Multi-attribute tag and score data asset recommendation method

Abstract

The invention relates to a multi-attribute tag and score data asset recommendation method, and belongs to the technical field of data asset recommendation. According to the method, the current situation that concerned data asset information is passively obtained through single keyword retrieval is changed, multi-dimensional attribute decomposition is carried out according to the characteristics of data assets, the recommendation process is designed through the relevance between the data assets, the relevance between the users and the relevance between the users, and the recommendation efficiency is improved. The relatively accurate data asset recommendation content is formed, the interested data assets are actively recommended to the user, and the accuracy of the recommendation result is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data asset recommendation, and specifically relates to a data asset recommendation method with multiple attribute labels and scores. Background Art

[0002] Data assets are formed through the process of collection, governance, and integration. Data assets provide important support for business applications and in-depth data analysis. As data is collected and updated, the categories and quantities of data assets are constantly changing. Usually, data asset lists and keyword searches can be used to find data asset information. This is a passive way to obtain data asset information, and when the number of data assets is large or the data assets are unfamiliar, the time cost of obtaining the required information is high. Actively recommending data asset information is a better way to meet the relevant needs of users.

[0003] The recommendation algorithm uses some user behaviors and some mathematical algorithms to infer the content that users may like. Existing recommendation algorithms include content-based recommendation, user-based collaborative filtering recommendation, project-based collaborative filtering recommendation, model-based collaborative filtering recommendation, etc. These algorithms have their own advantages and disadvantages, and in a certain scenario, a single recommendation algorithm is often used, which sometimes makes it difficult to meet user needs and obtain the information of interest. Moreover, data assets are different from commodities, movies, etc. They are relatively complex and have diverse attributes. Recommending them requires comprehensive analysis and trade-offs from many aspects. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] The technical problem to be solved by the present invention is to design a data asset recommendation method to form relatively accurate data asset recommendation content to recommend to users, thereby improving the accuracy of the recommendation results.

[0006] (II) Technical solution

[0007] In order to solve the above technical problems, the present invention provides a data asset recommendation method with multi-attribute labels and scores. In the method, first, multi-attribute labels for data assets are constructed, and the user's focus preferences are obtained at the same time; then, the association between data assets and data assets, and the association between users and similar users are analyzed; finally, the association between users and data assets, users and users, and data assets and data assets is used to construct a recommendation algorithm and return the recommendation results.

[0008] The invention also provides a system for implementing the method.

[0009] (III) Beneficial effects

[0010] The present invention proposes a data asset recommendation method with multi-attribute labels and scores, which changes the current situation of passively obtaining the data asset information of interest through single keyword retrieval, performs multi-dimensional attribute decomposition according to the characteristics of data assets, and uses the correlation between data assets and data assets, users and data assets, and users and users to design a recommendation process, form more accurate data asset recommendation content, actively recommend data assets of interest to users, and improve the accuracy of recommendation results. In the recommendation process, three elements are involved: users, other related users, and data assets. The core is to obtain the user's focus and establish an association relationship between the focus or multiple focuses and data assets. To this end, it is necessary to analyze the characteristics and attributes of data assets, establish relevant labels or perform scoring, so as to match the labels or scores of data assets with the user's focus, thereby forming a recommendation result. DETAILED DESCRIPTION

[0011] In order to make the purpose, content and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below in conjunction with embodiments.

[0012] The present invention proposes a data asset recommendation method with multi-attribute labels and scores. According to the characteristics of data assets, multi-dimensional attribute decomposition is performed, and the correlation between data assets and data assets, users and data assets, and users and users is used to design the recommendation process, so as to form more accurate data asset recommendation content and improve the accuracy of recommendation results.

[0013] The present invention changes the current situation of passively obtaining the data asset information of interest through single keyword retrieval, and actively recommends the data assets of interest to users. In the recommendation process, three elements are involved: users, other related users, and data assets. The core is to obtain the user's focus and establish an association between the focus or multiple focuses and data assets. To this end, it is necessary to analyze the characteristics and attributes of the data assets, establish relevant tags or perform scoring, so as to match the tags or scores of the data assets with the user's focus, thereby forming a recommendation result. The overall steps are:

[0014] First, we build multi-attribute labels for data assets and obtain users’ focus preferences.

