Intelligent retrieval system for film and television special effects materials

Abstract

The present invention discloses an intelligent retrieval system for film and television special effects materials, which relates to the field of intelligent retrieval technology. It includes a data storage module: used to establish a material library; a data collection module: collects user search records; a data processing module: calculates the first-level matching coefficient and the second-level matching coefficient of the input keyword; combines the first-level matching coefficient and the second-level matching coefficient to calculate the matching factor of the input keyword and update the search keyword; and a data output module: sorts film and television special effects materials for user selection. The present invention collects input keywords and matches film and television special effects materials with the input keywords, calculates the matching factor based on the time and frequency of the input keywords, updates the search keywords of the film and television special effects materials, and sorts the search keywords. It can change the matching between film and television special effects materials and search keywords according to the user's thinking habits, and ensures the accuracy of the search when new keywords appear.

Description

Technical Field

[0001] The present invention relates to the field of intelligent retrieval technology, and in particular to an intelligent retrieval system for film and television special effects materials. Background Art

[0002] The intelligent retrieval method for film and television special effects materials relies on artificial intelligence, computer vision and natural language processing technologies to standardize and annotate metadata (including manual semantic tags and AI automatic feature extraction) for multimodal materials such as videos, images, 3D models, and audio. It extracts low-level visual features (color, texture, shape) and high-level semantic features through models such as CNN and Transformer, analyzes audio spectrum features using MFCC or CNN, and establishes semantic associations between text and visual / audio materials with the help of cross-modal models such as CLIP. During retrieval, it integrates content-based feature matching (such as cosine distance calculation), semantic retrieval (models such as BERT parse user natural language queries and convert them into semantic vectors), and interactive feedback optimization (reinforcement learning adjusts retrieval strategies based on user tags) to achieve efficient and accurate matching of massive materials, supporting the full-process material screening needs in film and television production, from storyboard design to post-editing.

[0003] In material retrieval, what determines the retrieval efficiency is the matching mechanism between the retrieval keywords and the material library, such as the intelligent matching system for advertising design materials with patent publication number CN116861046A. First, the material storage end stores the basic materials, and then collects product information and design information through the design information collection end, retrieves the keywords, and obtains the adapted materials. Then, the priority of the adapted materials is calculated to obtain the keyword priority. At the same time, the proportion of blank areas and the standard proportion of adapted elements are extracted and compared. The complexity of the elements in the design screen is superimposed to obtain the screen complexity value, which is compared with the preset complexity value to obtain the element complexity value of the adapted elements and obtain the comprehensive priority value of the adapted elements. The display terminal then arranges and displays the adapted elements according to the size of the comprehensive priority value, so that when relevant personnel retrieve materials, they can give priority to displaying the advertisements of the products being designed and the corresponding design screens that best match the materials.

[0004] Materials for film and television special effects are identified by keywords manually or by intelligent recognition systems. In the matching of keywords and materials using the above-mentioned method and methods based on the same principle, the intelligent recognition system can only assign keywords based on the set keyword types. When new keywords appear or the user group's understanding of the same special effects shifts, inaccurate matching will occur, reducing the accuracy of the retrieval. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent retrieval system for film and television special effects materials to solve the problems raised in the above background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solution: an intelligent retrieval system for film and television special effects materials, comprising:

[0007] Data storage module: used to collect film and television special effects materials, establish a material library, and match film and television special effects materials with search keywords;

[0008] Data collection module: collects search records of different users within a continuous login period;

[0009] Data processing module:

[0010] According to the continuous search records before the user downloads the film and television special effects materials, the first-level matching coefficients of the input keywords entered by the user are assigned from high to low according to the order of search time;

[0011] Calculate the search frequency of the input keyword based on the user's continuous search records before downloading film and television special effects materials, and assign a secondary matching coefficient to the input keyword based on the search frequency;

[0012] The first-level matching coefficient and the second-level matching coefficient of the same input keyword are combined, and the matching factor of the input keyword is calculated in such a way that the weight of the first-level matching coefficient is greater than the weight of the second-level matching coefficient;

[0013] Sort the input keywords corresponding to the film and television special effects materials according to the size of the matching factors, and mark the input keywords as search keywords for the film and television special effects materials;

[0014] Data output module: When a new search is initiated, the film and television special effects materials are sorted according to the order of the input keywords entered by the user and the order of the search keywords corresponding to the film and television special effects materials for the user to choose.

