Database-based skill learning data quick search method
By calculating the correlation coefficients between search fields and the popularity of courses, and dynamically sorting them by the number of registrants and favorites, the problem of insufficient accuracy in existing skills learning data retrieval is solved, and efficient and accurate multi-skill integrated query and course recommendation are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TIANJIN JINLONG UNITED EDUCATION TECH GRP CO LTD
- Filing Date
- 2026-03-02
- Publication Date
- 2026-06-02
AI Technical Summary
Existing methods for retrieving skills learning data fail to fully explore the relationships between search fields, resulting in insufficient accuracy of search results, which is insufficient to meet the needs of complex skills. Furthermore, the course sorting does not take into account the popularity of courses and user learning preferences, which affects the user learning experience and training effectiveness.
By calculating the correlation coefficients between search fields, dynamically adjusting the sorting based on the number of course enrollments and favorites, a synergistic relationship between search field combinations is established. Furthermore, by continuously updating cached data through a preset search cycle, the search logic is optimized to improve accuracy and efficiency.
It enables accurate retrieval in scenarios involving multiple professions and skills, reduces system response latency, and improves user learning experience and the matching quality of training resources.
Smart Images
Figure CN122132425A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data retrieval technology, specifically to a database-based method for rapid retrieval of skills learning data. Background Technology
[0002] Against the backdrop of accelerated digital transformation in vocational skills training, academic management systems such as SkillPass have accumulated massive amounts of student data, course resources, and learning behavior records, covering multiple dimensions such as student information, class configuration, and learning progress, providing a rich foundation for skills learning data retrieval. However, existing retrieval methods mostly rely on single-field matching, failing to fully explore the relationships between search fields, resulting in insufficient accuracy of search results and difficulty in meeting users' demands for rapid matching of complex skills. Furthermore, the system's course resources are categorized by occupation, level, etc., and traditional retrieval methods do not dynamically sort them based on course popularity and user learning preferences, further reducing retrieval efficiency.
[0003] With the diversification of vocational skills training, user search needs are becoming increasingly complex. Single-field searches can no longer cover query scenarios involving multiple professions and skills. Existing technologies lack quantitative analysis of the frequency and correlation strength of fields within a search period, making it difficult to predict high-frequency search combinations in advance. This leads to increased duplicate search operations and system response delays. Furthermore, course ranking relies solely on fixed attributes, failing to consider real-time dynamic data such as the number of registrants and favorites, making it impossible to accurately push popular courses that meet market demands, thus impacting user learning experience and training effectiveness.
[0004] Therefore, there is an urgent need to develop an efficient database-based method for rapid retrieval of skills learning data. By mining the correlation patterns between search fields and dynamically adjusting the sorting logic based on course popularity, the accuracy and efficiency of search results can be achieved. Summary of the Invention
[0005] The purpose of this invention is to provide a database-based method for rapid retrieval of skills learning data, thereby solving the above-mentioned technical problems.
[0006] The objective of this invention can be achieved through the following technical solutions: A database-based method for rapid retrieval of skills learning data includes the following steps: Set a preset search period T, retrieve all search fields within the current search period T, count the number of times the same search field appears n, and calculate the frequency of occurrence. I represents the number of search fields within the search period T, and n i This represents the number of times the i-th search field appears.
[0007] As a further aspect of the present invention: calculating the correlation coefficient G between different search fields specifically includes the following steps: Get the search field combination "Ai+Aj", where Ai represents the i-th search field, Aj represents the j-th search field, i≠j, and "+" means inputting both simultaneously; Obtain the frequencies P(Ai) and P(Aj) corresponding to "Ai" and "Aj" in the search field combination "Ai+Aj", and calculate the frequencies. n (Ai∩Aj) represents the number of times that the search fields Ai and Aj are searched simultaneously; Calculate the correlation coefficient G(i,j) between the search fields Ai and Aj = P(Ai∩Aj) / [P(Ai)+P(Aj)].
[0008] As a further aspect of the present invention: Input the search field Ai and perform a search; mark the retrieved skill courses as initial courses Ai; obtain the number of enrollees R and the number of favorites Z for the initial courses; and calculate the popularity X = λ1 × log M R+λ2×Z, where λ1 and λ2 represent the preset first and second coefficients respectively, and M represents the preset number of people in the gradient. The initial courses are sorted in descending order of popularity X; Aj is obtained in the same way.