[0015] Then, analyze the association between data assets and data assets, and the association between users and similar users;

[0016] Finally, the recommendation algorithm is constructed by utilizing the association relationships between users and data assets, users and users, and data assets and data assets, and the recommendation results are returned.

[0017] The specific steps are as follows:

[0018] Step 1: Build multi-attribute labels for data assets

[0019] Establish a data asset labeling system from multiple dimensions and design a labeling method, which specifically includes the following steps:

[0020] Step 11: Divide the attributes of data assets into multiple types, each of which includes multiple sub-attributes; attribute types include time, data type, data size, data content, data value, etc.;

[0021] Step 12: Divide the time by type into generation time, update time, and update frequency, and perform the following steps:

[0022] Step 121: Generate a label and a corresponding score based on the generation time; the earlier the generation time, the higher the score; data assets generated more than one year ago are scored 0 points, and the score increases by 0.1 points for each month of delay. If the score is in the range of [0,0.5], the label is "Generation time: early", and if the score is in the range of [0.5,1], the label is "Generation time: late";

[0023] Step 122: Generate a label and a corresponding score according to the last update time. The following two aspects are considered when generating labels and corresponding scores in this step: First, the last update time. The later the time, the higher the score. No update is 0 points, and the update in the current week is 1 point. Every week earlier, the score decreases by 0.1 points until it reaches 0 points, or there is no update. Second, the amount of data updated last. The proportion of the updated data volume to the total data volume is counted. If the proportion is in the range of [0.9, 1], 1 point is given, and if it is in the range of [0.8, 0.9], 0.8 points are given, and so on. No update is 0 points. The scores of the two aspects are weighted and summed with equal weights. If the score is in the range of [0.8, 1], the label is "update time: new". If the score is in the range of [0.6, 0.8], the label is "update time: relatively new". If the score is in the range of [0.6, 0.4], the label is "update time: relatively early". If the score is lower than 0.4, the label is "update time: early".

[0024] Step 123: Generate labels and corresponding scores based on update frequency; this step considers the following three aspects when generating labels and corresponding scores: First, the total update frequency. If the frequency is greater than 10 times, the score is 1; if the frequency is greater than 9 times, the score is 0.9, and so on; second, the average frequency. The average frequency is calculated based on the total update frequency, the last update time, and the generation time interval. If the average frequency is less than 3 days, the score is 1; if the average frequency is less than a week, the score is 0.8; if the average frequency is less than ten days, the score is 0.6; if the average frequency is less than a month, the score is 0.4; if the average frequency is less than two months, the score is 0.2; if it is higher than a month, the score is 0; third, the average update data volume, which is divided by the sum of the data volume of each update Take the total number of updates and calculate the ratio of the average updated data volume to the total data volume. If the ratio is in the range of [0.9,1], it will be scored 1 point, if it is in the range of [0.8,0.9], it will be scored 0.8 points, and so on. No update will be scored 0 points. The two scores are weighted and summed with equal weights. The scores of the three aspects are weighted and summed with equal weights. If the score is in the range of [0.8,1], the label is "frequent updates", if the score is in the range of [0.6,0.8], the label is "relatively frequent updates", if the score is in the range of [0.6,0.4], the label is "average updates", if the score is in the range of [0.4,0.2], the label is "infrequent updates", and if the score is lower than 0.2, the label is "basically no updates";

[0025] Step 13: Generate labels and corresponding scores based on data size, data type, and data size changes:

[0026] Step 131: Generate labels and corresponding scores according to the data scale; the data scale is determined by the total number of records and files. If the data scale is greater than 10 million, or between 10 million and 1 million, the score is between [1, 0.8], decreasing in proportion to the data volume; if the data scale is between 1 million and 100,000, the score is between [0.8, 0.2], decreasing in proportion to the data volume; if the data scale is between 100,000 and 10,000, the score is between [0, 2, 0], decreasing in proportion to the data volume; if the data scale is less than 10,000, the score is 0; if the score is in the range of [0.8, 1], the label is "many", if the score is in the range of [0.6, 0.8], the label is "more", if the score is in the range of [0.4, 0.6], the label is "general", if the score is in the range of [0.2, 0.4], the label is "few", and if the score is in the range of [0, 0.2], the label is "very few";