[0015] Preferably, the specific calculation formula of the matching factor is:

[0016]

[0017] in represents the matching factor, represents the first-level matching coefficient, Represents the secondary matching coefficient.

[0018] Preferably, the specific calculation method of the matching factor is:

[0019] The user's primary matching factor is calculated according to the formula:

[0020]

[0021] in represents the primary matching factor, represents the first-level matching coefficient, represents the secondary matching coefficient;

[0022] Combine the primary matching factors of each user based on their frequency. The specific formula is:

[0023] ,

[0024] in Indicates the user's registration duration in months. Indicates the average number of searches per day on user login days. Indicates the user's idle time, that is, the average time of continuous non-login, in days. represents the matching factor, represents the primary matching factor, represents the secondary matching factor, Indicates calculation of all The average value of .

[0025] Preferably, the specific calculation method of the matching factor is:

[0026] The user's primary matching factor is calculated according to the formula:

[0027]

[0028] in represents the primary matching factor, represents the first-level matching coefficient, represents the secondary matching coefficient;

[0029] Combine the primary matching factors of each user based on their frequency. The specific formula is:

[0030] ,

[0031] in Indicates the user's registration duration in months. Indicates the average number of searches per day on user login days. Indicates the user's idle time, that is, the average time of continuous non-login, in days. represents the matching factor, represents the primary matching factor, represents the secondary matching factor, Indicates calculation of all The average value of Indicates the number of film and television special effects material templates uploaded by users. Indicates the total download volume of film and television special effects templates uploaded by users.

[0032] Preferably, the sorting of film and television special effects materials according to the input keywords input by the user specifically includes:

[0033] According to the input keywords entered by the user, all the film and television special effects materials in the material library are traversed and all the film and television special effects materials that match the input keywords are selected;

[0034] The search keywords of the selected film and television special effects materials are compared according to the priority sorting, and the selected film and television special effects materials are sorted according to the time when the input keywords appear.

[0035] Preferably, the sorting of film and television special effects materials according to the input keywords input by the user specifically includes:

[0036] According to the order of the input keywords entered by the user, the first input keyword is assigned a weight coefficient of 0.5, and the weight coefficient of each subsequent input keyword is half of the weight coefficient of the previous input keyword;

[0037] For each film and television special effects material, retain the first several search keywords, traverse all film and television special effects materials in the material library, select all film and television special effects materials that match the input keywords, and reversely label the search keywords of each selected film and television special effects material;

[0038] The selected film and television special effects materials are sorted from large to small according to the sum of the product of the weight coefficient of the input keyword and the search keyword label.

[0039] Preferably, the specific calculation process of the first-level matching coefficient includes:

[0040] Record the keywords entered by the user in the search history over a continuous period of time, and re-record them when the user downloads film and television special effects materials;

[0041] The input keywords are numbered in reverse order according to the search order. For repeated input keywords, the one with the earliest time is used as the reference. The labels of the input keywords are normalized to obtain the first-level matching coefficient.

[0042] Preferably, the specific calculation process of the secondary matching coefficient includes:

[0043] Record the keywords entered by the user in the search history over a continuous period of time, and re-record them when the user downloads film and television special effects materials;

[0044] Sort the input keywords by the number of times they appear, and normalize the number of times the input keywords appear to obtain the secondary matching coefficient.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] The input keywords are collected and matched with the film and television special effects materials. The matching factor is calculated according to the time and frequency of the input keywords, the search keywords of the film and television special effects materials are updated, and the search keywords are sorted. It can change the matching of film and television special effects materials and search keywords according to the user's thinking habits, and ensure the accuracy of the search when new keywords appear and the user's search cognition changes.

[0047] At the same time, it can sort the film and television special effects materials according to the order of input keywords and the order of the film and television special effects materials' own search keywords, and put the film and television special effects materials with the highest matching degree at the front of the search list, so that users can quickly filter.