[0009] As a further aspect of this invention: In the search field combination "Ai+Aj", the first search field Ai is designated as the primary field, and the remaining search fields are designated as secondary fields. The popularity X of the initial course corresponding to the secondary fields is modified as follows: Based on the current original value of popularity X, multiply it with the correlation coefficient G(i,j), and the result is the updated value of popularity X. Reorder the initial courses in the search fields Ai and Aj based on the current popularity X, sort them from highest to lowest popularity X, select the top K initial courses as popular courses, store the popular courses in the cache, K represents the preset number of courses, and record the search field combination "Ai+Aj" and the K popular courses as a linkage relationship. Repeat the above operation to match the correlation relationships of all search field combinations within the current search period T.
[0010] As a further aspect of the present invention: when a user inputs the search field combination "Ai+Aj", popular courses with related relationships are displayed first; When the current search period T ends, the next search period T begins immediately, and the search field combinations and corresponding relationships are updated continuously.
[0011] As a further aspect of the present invention: when the number of search fields in the search field combination is greater than 2, the following steps are performed: Obtain the search field combination "Ai+Aj+...+Am" and divide it into search field combinations "Ai+Aj", ..., and search field combination "Ai+Am". Calculate the correlation coefficients corresponding to the divided search field combinations. Repeat the above steps to filter K popular courses and match the correlation between the search field combination "Ai+Aj+...+Am" and the K popular courses.
[0012] As a further aspect of the present invention: if the total number of all search fields within the current search period T is less than a preset search threshold, then the subsequent steps are stopped.
[0013] As a further aspect of the present invention: when initial courses with equal popularity X appear, the initial course with a shorter publication time is placed in a higher ranking position.
[0014] The beneficial effects of the present invention are: (1) By calculating the correlation coefficient between the search fields, the correlation strength of the field combination is quantified, and the course popularity is corrected by combining the main and auxiliary fields, so that the search results are more in line with the user's composite skill needs, avoiding the one-sidedness of single field search, effectively matching the query scenarios of multiple occupational intersections and multiple skill integration, and solving the pain point of insufficient accuracy of traditional search.
[0015] (2) By updating the correlation between field combinations and popular courses in a preset retrieval cycle, popular courses corresponding to high-frequency combinations are stored in the cache. When users search again, they can prioritize the use of cached data, reducing redundant calculations and database query operations, reducing system response latency, and improving the search experience. At the same time, when the number of search fields is less than the threshold, subsequent steps are stopped to avoid invalid calculations and save system resources.
[0016] (3) The popularity of courses is dynamically calculated by combining the number of registrants and the number of favorites, while taking into account the publication time sorting rules, to ensure that the pushed courses meet market demand and are timely, helping users quickly lock in high-quality and popular resources. For multi-field combination searches, the association relationship is calculated by reasonably splitting the data to ensure the recommendation quality in complex search scenarios, and to provide support for students' personalized learning and optimized allocation of training resources.
[0017] (4) This method can be seamlessly applied to academic management systems such as SkillsPass, adapting to massive student data and diverse course resources. Through dynamically optimized retrieval logic, it helps the platform better respond to the market's demand for vocational training and enhance the platform's service capabilities and core competitiveness. Attached Figure Description
[0018] The invention will now be further described with reference to the accompanying drawings.
[0019] Figure 1 This is a schematic diagram of the structure of the database-based rapid retrieval method for skills learning data according to the present invention; Figure 2 This is a flowchart illustrating the database-based method for rapid retrieval of skills learning data according to the present invention. Detailed Implementation
[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 skilled in the art without creative effort are within the scope of protection of the present invention.
[0021] Please see Figure 1 As shown, this invention is a database-based method for rapid retrieval of skills learning data, comprising the following steps: Set a preset search period T, retrieve all search fields within the current search period T, count the number of times the same search field appears n, and calculate the frequency of occurrence. I represents the number of search fields within the search period T, and n i This represents the number of times the i-th search field appears.