[0027] Step 132: Generate labels and corresponding scores based on data types; divide data types into structured, unstructured, and semi-structured data; based on the proportion of the three types of data to the total data volume, if the proportion of a certain type of data is >0.6, then the label is "a certain type of data accounts for a high proportion"; if the proportion of a certain type of data is between [0.3, 0.6] and the proportion of two types of data exceeds 0.6, then the label is "two types of data account for a high proportion".

[0028] Step 133: Generate labels and corresponding scores according to the change in data scale; count the current data volume and the data volume when it was generated; if there is no change, the label is "the total amount remains unchanged"; if the data volume increases and the ratio to the total data volume is greater than 0.2, 0.4, 0.6, and 0.8 respectively, the corresponding labels are "the overall growth rate is small", "the overall growth rate is general", "the overall growth rate is large", and "the overall growth rate is very large"; if the data volume decreases and the ratio to the total data volume is greater than 0.2, 0.4, 0.6, and 0.8 respectively, the corresponding labels are "the overall reduction rate is small", "the overall reduction rate is general", "the overall reduction rate is large", and "the overall reduction rate is very large";

[0029] Step 14: Generate labels and corresponding scores based on the data content; classify data types by business or a certain dimension, such as operation data, sales data, production data, customer data, employee data, market data, etc.; count the proportion of a certain type of data to the total data volume. If the proportion of a certain type of data is > 0.6, the label is "a certain type of data accounts for a high proportion"; if the proportion of a certain type of data is in [0.3, 0.6] and the proportion of two types of data exceeds 0.6, the label is "a certain two types of data account for a high proportion"; if the proportion of a certain type of data is in [0.2, 0.3] and the proportion of three types of data exceeds 0.6, the label is "a certain three types of data account for a high proportion";

[0030] Step 15: Generate labels and corresponding scores based on data value; if the value of the assessed data asset is greater than 10 million, or between 10 million and 1 million, the score is between [1, 0.8], decreasing in proportion to the amount of data; if it is between 1 million and 100,000, the score is between [0.8, 0.2], decreasing in proportion to the amount of data; if it is between 100,000 and 10,000, the score is between [0, 0.2], decreasing in proportion to the amount of data; if it is less than 10,000, the score is 0; if the score is between [0.8, 1], the label is "high value", if it is between [0.6, 0.8], the label is "high value", if it is between [0.4, 0.6], the label is "average value", if it is between [0.2, 0.4], the label is "low value", if it is between [0, 0, 2], the label is "very low value";

[0031] Step 2: Recommend based on current user interests

[0032] This step is based on step 1, which explores the user's focus, associates the focus with the data asset tag system, and forms a recommendation result.

[0033] Step 21: Create a user profile based on the user's interests and preferences, including the following steps:

[0034] Step 211: Obtain the user's preferences for the labels of the above data assets by means of a questionnaire;

[0035] Step 212: According to the user's search history, the data asset type, content, search times, browsing time and other factors are counted, and sorted from high to low by frequency to obtain the user's preference priority;

[0036] Step 22: Cross-check the information in the questionnaire with the information obtained from the search history, extract the common content and obtain the corresponding data assets;

[0037] Step 23: Sort the data assets obtained in step 22 according to the scores;

[0038] Step 24: Sort the data assets according to storage time, update time, and value to obtain the corresponding data asset order;

[0039] Step 25: For the four categories of data assets obtained in steps 23 and 24 (four asset sets obtained in different sorting methods), weighted sums are performed according to weights W1=0.3, W2=0.2, W3=0.2, and W3=0.3, respectively, and sorted from large to small according to scores;

[0040] Step 26: Recommend the top K data assets in the ranking obtained in step 25 to the user.