[0048] In addition, the overall retrieval mechanism of the retrieval system can be changed according to the usage habits of many users and the degree of user dependence on the retrieval system, making the retrieval system more suitable for users who use the system for a long time and improving user stickiness. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic diagram of the process of intelligent retrieval of film and television special effects materials according to the present invention;

[0050] Figure 2 Schematic diagram of the calculation process of the matching factor in the first embodiment of the present invention;

[0051] Figure 3 Schematic diagram of the calculation process of the matching factor in the second and third embodiments of the present invention;

[0052] Figure 4 A schematic diagram of the process of sorting materials in the present invention;

[0053] Figure 5 Schematic diagram of the user search interface of the present invention. DETAILED DESCRIPTION

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0055] In this application, for ease of understanding, the method steps used do not need to be executed in the order of the steps in this embodiment during actual operation. In other embodiments, these steps may be performed simultaneously or in a different order.

[0056] Example 1:

[0057] Including the user's input keywords in the search system and updating the search keywords of film and television special effects materials according to the time and order of the input keywords can make the search system more adaptable to the user's search habits and improve the search efficiency.

[0058] like Figure 1 and Figure 2 As shown, the present invention provides a technical solution: an intelligent retrieval system for film and television special effects materials, comprising:

[0059] Data storage module: used to collect film and television special effects materials, establish a material library, and match film and television special effects materials with search keywords;

[0060] Data collection module: collects search records of different users within a continuous login period;

[0061] Data processing module:

[0062] According to the continuous search records before the user downloads the film and television special effects materials, the user's input keywords are assigned a first-level matching coefficient from high to low according to the order of search time;

[0063] Calculate the search frequency of the input keyword based on the user's continuous search records before downloading film and television special effects materials, and assign a secondary matching coefficient to the input keyword based on the search frequency;

[0064] The first-level matching coefficient and the second-level matching coefficient of the same input keyword are combined, and the matching factor of the input keyword is calculated in such a way that the weight of the first-level matching coefficient is greater than the weight of the second-level matching coefficient;

[0065] Sort the input keywords corresponding to the film and television special effects materials according to the size of the matching factors, and mark the input keywords as search keywords for the film and television special effects materials;

[0066] Data output module: When a new search is initiated, the film and television special effects materials are sorted according to the order of the input keywords entered by the user and the order of the search keywords corresponding to the film and television special effects materials for user selection.

[0067] It should be noted that establishing a material library requires collecting a certain amount of film and television special effects materials, and then manually matching the search keywords, or using an intelligent matching system (such as AI, visual recognition) to intelligently identify and automatically match the search keywords. These are all existing technologies and will not be elaborated here.

[0068] The specific calculation process of the first-level matching coefficient includes:

[0069] Record the keywords entered by the user in the search history over a continuous period of time, and re-record them when the user downloads film and television special effects materials;

[0070] The input keywords are numbered in reverse order according to the search order. For repeated input keywords, the one with the earliest time is used as the reference. The labels of the input keywords are normalized to obtain the first-level matching coefficient.

[0071] It should be noted that for ease of understanding, simulated data is used as follows:

[0072] Assume that a user's continuous time retrieval records are as follows:

[0073] First search (keyword input): explosion, particle, flame;

[0074] Second search (keyword input): explosion, technology, flame;

[0075] The third search (enter keywords): explosion, flame, splash.

[0076] After three searches, the user downloaded a film and television special effects material named "Wasteland Wilderness Explosion-Second Edition". The original corresponding search keywords for this material were explosion, wasteland, flame, and ultra-large range.

[0077] Sort and label the input keywords: explosion (5), particle (4), flame (3), technology (2), splash (1).

[0078] Then perform normalization (minimum-maximum normalization is used here for ease of calculation), the formula is:

[0079]

[0080] in Represents the normalized value. Indicates the data that needs to be normalized. Indicates the minimum value of the values ​​that need to be normalized. Indicates the maximum value of the values ​​that need to be normalized.

[0081] In order to facilitate calculation (to prevent the data from being at the extreme values ​​of 0 and 1), when normalizing, one data is added before and after the label, and the label can be changed to 6, 5, 4, 3, 2, 1, 0. After normalization, it is 1, 0.83, 0.67, 0.5, 0.33, 0.17, 0. The corresponding first-level matching coefficients of the input keywords explosion, particle, flame, technology, and splash are 0.83, 0.67, 0.5, 0.33, and 0.17 respectively.

[0082] The specific calculation process of the secondary matching coefficient includes:

[0083] Record the keywords entered by the user in the search history over a continuous period of time, and re-record them when the user downloads film and television special effects materials;

[0084] Sort the input keywords by the number of times they appear, and normalize the number of times the input keywords appear to obtain the secondary matching coefficient.