[0022] It's important to note that this section is the foundational step in the rapid retrieval of skills learning data. Its core lies in clarifying the popularity distribution of search fields through periodic statistics. First, a fixed search period T needs to be preset. The period length can be flexibly set according to the platform's search volume, ensuring that the data statistics are both timely and cover a sufficient number of search samples. Then, the system automatically collects search behavior data from all users within the current period T, extracts all search fields, removes duplicates, and categorizes them to determine the number of search field types I, i.e., the total number of unique search fields within the period. Next, for each search field (the i-th type), its total occurrence count ni within period T is counted. This count covers all cases of the same user searching the same field multiple times and different users searching the same field. Finally, the frequency of each search field is calculated. The frequency calculation logic is the occurrence count ni of the i-th field divided by the total occurrence count of all search fields within period T (the sum of all ni). This frequency quantifies the search popularity of each field, providing basic data support for subsequent field association analysis and popular course selection, ensuring that the search logic aligns with users' actual needs.
[0023] In a preferred embodiment, calculating the correlation coefficient G between different search fields specifically includes the following steps: Get the search field combination "Ai+Aj", where Ai represents the i-th search field, Aj represents the j-th search field, i≠j, and "+" means inputting both simultaneously; Obtain the frequencies P(Ai) and P(Aj) corresponding to "Ai" and "Aj" in the search field combination "Ai+Aj", and calculate the frequencies. n (Ai∩Aj) represents the number of times that the search fields Ai and Aj are searched simultaneously; Calculate the correlation coefficient G(i,j) between the search fields Ai and Aj = P(Ai∩Aj) / [P(Ai)+P(Aj)].
[0024] It is worth noting that the correlation strength is determined by quantifying the collaborative retrieval characteristics of the two fields. First, the definition of the retrieval field combination "Ai+Aj" is clarified, where Ai represents the i-th retrieval field and Aj represents the j-th retrieval field (i≠j, ensuring field uniqueness). The "+" specifically refers to the composite retrieval behavior of the user simultaneously inputting both Ai and Aj fields in a single retrieval operation, such as the user simultaneously inputting combined queries like "machine tool assembly and adjustment + advanced" or "pharmaceutical purchase and sales + basic". Next, key parameters are obtained and calculated. First, based on the preset retrieval period T, the total occurrence count ni of Ai and the total occurrence count nj of Aj within the period are counted. Then, ni and nj are divided by the total occurrence count N of all retrieval fields within the period, respectively, to obtain the retrieval frequency of a single field P(Ai) = ni / N and P(Aj) = nj / N (for simplification, let...). This frequency value directly reflects the search popularity of a single field; The core parameter n(Ai∩Aj) specifically refers to the total number of times a user simultaneously inputs Ai and Aj for a search within a search period T. The scope must be strictly defined during statistics, including only search records where both fields are entered simultaneously, completely excluding searches for Ai or Aj alone. For example, if there are 80 combined searches for "machine tool assembly and adjustment + advanced" within the period, then n(machine tool assembly and adjustment ∩ advanced) = 80; if there are 50 combined searches for "pharmaceutical purchase and sales + primary", then n(pharmaceutical purchase and sales ∩ primary) = 50, ensuring that the data accurately maps to the collaborative search needs of the two fields. Then, n(Ai∩Aj) is divided by the total number of all search behaviors N within the period to obtain the joint frequency of the two fields P(Ai∩Aj) = n(Ai∩Aj) / N. Finally, the correlation coefficient was calculated using the formula G(i,j)=P(Ai∩Aj) / [P(Ai)+P(Aj)]. The result is between 0 and 0.5. The closer the value is to 0.5, the higher the probability that the two fields are retrieved at the same time and the stronger the correlation. The closer the value is to 0, the weaker the correlation. This provides a quantitative basis for subsequent course recommendations based on correlation.
[0025] In a preferred embodiment, the user inputs the search field Ai and performs a search. The retrieved skill courses are marked as initial courses (Ai). The number of enrollees (R) and the number of favorites (Z) for the initial courses are obtained. The popularity of the initial courses is then calculated as X = λ1 × log MR+λ2×Z, where λ1 and λ2 represent the preset first and second coefficients respectively, and M represents the preset number of people in the gradient. The initial courses are sorted in descending order of popularity X; Aj is obtained in the same way.