[0041] Step 3: Recommendations based on data asset relevance

[0042] In order to prevent users from being trapped in information cocoons, the relationship between data assets is constructed based on the correlation between data assets, and the recommended content in step 2 is expanded to form a virtuous circle.

[0043] Step 31: Establish associations between data assets:

[0044] Step 311: Establish association based on generation time; obtain the generation time range T10 of the current data asset and the average score S10 of the generation time, and count the generation time ranges T10 of other data assets i And the average score S1 of the generation time i , filter out T1 i ∈T10 and S1 i ∈[0.8S10,1.2S10], and filter out the Top K data assets according to the original scores of these data assets;

[0045] Step 312: Establish association based on update time; obtain the update time range T20 of the current data asset and the average score S20 of the update time, and count the generation time range T2 of other data assetsi and the average score S2 of the update time i , filter out T2 i ∈T20 and S2 i ∈[0.8S20,1.2S20], and filter out the Top K data assets according to the original scores of these data assets;

[0046] Step 313: Establish association according to data type; obtain the average score of the current data asset type S30, and calculate the average scores of other data asset types S3 i , filter out S3 i ∈[0.8S30,1.2S30], and filter out the Top K data assets according to the original scores of these data assets;

[0047] Step 314: Establish association based on data size, obtain the average score S40 of the current data asset size, and calculate the average score S40 of other data asset sizes i , filter out S4 i ∈[0.8S40,1.2S40], and filter out the Top K data assets according to the original scores of these data assets;

[0048] Step 315: Establish association based on data content, obtain the average score S50 of the current data asset content, and calculate the average score S50 of other data asset content i , filter out S5 i ∈[0.8S50,1.2S50], and filter out the Top K data assets according to the original scores of these data assets;

[0049] Step 316: Establish association based on data value, that is, obtain the average score S60 of the current data asset value, and calculate the average score S60 of the data asset value. i , filter out S i ∈[0.8S60,1.2S60], and filter out the Top K data assets according to the original scores of these data assets;

[0050] Step 32: According to the generation time weight w1=0.05, the update time weight w2=0.05, the data type weight w3=0.2, the data scale weight w4=0.2, the data content weight w5=0.3, and the data value weight w6=0.2, the original scores of the data assets screened in steps 311-316 are weighted and summed, and sorted from large to small by score to select the top K data assets;

[0051] Step 33: Recommend the Top K data assets obtained in step 32 to the user;

[0052] Step 4: Recommend based on user similarity

[0053] Consider the association and similarity between users, build the association relationship between users, expand the recommendation results of step 2, and use the similarity between users to recommend data assets.

[0054] Step 41: Group users and establish associations:

[0055] Step 411: Establish association based on generation time; obtain the generation time range T30 of the user's browsing data assets and the average score S70 of the generation time, and count the generation time ranges T30 of other users' browsing data assets. i And the average score S7 of the generation time i , filter out T3 i ∈T30 and S7 i ∈[0.8S70,1.2S70] corresponds to the user set, accumulate the data assets browsed by these users and sort them from least to least, and filter out the Top K data assets;

[0056] Step 412: Establish association based on update time; obtain the update time range T40 of the user's browsing data assets and the average score S80 of the update time, and count the update time range T40 of other users' browsing data assets i And the average score of the update time S8 i , filter out T4 i ∈T40 and S8 i ∈[0.8S80,1.2S80] corresponds to the user set, accumulate the data assets browsed by these users by the number of times and sort them from least to least, and filter out the Top K data assets;

[0057] Step 413: Establish associations based on data types; obtain the average score of the data asset type browsed by the user S90, and count the average scores of the data asset types browsed by other users S90 i , filter out S9 i ∈[0.8S90,1.2S90] corresponds to the user set, accumulate the data assets browsed by these users and sort them from least to least, and filter out the Top K data assets;

[0058] Step 414: Establish association based on data size; obtain the average score S100 of the data asset size browsed by the user, and count the average score S10 of the data asset size browsed by other users i , filter out S10 i∈[0.8S100,1.2S100] corresponds to the user set, accumulate the data assets browsed by these users by the number of times and sort them from least to least, and filter out the Top K data assets;