[0085] It should be noted that, for the convenience of calculation, the above simulation data is continued to be used to demonstrate the calculation of the secondary matching coefficient:

[0086] The input keywords are sorted by the number of times they appear (the number is in brackets, and input keywords with the same number of times are sorted randomly): explosion (3), flame (3), technology (1), particle (1), splash (1).

[0087] The same normalization method is used as the first-level matching coefficient, so it will not be demonstrated here (when normalizing, add a data before and after the number, and the label can be changed to 4, 3, 3, 1, 1, 1, 0). The normalized results are: the second-level matching coefficients of explosion, flame, particle, technology, and splash are 0.75, 0.75, 0.25, 0.25, and 0.25 respectively.

[0088] like Figure 2 As shown in Figure 2, the specific calculation formula for the matching factor is:

[0089]

[0090] in represents the matching factor, represents the first-level matching coefficient, Represents the secondary matching coefficient.

[0091] According to the secondary matching coefficient Different value ranges ( <0.3, 0.3≤ ≤0.6, ≥0.6) using different expressions. This segmented processing method allows the secondary matching coefficients in different intervals to be The impact can be modeled more accurately because the different secondary matching coefficients Within the interval, the secondary matching coefficient Matching Factor The contribution proportion is different. For example, when the user frequently enters the same input keyword, it means that the input keyword has a high degree of matching with the film and television special effects material.

[0092] Exponential function part: ( represent The coefficients of right The exponential function has the property of monotonically increasing and can amplify The changes in The impact, especially in When the deviation is 0.5. In different segments The values ​​of different Within the range Adjustment of impact.

[0093] Logarithmic function part: ( represent The coefficients) form is used to process The logarithmic function can compress The impact of the larger value of The square operation further adjusts the amplitude and shape of the logarithmic function output, making The impact is more in line with actual needs.

[0094] It should be noted that, for ease of understanding, the first-level matching coefficient obtained by the above calculation is used ( =0.83, 0.67, 0.5, 0.33, 0.17) and the secondary matching coefficient ( =0.75, 0.75, 0.25, 0.25, 0.25) into the formula to get the matching factors of explosion, flame, particle, technology, and splash They are 1.548, 1.313, 1.122, 0.911 and 0.826 respectively.

[0095] Assuming that the original search keywords (matching factors are in brackets) corresponding to the material "Wasteland Wilderness Explosion - Second Edition" are: Explosion (1.441), Wasteland (1.323), Flame (0.991), and Ultra-Large Range (0.642), the new search keyword combination for the material can be obtained by reordering the search keywords (adding new search keywords and updating the matching factors of the previous search keywords): Explosion (1.548), Wasteland (1.323), Flame (1.313), Particles (1.122), Technology (0.911), Splash (0.826), and Ultra-Large Range (0.642).

[0096] Sorting of film and television special effects materials based on the keywords input by the user specifically includes:

[0097] According to the input keywords entered by the user, all the film and television special effects materials in the material library are traversed and all the film and television special effects materials that match the input keywords are selected;

[0098] The search keywords of the selected film and television special effects materials are compared according to the priority sorting, and the selected film and television special effects materials are sorted according to the time when the input keywords appear.

[0099] It should be noted that for ease of understanding, simulated data is used as follows:

[0100] Material 1: The search keywords for "Wasteland Explosion-Second Edition" are:

[0101] Explosions, wasteland, fire, particles, technology, splashes, and a huge range.

[0102] Material 2: The search keywords for "Technology Building Explosion-Homemade Version" are:

[0103] Explosion, building, technology, particles, splash, flame.

[0104] Assume that the keywords input by the user are: explosion, building, and flame. The above two materials have the same search keywords and can appear in the search list.

[0105] The two materials are compared starting from the first search keyword. The first search keyword of Material 1 is "explosion", and the first search keyword of Material 2 is also "explosion", which are the same as the first input keyword. Then compare the second search keyword. Material 1 is "wasteland" and Material 2 is "building". "Building" appears first in the input keyword ("wasteland" does not appear, meaning infinite delay), so Material 2 is ranked before Material 1 in the search list, indicating that it is more in line with the multiple input keywords entered by the user (only two materials are selected here for demonstration. When the user searches, all retrieved materials need to be sorted in this way). It is ranked at the front of the list for user selection, thereby improving search efficiency.