[0026] Understandably, ranking courses corresponding to a single search field by quantifying course popularity lays the foundation for subsequent multi-field related searches. First, when a user enters a search field Ai (such as "auto mechanic" or "senior technician"), the system matches and searches the skills learning database based on this field, filtering out all skills courses related to Ai and marking these courses as initial Ai courses, ensuring a basic relevance between the initial courses and the search requirements. Next, the system automatically collects two core dynamic data points for each initial Ai course: first, the number of enrollees R, the total number of students who have successfully enrolled in the course, reflecting the actual participation rate; and second, the number of times the course has been added to users' favorites Z, reflecting user preference for the course. Then, the popularity X of each course is calculated according to a preset formula, where λ1 and λ2 are preset coefficients whose weights can be adjusted according to the platform's training positioning (e.g., increasing λ1 if participation is emphasized, increasing λ2 if user preference is emphasized), and M is the preset gradient number of participants (e.g., 100 people). Introducing the logM function avoids numerical imbalance caused by excessively large enrollment numbers, making the popularity calculation more balanced. After calculation, all initial AI courses are sorted from highest to lowest popularity (X), prioritizing high-popularity courses for users. For the search field Aj, the system repeats the entire process: retrieving and matching initial courses for Aj, collecting the number of enrollees (R) and the number of favorites (Z), calculating and sorting popularity (X), and finally obtaining the sorted initial courses for Aj. This provides independent and accurate single-field course popularity data support for subsequent optimization and sorting of courses with multiple field associations.
[0027] In a preferred embodiment, the first search field Ai in the search field combination "Ai+Aj" is designated as the primary field, and the remaining search fields are designated as secondary fields. The popularity X of the initial course corresponding to the secondary fields is then modified as follows: Based on the current original value of popularity X, multiply it with the correlation coefficient G(i,j), and the result is the updated value of popularity X. Reorder the initial courses in the search fields Ai and Aj based on the current popularity X, sort them from highest to lowest popularity X, select the top K initial courses as popular courses, store the popular courses in the cache, K represents the preset number of courses, and record the search field combination "Ai+Aj" and the K popular courses as a linkage relationship. Repeat the above operation to match the correlation relationships of all search field combinations within the current search period T.
[0028] It's important to note that by differentiating between primary and secondary fields, adjusting popularity, and establishing correlations, the accuracy and efficiency of search results can be improved. First, for the search field combination "Ai+Aj", a clear rule for dividing primary and secondary fields is established: the first search field Ai is defined as the primary field, and the remaining field Aj is designated as the secondary field. The primary field guides the core search direction, while the secondary fields supplement the filtering dimensions, ensuring that the search logic aligns with the user's priority needs. Next, the popularity of the initial courses corresponding to the secondary field Aj is adjusted. The adjustment logic is based on the original popularity X of the initial courses Aj, multiplied by the previously calculated correlation coefficient G(i,j) between Ai and Aj to obtain the updated popularity X value. A higher correlation coefficient G(i,j) indicates a stronger correlation between Ai and Aj, resulting in a higher adjusted popularity for the initial courses Aj. This binds course popularity with field correlation, making the sorting more aligned with complex search needs.
[0029] Subsequently, the system integrates the initial courses (Ai) with the revised initial courses (Aj) based on their original popularity (X), and re-sorts them uniformly according to their popularity (X) from highest to lowest. The top K courses are then selected and marked as the most popular courses corresponding to this field combination. To improve subsequent search efficiency, the system caches these popular courses and establishes a link between the search field combination "Ai+Aj" and these K popular courses, forming a unique association mapping. Finally, for all possible two-field search combinations within the current search period T, the entire process of primary and secondary field division, secondary field popularity correction, course re-sorting, popular course filtering, cache storage, and link relationship establishment is repeated. This achieves comprehensive matching of all two-field combinations with their corresponding popular courses within the period, providing data support for rapid response when users subsequently input the same field combination, balancing search accuracy and efficiency.
[0030] In a preferred embodiment, when a user enters the search field combination "Ai+Aj", popular courses with related relationships are displayed first. When the current search period T ends, the next search period T begins immediately, and the search field combinations and corresponding relationships are updated continuously.
[0031] It's important to note that a cache reuse and rolling update mechanism balances search efficiency with result timeliness. When a user inputs the search field combination "Ai+Aj", the system first checks the cache for a corresponding relationship (i.e., the previously bound top K popular courses). If it exists, the system directly retrieves and displays the popular courses from the cache, skipping redundant steps such as field association calculations and course sorting, significantly reducing response latency and improving the user's search experience. If this is the first time the system has searched for this combination, it follows a preset process to divide the primary and secondary fields, correct the popularity of the secondary fields, and filter the course sorting, generating popular courses and establishing a relationship to store in the cache, providing a basis for reuse in subsequent similar searches.
