is the reverse of the priority order only when determining the end classification. Identify the probability of calling the animation material that is ranked first in the combination order; When the calling probability is not lower than the lowest calling probability of the layer, the animation material is treated as an independent animation material combination; When the call probability is lower than the minimum call probability of the layer, the animation material ranked second in the combination order is combined with the animation material ranked first to obtain an initial combination, and the call probability of the initial combination is estimated. When the call probability of the initial combination is not lower than the minimum call probability of the layer, the initial combination is marked as an animation material combination. When the call probability of the initial combination is lower than the minimum call probability of the layer, the animation material ranked third in the combination order is combined with the initial combination to obtain a new initial combination, and the call probability of the new initial combination is estimated. When the call probability of the initial combination is not lower than the minimum call probability of the layer, the initial combination is marked as an animation material combination. When the call probability of the initial combination is lower than the minimum call probability of the layer, the next animation material in the combination order is combined with the initial combination to obtain a new initial combination. This process continues until an animation material combination is obtained.

8. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 6, characterized in that, Distance formula: In the formula: L is the combination distance; QA is the priority value of the combination analysis material; QB is the combination priority value of the animation material combination.

9. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 1, characterized in that, Add implicit tags to the animation assets in the asset library. Implicit tags are asset tags that are not explicit tags. When explicit tags alone cannot yield search results that meet the search requirements, a secondary search is performed based on the implicit tags of the animation materials.

10. The intelligent classification, retrieval, and reuse management system for animation material assets according to claim 9, characterized in that, Secondary retrieval based on implicit tags in animation materials, including: The search is performed by adding implicit tags according to the hierarchical priority.

Images