[0106] Example 2:

[0107] In Example 1, the search keywords for film and television special effects materials are updated through the search records of a single user. The data is mainly for a single user and can only be more suitable for a single user, thereby improving the user experience of a single user who has used this search system for a long time. However, for new users, the search mechanism has not been updated. Based on this, this embodiment provides another matching factor calculation method for new users.

[0108] like Figure 3 As shown in the figure, the specific calculation method of the matching factor is:

[0109] The user's primary matching factor is calculated according to the formula:

[0110]

[0111] in represents the primary matching factor, represents the first-level matching coefficient, represents the secondary matching coefficient;

[0112] Combine the primary matching factors of each user based on their frequency. The specific formula is:

[0113] ,

[0114] in Indicates the user's registration duration in months. Indicates the average number of searches per day on user login days. Indicates the user's idle time, that is, the average time of continuous non-login, in days. represents the matching factor, represents the primary matching factor, represents the secondary matching factor, Indicates calculation of all The average value of .

[0115] Take the natural logarithm of the search frequency to smooth out high-frequency users and avoid = 0. The addition of 1 is to ensure the domain of the logarithmic function.

[0116] It should be noted that for the convenience of calculation, the simulated data are as follows:

[0117] Assuming that the usage data of two old users is used, the primary matching factors of the search keywords "explosion" and "wasteland" for the material 1 "Wasteland Wilderness Explosion - Second Edition" in Example 1 are calculated as shown in Table 1 below (the specific calculation process is the same as that of Example 1 and will not be demonstrated here):

[0118] Table 1: Old user data table 1

[0119]

[0120] For new users, the ranking calculation method for the two search keywords in Material 1 "Wasteland Explosion - Second Edition" is:

[0121] For the search keyword "explosion":

[0122] Secondary matching factor for user 1 =1.600, the primary matching factor for user 2 =0.668, so the matching factor for new users is =0.5×(1.600+0.668)=1.134;

[0123] For the search keyword "wasteland":

[0124] Secondary matching factor for user 1 =1.368, the primary matching factor of user 2 =0.951, so the matching factor for new users is =0.5×(1.368+0.951)≈1.160.

[0125] Therefore, for new users, the search keyword "wasteland" for material 1 ranks before "explosion".

[0126] For the convenience of calculation, only the data of two old users are selected here. In actual use, more old users can be selected for calculation based on certain screening conditions to obtain more accurate results. The screening conditions can be that the registration time is greater than a certain value or the number of daily searches is greater than a certain value. There is no specific restriction.

[0127] The matching factors of several old users are used as the primary matching factors for new users. Different weights are assigned to the primary matching factors based on each user's registration duration, window time, and average search times to obtain the secondary matching factors. The average value is then calculated as the matching factor for the new user, making up for the lack of new user data.

[0128] Example 3:

[0129] In the second embodiment, when calculating the user's secondary matching factor, only the user's registration duration, average search times, and window time are considered. When the system supports users to upload materials independently and download materials uploaded by others, this statistical method is not accurate enough. Based on this, this embodiment provides another method for calculating the secondary matching factor to improve the calculation accuracy.

[0130] The specific calculation method of the matching factor is:

[0131] The user's primary matching factor is calculated according to the formula:

[0132]

[0133] in represents the primary matching factor, represents the first-level matching coefficient, represents the secondary matching coefficient;

[0134] Combine the primary matching factors of each user based on their frequency. The specific formula is:

[0135] ,

[0136] in Indicates the user's registration duration in months. Indicates the average number of searches per day on user login days. Indicates the user's idle time, that is, the average time of continuous non-login, in days. represents the matching factor, represents the primary matching factor, represents the secondary matching factor, Indicates calculation of all The average value of Indicates the number of film and television special effects material templates uploaded by users. Indicates the total download volume of film and television special effects templates uploaded by users.