[0032] For cycle management, a seamless rolling update mode is adopted. When the current search cycle T ends, the system immediately starts the next search cycle T of the same length, ensuring uninterrupted data updates. Within the new cycle, the system re-collects all search behavior data, updates the frequency of field occurrences and the correlation coefficients between fields, and then recalculates the popular courses corresponding to each combination, overwriting the linkage relationships cached in the old cycle. This approach avoids data lag in a single cycle and ensures sample validity through periodic statistics, ensuring that linkage relationships always align with the latest search preferences and course popularity, thus guaranteeing the continuous accuracy of composite search results.
[0033] When a user enters only a single search field Ai, the system first matches related courses in the database based on the Ai field, marking them as initial Ai courses to ensure a basic relevance between the courses and the search request. Then, the previously calculated popularity X of the initial Ai courses is reused, eliminating the need for repetitive calculations and improving response efficiency. Next, courses are sorted from highest to lowest popularity X; if popularity scores are equal, they are sorted from newest to oldest publication date, prioritizing popular and timely courses. Simultaneously, the system records the number of times Ai is searched, incorporating it into the field frequency statistics for the current period T. This accumulates foundational data for subsequent correlation analysis of potential field combinations (such as Ai+Aj), achieving a logical connection between single-field and multi-field searches.
[0034] In a preferred embodiment, when the number of search fields in the search field combination is greater than 2, the following steps are performed: Obtain the search field combination "Ai+Aj+...+Am" and divide it into search field combinations "Ai+Aj", ..., and search field combination "Ai+Am". Calculate the correlation coefficients corresponding to the divided search field combinations. Repeat the above steps to filter K popular courses and match the correlation between the search field combination "Ai+Aj+...+Am" and the K popular courses.
[0035] Understandably, this approach is designed for complex combinations with more than two search fields. First, the first search field Ai in the combination is used as the base primary field. All other auxiliary search fields are paired with Ai to form several binary search field combinations, namely "Ai+Aj", "Ai+Ak", ... "Ai+Am". By decomposing complex combinations into basic binary combinations, the complexity of association calculations is reduced, while ensuring that the core search requirements of the primary field Ai remain unchanged.
[0036] Subsequently, the system calculates the correlation coefficient for each binary combination: for the "Ai+Aj" combination, it calls the previously statistical field frequency data and the number of joint searches, and calculates the correlation coefficient using the formula G(i,j)=P(Ai∩Aj) / [P(Ai)+P(Aj)]; similarly, it completes the correlation coefficient calculation for all binary combinations "Ai+Ak"..."Ai+Am" in turn, forming multiple sets of quantitative data on correlation strength.
[0037] Based on the correlation coefficients of each binary combination, the system repeats the course processing flow for single fields and binary combinations: First, obtain the initial courses corresponding to Ai, Aj, Ak...Am respectively. Multiply the popularity X of the initial courses corresponding to Aj, Ak...Am with the correlation coefficients (G(i,j), G(i,k)...G(i,m)) of the corresponding binary combinations to complete the popularity correction. Then, integrate all the corrected initial courses of auxiliary fields and the initial courses of the Ai main field, sort them uniformly from high to low according to the popularity X, and select the top K courses (K is the preset number of courses) with the highest overall popularity, and mark them as the popular courses corresponding to the multi-field combination "Ai+Aj+...+Am".
[0038] Finally, the system caches these popular courses and establishes a link between the search field combination "Ai+Aj+...+Am" and these K popular courses. When users subsequently input this multi-field combination, the cached data can be reused for a quick response. This solution, through the logic of "fixed main field + binary decomposition," ensures the accuracy of multi-field correlation analysis while reusing existing processing procedures to reduce system complexity, adapting to complex search scenarios such as multiple skill intersections and multiple level superpositions.
[0039] In another preferred embodiment of the present invention, if the total number of all search fields within the current search period T is less than a preset search threshold, then the subsequent steps are stopped.