[0059] Step 415: Establish association based on data content, obtain the average score of users browsing data asset content S110, and count the average score of other users browsing data asset content S11 i , filter out S i ∈[0.8S110,1.2S110] corresponds to the user set, accumulate the data assets browsed by these users by the number of times and sort them from least to least, and filter out the Top K data assets;

[0060] Step 416: Establish association based on data value; obtain the average score of the data asset value browsed by the user S120, and calculate the average score of the data asset value browsed by other users S12 i , filter out S12 i ∈[0.8S120,1.2S120] corresponds to the user set, accumulate the number of data assets browsed by these users and sort them from least to least, and filter out the Top K data assets;

[0061] Step 42: According to the generation time weight w1=0.05, the update time weight w2=0.05, the data type weight w3=0.2, the data scale weight w4=0.2, the data content weight w5=0.3, and the data value weight w6=0.2, the original scores of the data assets screened in steps 411-416 are weighted and summed, and sorted from large to small by score to select the top K data assets;

[0062] Step 43: Recommend the Top K data assets obtained in step 42 to the user.

[0063] Step 5: Comprehensively evaluate multiple recommendation results and make recommendations

[0064] Comprehensively evaluate the multiple recommendation results obtained in step 2, step 3, and step 4 from multiple dimensions to generate the final recommendation result and form a more accurate recommendation result. The steps are as follows:

[0065] Step 51: Get the top K data assets with the most total views, and map the views to the interval [0,10];

[0066] Step 52: Obtain the top K data assets with the highest total scores, and map the scores to the interval [0,10];

[0067] Step 53: Get the top K data assets of step 2, step 3, and step 4, and map the serial numbers to the interval [0,10] according to the sorting, that is, set the value of the top 1 data asset to 10 and the value of the top K data asset to 0;

[0068] Step 54: Perform weighted summation of the five types of data assets obtained in steps 51, 52, and 53 according to equal weights, calculate the score of each type of data asset, and accumulate points for the same data assets in different sets;

[0069] Step 55: Sort the accumulated scores of the data assets from large to small, and select the top K data assets;

[0070] Step 56: Recommend the Top K data assets selected in step 55 to the user.

[0071] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A data asset recommendation method with multiple attribute labels and scores, characterized in that: In this method, first, multi-attribute labels for data assets are constructed, and user focus preferences are obtained at the same time; then, the association between data assets and data assets, and the association between users and similar users are analyzed; finally, the association between users and data assets, users and users, and data assets and data assets is used to construct a recommendation algorithm and return the recommendation results.

2. The method according to claim 1, characterized in that The method specifically comprises the following steps: Step 1: Build multi-attribute labels for data assets This step establishes a data asset label system from multiple dimensions and designs a label determination method; Step 2: Recommend based on current user interests This step is based on step 1, mining the user's focus, associating the focus with the data asset tag system, and forming a recommendation result; Step 3: Recommendations based on data asset relevance This step builds the association relationship between data assets based on the relevance of data assets, and expands the recommended content of step 2; Step 4: Recommend based on user similarity This step considers the association and similarity between users, builds the association relationship between users, expands the recommendation results of step 2, and uses the similarity between users to recommend data assets; Step 5: Comprehensively evaluate multiple recommendation results and make recommendations Comprehensively evaluate the multiple recommendation results obtained in step 2, step 3, and step 4 from multiple dimensions to generate the final recommendation result.