[0137] use To handle the number of uploaded files, add 1 to avoid =0, the square root is undefined. The square root function can give a certain positive incentive to the number of uploaded files, but the growth rate will gradually slow down. To handle the download volume, e (the base of the natural logarithm, approximately equal to 2.71828) is added to ensure that = 0, the logarithmic function is defined and has a value of 1 (because ln(e) = 1). The logarithmic function can smooth out the high-download users and prevent a small number of high-download users from having too much influence on the secondary matching factor. This item is used to fine-tune the situation when the number of uploaded files and the download amount are small. or When it is very small, this term will be close to 1 and have little effect on the secondary matching factor; or As increases, this term gradually decreases, thus improving the secondary matching factor somewhat (because the denominator decreases). However, the adjustment here is gentle, mainly smoothing this effect through the fourth root.

[0138] It should be noted that for the convenience of calculation, the simulated data is as follows:

[0139] Assuming that the usage data of two old users is used, the primary matching factor of the search keyword "explosion" for the material 1 "Wasteland Explosion - Second Edition" in Example 2 is calculated as shown in Table 2 below (the specific calculation process is the same as that of Example 1 and will not be demonstrated here):

[0140] Table 2: Old user data table 2

[0141]

[0142] Substitute the data of the two users in Table 2 to calculate the secondary matching factor of user 1 ≈1.587, the secondary matching factor for user 2 ≈0.526, matching factor for new users =0.5×(1.587+0.526)≈1.057.

[0143] The same method can also be used to calculate the matching factors of other search keywords of material 1. The search keywords of material 1 can be sorted based on the size of the calculated matching factors. This is suitable for new users, solves the problem of new users having fewer search records, and improves the search speed of new users.

[0144] Example 4:

[0145] In Example 1, when sorting film and television special effects materials based on the input keywords entered by the user, the retrieved film and television special effects materials are sorted according to the order of the input keywords. The calculation is simple, but the sorting sensitivity and accuracy are low simply based on the order of order. Based on this, this embodiment provides another sorting method to improve the accuracy of sorting and facilitate users to select film and television special effects materials.

[0146] like Figure 4 As shown, the sorting of film and television special effects materials based on the input keywords entered by the user specifically includes:

[0147] According to the order of the input keywords entered by the user, the first input keyword is assigned a weight coefficient of 0.5, and the weight coefficient of each subsequent input keyword is half of the weight coefficient of the previous input keyword;

[0148] For each film and television special effects material, retain the first several search keywords, traverse all film and television special effects materials in the material library, select all film and television special effects materials that match the input keywords, and reversely label the search keywords of each selected film and television special effects material;

[0149] The selected film and television special effects materials are sorted from large to small according to the sum of the product of the weight coefficient of the input keyword and the search keyword label.

[0150] It should be noted that for the convenience of calculation, the following simulated data is used (the same data as in Example 1 is used, but only the first four search keywords are retained. In actual use, the specific number of retained keywords can be set according to the hardware specifications. The more retained keywords, the more accurate the calculation, but the higher the calculation requirements for the hardware):

[0151] Material 1: The search keywords for "Wasteland Explosion - Second Edition" are (the numbers in brackets are in reverse order):

[0152] Explosion (4), Wasteland (3), Fire (2), Particles (1).

[0153] Material 2: The search keywords for "Technology Building Explosion-Homemade Version" are:

[0154] Explosion (4), Building (3), Technology (2), Particle (1).

[0155] Assume that the keywords input by the user are: explosion, building, and flame. The above two materials have the same search keywords and can appear in the search list.

[0156] The three input keywords are assigned weights of 0.5, 0.25, and 0.125 respectively;

[0157] For material 1, the sum of the weight coefficient of the input keyword and the product of the search keyword label is 0.5×4+0.25×0+0.125×2=2.25;

[0158] For material 2, the sum of the weight coefficient of the input keyword and the product of the search keyword label is 0.5×4+0.25×3+0.125×0=2.75.

[0159] Since 2.25 is less than 2.75, material 2 is ranked before material 1 in the search list.

[0160] By quantifying the sorting priority of each material based on the input keywords, the retrieved materials can be sorted more accurately in the search list, further improving the efficiency of users in retrieving the required materials.

[0161] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is limited by the accompanying embodiments and their equivalents.