[0040] It is worth noting that when the total number of search fields within a search period T is lower than a preset threshold, it indicates insufficient sample size. Performing field frequency statistics, correlation analysis, and popular course filtering at this point is prone to randomness and bias, failing to accurately reflect user needs. Stopping subsequent steps avoids generating low-quality correlations based on invalid data, reduces unnecessary computational overhead, ensures the search mechanism always operates based on a sufficient number of valid samples, and guarantees the accuracy and reliability of search results.
[0041] In a preferred embodiment, when initial courses with equal popularity (X) are available, the initial course with a shorter publication time is ranked higher.
[0042] It's worth noting that when multiple initial courses have the same popularity (X), the publication date is used as a supplementary sorting criterion, prioritizing courses that are newer and more aligned with current technological trends and training needs. This resolves sorting disputes when popularity is equal and also pushes timely, high-quality courses, aligning with the rapidly evolving nature of knowledge in the vocational skills field. This better meets users' needs for learning cutting-edge skills, enhancing the search experience and the practicality of training.
[0043] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.
Claims
1. A database-based method for rapid retrieval of skills learning data, characterized in that: Includes the following steps: Set a preset search period T, retrieve all search fields within the current search period T, count the number of times the same search field appears n, and calculate the frequency of occurrence. I represents the number of search fields within the search period T, and n i This represents the number of times the i-th search field appears.
2. The database-based method for rapid retrieval of skills learning data according to claim 1, characterized in that, Calculating the association coefficient G between different search fields involves the following steps: Get the search field combination "Ai+Aj", where Ai represents the i-th search field, Aj represents the j-th search field, i≠j, and "+" means inputting both simultaneously; Obtain the frequencies P(Ai) and P(Aj) corresponding to "Ai" and "Aj" in the search field combination "Ai+Aj", and calculate the frequencies. n (Ai∩Aj) represents the number of times that the search fields Ai and Aj are searched simultaneously; Calculate the correlation coefficient G(i,j) between the search fields Ai and Aj = P(Ai∩Aj) / [P(Ai)+P(Aj)].
3. The database-based method for rapid retrieval of skills learning data according to claim 2, characterized in that, Input the search field Ai and perform the search. Mark the retrieved skill courses as Ai initial courses, obtain the number of enrollees R and the number of favorites Z for the initial courses, and calculate the popularity of the initial courses X=λ1×log M R+λ2×Z, where λ1 and λ2 represent the preset first and second coefficients respectively, and M represents the preset number of people in the gradient. The initial courses are sorted in descending order of popularity X; Aj is obtained in the same way.
4. The database-based method for rapid retrieval of skills learning data according to claim 3, characterized in that, In the search field combination "Ai+Aj", the first search field Ai is designated as the primary field, and the remaining search fields are designated as secondary fields. The secondary fields corresponding to the initial course popularity X are modified as follows: Based on the current original value of popularity X, multiply it with the correlation coefficient G(i,j), and the result is the updated value of popularity X. Reorder the initial courses in the search fields Ai and Aj based on the current popularity X, sort them from largest to smallest popularity X, select the top K initial courses as popular courses, store the popular courses in the cache, K represents the preset number of courses, and record the search field combination "Ai+Aj" with the K popular courses as a linkage relationship; Repeat the above operation to match the correlation relationships of all search field combinations within the current search period T.
5. The database-based method for rapid retrieval of skills learning data according to claim 4, characterized in that, When a user enters the search field combination "Ai+Aj", popular courses with related information will be displayed first. When the current search period T ends, the next search period T begins immediately, and the search field combinations and corresponding relationships are updated continuously.
6. The database-based method for rapid retrieval of skills learning data according to claim 4, characterized in that, When the number of search fields in a search field combination is greater than 2, perform the following steps: Obtain the search field combination "Ai+Aj+...+Am" and divide it into search field combinations "Ai+Aj", ..., and search field combination "Ai+Am". Calculate the correlation coefficients corresponding to the divided search field combinations. Repeat the above steps to filter K popular courses and match the correlation between the search field combination "Ai+Aj+...+Am" and the K popular courses.
7. The database-based method for rapid retrieval of skills learning data according to claim 1, characterized in that, If the total number of all search fields within the current search period T is less than the preset search threshold, then the subsequent steps will be stopped.
8. The database-based method for rapid retrieval of skills learning data according to claim 4, characterized in that, When initial courses with the same popularity (X) are presented, the initial course with the shorter publication date is ranked higher.