3. The method according to claim 2, characterized in that Step 1 specifically includes the following steps: Step 11: Divide the attributes of data assets into multiple types, each of which includes multiple sub-attributes; attribute types include time, data type, data size, data content, and data value; Step 12: Divide the time by type into generation time, update time, and update frequency, and perform the following steps: Step 121: Generate a label and a corresponding score according to the generation time; The earlier the generation time, the higher the score. Data assets generated more than one year ago will get 0 points. For every month of delay, the score will increase by 0.1 points. If the score is in the range of [0,0.5], the label will be "Generation time: early". If the score is in the range of [0.5,1], the label will be "Generation time: late". Step 122: Generate a label and a corresponding score based on the last update time. The following two aspects are considered when generating labels and corresponding scores in this step: First, the last update time. The later the time, the higher the score. No update is 0 points, and the update in the current week is 1 point. Every week earlier, the score decreases by 0.1 points until it reaches 0 points, or there is no update. Second, the amount of data updated last. The proportion of the updated data volume to the total data volume is counted. If the proportion is in the range of [0.9, 1], 1 point is given. If the proportion is in the range of [0.8, 0.9], 0.8 points are given. And so on. No update is 0 points. The scores of the two aspects are weighted and summed with equal weights. If the score is in the range of [0.8, 1], the label is "update time: new". If the score is in the range of [0.6, 0.8], the label is "update time: relatively new". If the score is in the range of [0.6, 0.4], the label is "update time: relatively early". If the score is lower than 0.4, the label is "update time: early". Step 123: Generate labels and corresponding scores based on update frequency; this step considers the following three aspects when generating labels and corresponding scores: First, the total update frequency. If the frequency is greater than 10 times, the score is 1; if the frequency is greater than 9 times, the score is 0.9, and so on; second, the average frequency. The average frequency is calculated based on the total update frequency, the last update time, and the generation time interval. If the average frequency is less than 3 days, the score is 1; if the average frequency is less than a week, the score is 0.8; if the average frequency is less than ten days, the score is 0.6; if the average frequency is less than a month, the score is 0.4; if the average frequency is less than two months, the score is 0.2; if it is higher than a month, the score is 0; third, the average update data volume, which is divided by the sum of the data volume of each update Take the total number of updates and calculate the ratio of the average updated data volume to the total data volume. If the ratio is in the range of [0.9,1], it will be scored 1 point, if it is in the range of [0.8,0.9], it will be scored 0.8 points, and so on. No update will be scored 0 points. The two scores are weighted and summed with equal weights. The scores of the three aspects are weighted and summed with equal weights. If the score is in the range of [0.8,1], the label is "frequent updates", if the score is in the range of [0.6,0.8], the label is "relatively frequent updates", if the score is in the range of [0.6,0.4], the label is "average updates", if the score is in the range of [0.4,0.2], the label is "infrequent updates", and if the score is lower than 0.2, the label is "basically no updates"; Step 13: Generate labels and corresponding scores based on data size, data type, and data size changes: Step 131: Generate labels and corresponding scores according to the data scale; the data scale is determined by the total number of records and files. If the data scale is greater than 10 million, or between 10 million and 1 million, the score is between [1, 0.8], decreasing in proportion to the data volume; if the data scale is between 1 million and 100,000, the score is between [0.8, 0.2], decreasing in proportion to the data volume; if the data scale is between 100,000 and 10,000, the score is between [0, 2, 0], decreasing in proportion to the data volume; if the data scale is less than 10,000, the score is 0; if the score is in the range of [0.8, 1], the label is "many", if the score is in the range of [0.6, 0.8], the label is "more", if the score is in the range of [0.4, 0.6], the label is "general", if the score is in the range of [0.2, 0.4], the label is "few", and if the score is in the range of [0, 0.2], the label is "very few"; Step 132: Generate labels and corresponding scores according to data types; divide data types into structured, unstructured, and semi-structured data; according to the proportion of the three types of data to the total data volume, if the proportion of a certain type of data is > 0.6, then the label is "a certain type of data accounts for a high proportion"; if the proportion of a certain type of data is between [0.3, 0.6] and the proportion of two types of data exceeds 0.6, then the label is "two types of data account for a high proportion"; Step 133: Generate labels and corresponding scores according to the change in data scale; count the current data volume and the data volume when it was generated; if there is no change, the label is "total unchanged"; if the data volume increases and the ratio to the total data volume is greater than 0.2, 0.4, 0.6, and 0.8 respectively, the corresponding labels are "overall small growth", "overall normal growth", "overall large growth", and "overall large growth"; if the data volume decreases and the ratio to the total data volume is greater than 0.2, 0.4, 0.6, and 0.8 respectively, the corresponding labels are "overall small reduction", "overall normal reduction", "overall large reduction", and "overall large reduction" respectively; Step 14: Generate labels and corresponding scores based on the data content; classify data types by business or a certain dimension, such as operation data, sales data, production data, customer data, employee data, market data, etc.; count the proportion of a certain type of data to the total data volume. If the proportion of a certain type of data is > 0.6, the label is "a certain type of data accounts for a high proportion"; if the proportion of a certain type of data is in [0.3, 0.6] and the proportion of two types of data exceeds 0.6, the label is "a certain two types of data account for a high proportion"; if the proportion of a certain type of data is in [0.2, 0.3] and the proportion of three types of data exceeds 0.6, the label is "a certain three types of data account for a high proportion"; Step 15: Generate labels and corresponding scores based on data value; if the value of the assessed data asset is greater than 10 million, or between 10 million and 1 million, the score is between [1,0.8], decreasing in proportion to the amount of data; if the value of the assessed data asset is between 1 million and 100,000, the score is between [0.8,0.2], decreasing in proportion to the amount of data; if the value of the assessed data asset is between 100,000 and 10,000, the score is between [0,0.2], decreasing in proportion to the amount of data; if the value of the assessed data asset is less than 10,000, the score is 0; if the score is between [0.8,1], the label is "high value", if it is between [0.6,0.8], the label is "high value", if it is between [0.4,0.6], the label is "average value", if it is between [0.2,0.4], the label is "low value", and if it is between [0,0,2], the label is "very low value".