Claims

1. Intelligent retrieval system for film and television special effects materials, including: Data storage module: used to collect film and television special effects materials, establish a material library, and match film and television special effects materials with search keywords; Its characteristics are: Data collection module: collects search records of different users within a continuous login period; Data processing module: According to the continuous search records before the user downloads the film and television special effects materials, the user's input keywords are assigned a first-level matching coefficient from high to low according to the order of search time; Calculate the search frequency of the input keyword based on the user's continuous search records before downloading film and television special effects materials, and assign a secondary matching coefficient to the input keyword based on the search frequency; The first-level matching coefficient and the second-level matching coefficient of the same input keyword are combined, and the matching factor of the input keyword is calculated in such a way that the weight of the first-level matching coefficient is greater than the weight of the second-level matching coefficient; Sort the input keywords corresponding to the film and television special effects materials according to the size of the matching factors, and mark the input keywords as search keywords for the film and television special effects materials; Data output module: When a new search is initiated, the film and television special effects materials are sorted according to the order of the input keywords entered by the user and the order of the search keywords corresponding to the film and television special effects materials for user selection.

2. The intelligent retrieval system for film and television special effects materials according to claim 1, characterized in that: The specific calculation formula of the matching factor is: ； in represents the matching factor, represents the first-level matching coefficient, Represents the secondary matching coefficient.

3. The intelligent retrieval system for film and television special effects materials according to claim 1, characterized in that: The specific calculation method of the matching factor is: The user's primary matching factor is calculated according to the formula: ； in represents the primary matching factor, represents the first-level matching coefficient, represents the secondary matching coefficient; Combine the primary matching factors of each user based on their frequency. The specific formula is: ， ； in Indicates the user's registration duration in months. Indicates the average number of searches per day on user login days. Indicates the user's idle time, that is, the average time of continuous non-login, in days. represents the matching factor, represents the primary matching factor, represents the secondary matching factor, Indicates calculation of all The average value of .

4. The intelligent retrieval system for film and television special effects materials according to claim 1, characterized in that: The specific calculation method of the matching factor is: The user's primary matching factor is calculated according to the formula: ； in represents the primary matching factor, represents the first-level matching coefficient, represents the secondary matching coefficient; Combine the primary matching factors of each user based on their frequency. The specific formula is: ， ； in Indicates the user's registration duration in months. Indicates the average number of searches per day on user login days. Indicates the user's idle time, that is, the average time of continuous non-login, in days. represents the matching factor, represents the primary matching factor, represents the secondary matching factor, Indicates calculation of all The average value of Indicates the number of film and television special effects material templates uploaded by users. Indicates the total download volume of film and television special effects templates uploaded by users.

5. The intelligent retrieval system for film and television special effects materials according to claim 1, characterized in that: The sorting of film and television special effects materials according to the input keywords input by the user specifically includes: According to the input keywords entered by the user, all the film and television special effects materials in the material library are traversed and all the film and television special effects materials that match the input keywords are selected; The search keywords of the selected film and television special effects materials are compared according to the priority sorting, and the selected film and television special effects materials are sorted according to the time when the input keywords appear.

6. The intelligent retrieval system for film and television special effects materials according to claim 1, characterized in that: The sorting of film and television special effects materials according to the input keywords input by the user specifically includes: According to the order of the input keywords entered by the user, the first input keyword is assigned a weight coefficient of 0.5, and the weight coefficient of each subsequent input keyword is half of the weight coefficient of the previous input keyword; For each film and television special effects material, retain the first several search keywords, traverse all film and television special effects materials in the material library, select all film and television special effects materials that match the input keywords, and reversely label the search keywords of each selected film and television special effects material; The selected film and television special effects materials are sorted from large to small according to the sum of the product of the weight coefficient of the input keyword and the search keyword label.

7. The intelligent retrieval system for film and television special effects materials according to claim 1, characterized in that: The specific calculation process of the first-level matching coefficient includes: Record the keywords entered by the user in the search history over a continuous period of time, and re-record them when the user downloads film and television special effects materials; The input keywords are numbered in reverse order according to the search order. For repeated input keywords, the one with the earliest time is used as the reference. The labels of the input keywords are normalized to obtain the first-level matching coefficient.

8. The intelligent retrieval system for film and television special effects materials according to claim 1, characterized in that: The specific calculation process of the secondary matching coefficient includes: Record the keywords entered by the user in the search history over a continuous period of time, and re-record them when the user downloads film and television special effects materials; Sort the input keywords by the number of times they appear, and normalize the number of times the input keywords appear to obtain the secondary matching coefficient.

Images