4. The method according to claim 3, characterized in that Step 2 specifically includes the following steps: Step 21: Create a user profile based on the user's interests and preferences, including the following steps: Step 211: Obtain the user's preferences for the labels of the above data assets by means of a questionnaire; Step 212: According to the user's search history, the data asset type, content, search times, browsing time and other factors are counted, and sorted from high to low by frequency to obtain the user's preference priority; Step 22: Cross-check the information in the questionnaire with the information obtained from the search history, extract the common content and obtain the corresponding data assets; Step 23: Sort the data assets obtained in step 22 according to the scores; Step 24: Sort the data assets according to storage time, update time, and value to obtain the corresponding data asset order; Step 25: For the four types of data assets obtained in step 23 and step 24, weighted sum is performed according to weights W1=0.3, W2=0.2, W3=0.2, and W3=0.3 respectively, and sorted from large to small according to the scores; Step 26: Recommend the top K data assets in the ranking obtained in step 25 to the user.

5. The method according to claim 4, characterized in that Step 3 specifically includes: Step 31: Establish associations between data assets: Step 311: Establish association based on generation time; obtain the generation time range T10 of the current data asset and the average score S10 of the generation time, and count the generation time ranges T10 of other data assets i And the average score S1 of the generation time i , filter out T1 i ∈T10 and S1 i ∈[0.8S10,1.2S10], and filter out the Top K data assets according to the original scores of these data assets; Step 312: Establish association based on update time; obtain the update time range T20 of the current data asset and the average score S20 of the update time, and count the generation time range T2 of other data assets i and the average score S2 of the update time i , filter out T2 i ∈T20 and S2 i ∈[0.8S20,1.2S20], and filter out the Top K data assets according to the original scores of these data assets; Step 313: Establish association according to data type; obtain the average score of the current data asset type S30, and calculate the average scores of other data asset types S3 i , filter out S3 i ∈[0.8S30,1.2S30], and filter out the Top K data assets according to the original scores of these data assets; Step 314: Establish association based on data size, obtain the average score S40 of the current data asset size, and calculate the average score S40 of other data asset sizes i , filter out S4 i ∈[0.8S40,1.2S40], and filter out the Top K data assets according to the original scores of these data assets; Step 315: Establish association based on data content, obtain the average score S50 of the current data asset content, and calculate the average score S50 of other data asset content i , filter out S5 i ∈[0.8S50,1.2S50], and filter out the Top K data assets according to the original scores of these data assets; Step 316: Establish association based on data value, that is, obtain the average score S60 of the current data asset value, and calculate the average score S60 of the data asset value. i , filter out S i ∈[0.8S60,1.2S60], and filter out the Top K data assets according to the original scores of these data assets; Step 32: According to the generation time weight w1=0.05, the update time weight w2=0.05, the data type weight w3=0.2, the data scale weight w4=0.2, the data content weight w5=0.3, and the data value weight w6=0.2, the original scores of the data assets screened in steps 311-316 are weighted and summed, and sorted from large to small by score to select the top K data assets; Step 33: Recommend the Top K data assets obtained in step 32 to the user.

6. The method according to claim 5, characterized in that Step 4 specifically includes: Step 41: Group users and establish associations: Step 411: Establish association based on generation time; obtain the generation time range T30 of the user's browsing data assets and the average score S70 of the generation time, and count the generation time ranges T30 of other users' browsing data assets. i And the average score S7 of the generation time i , filter out T3 i ∈T30 and S7 i ∈[0.8S70,1.2S70] corresponds to the user set, accumulate the data assets browsed by these users and sort them from least to least, and filter out the Top K data assets; Step 412: Establish association based on update time; obtain the update time range T40 of the user's browsing data assets and the average score S80 of the update time, and count the update time range T40 of other users' browsing data assets i And the average score of the update time S8 i , filter out T4 i ∈T40 and S8 i ∈[0.8S80,1.2S80] corresponds to the user set, accumulate the data assets browsed by these users by the number of times and sort them from least to least, and filter out the Top K data assets; Step 413: Establish associations based on data types; obtain the average score of the data asset type browsed by the user S90, and count the average scores of the data asset types browsed by other users S90 i , filter out S9 i ∈[0.8S90,1.2S90] corresponds to the user set, accumulate the data assets browsed by these users and sort them from least to least, and filter out the Top K data assets; Step 414: Establish association based on data size; obtain the average score S100 of the data asset size browsed by the user, and count the average score S10 of the data asset size browsed by other users i , filter out S10 i ∈[0.8S100,1.2S100] corresponds to the user set, accumulate the data assets browsed by these users by the number of times and sort them from least to least, and filter out the Top K data assets; Step 415: Establish association based on data content, obtain the average score of users browsing data asset content S110, and count the average score of other users browsing data asset content S11 i , filter out S i ∈[0.8S110,1.2S110] corresponds to the user set, accumulate the data assets browsed by these users by the number of times and sort them from least to least, and filter out the Top K data assets; Step 416: Establish association based on data value; obtain the average score of the data asset value browsed by the user S120, and calculate the average score of the data asset value browsed by other users S12 i , filter out S12 i ∈[0.8S120,1.2S120] corresponds to the user set, accumulate the number of data assets browsed by these users and sort them from least to least, and filter out the Top K data assets; Step 42: According to the generation time weight w1=0.05, the update time weight w2=0.05, the data type weight w3=0.2, the data scale weight w4=0.2, the data content weight w5=0.3, and the data value weight w6=0.2, the original scores of the data assets screened in steps 411-416 are weighted and summed, and sorted from large to small by score to select the top K data assets; Step 43: Recommend the Top K data assets obtained in step 42 to the user.

7. The method according to claim 6, characterized in that Step 5 specifically includes: Step 51: Get the top K data assets with the most total views, and map the views to the interval [0,10]; Step 52: Obtain the top K data assets with the highest total scores and map the scores to the interval [0,10]; Step 53: Get the top K data assets of step 2, step 3, and step 4, and map the serial numbers to the interval [0,10] according to the sorting, that is, set the value of the top 1 data asset to 10 and the value of the top K data asset to 0; Step 54: Perform weighted summation of the five types of data assets obtained in steps 51, 52, and 53 according to equal weights, calculate the score of each type of data asset, and accumulate points for the same data assets in different sets; Step 55: Sort the accumulated points of the data assets from large to small, and select the top K data assets; Step 56: Recommend the Top K data assets selected in step 55 to the user.

8. The method according to any one of claims 1 to 7, characterized in that This method is applied in data processing.

9. A system for implementing the method according to any one of claims 1 to 8.

10. The system according to claim 9, characterized in that The system is used in data processing.