Real estate registration file sorting service method and system

Abstract

The present invention belongs to the technical field of archive organization. The present invention provides a real estate registration archive organization service method and system, including: dividing real estate registration archives into high-frequency query archives and low-frequency query archives, and screening out high-frequency stable query users corresponding to the high-frequency query archives, dividing the high-frequency query archives into high-efficiency archives and low-efficiency archives by calculating the query efficiency of the high-frequency query archives, and dividing the users corresponding to the high-efficiency archives into high-efficiency users and low-efficiency users, and judging whether to change the classification label of the archive by analyzing the difference between the query information of the high-efficiency users and the low-efficiency users and the archives and the number of high-efficiency users, identifying keywords that are consistent with both high-efficiency users and low-efficiency users as new classification labels, thereby solving the problem of low efficiency of user queries due to unreasonable label classification of archives in the process of real estate registration archive organization and improving the efficiency of user queries.

Description

Technical Field

[0001] The present invention belongs to the technical field of file organization, and in particular to a real estate registration file organization service method and system. Background Art

[0002] Existing technologies have many problems in grasping the needs of core users and classifying archive tags. For example, there is a lack of effective methods to screen out users who frequently query archives, which makes it difficult to deeply understand the needs of these core users, provide targeted services, and optimize archive management strategies, resulting in inaccurate archive tag classification. In terms of archive tag classification, existing technologies may lack scientific and effective methods to accurately classify archive tags, resulting in a mismatch between archive classification tags and users' actual query needs, which may cause users to spend too much time on queries, have low accuracy, and have low query efficiency.

[0003] To this end, the present invention provides a real estate registration file sorting service method and system. Summary of the Invention

[0004] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0005] The technical solution adopted by the present invention to solve its technical problem is:

[0006] Step 1: Count the number of queries for each real estate registration file in the historical period and the total number of real estate files, calculate the relative query ratio of the real estate registration files, and divide the real estate registration files into high-frequency query files and low-frequency query files;

[0007] Step 2: For high-frequency query files, filter out users with high-frequency and stable queries and establish a corresponding user database;

[0008] Step 3: Calculate the query efficiency of high-frequency query files and compare it with the threshold. Mark the files with query efficiency greater than or equal to the threshold as high-efficiency files. For high-efficiency files, divide the corresponding users in the user database into high-efficiency users and low-efficiency users.

[0009] Step 4: Determine whether the difference between the query information of efficient users and inefficient users and the classification labels of the archives is large. If so, determine whether the classification labels need to be changed based on the ratio of the actual number of efficient users to the ideal number of efficient users.

[0010] Step 5: If the classification label needs to be changed, establish a query information database based on efficient users and inefficient users, extract keywords, and identify keywords that are consistent with both efficient users and inefficient users as new classification labels.

[0011] Furthermore, the method of dividing the real estate registration files into high-frequency query files and low-frequency query files is:

[0012] Comparing the relative query ratio of real estate registration files with a threshold value;

[0013] Classify real estate registration files with a relative query ratio greater than or equal to a threshold as high-frequency query files;

[0014] Real estate registration files with a relative query ratio less than a threshold are classified as low-frequency query files.

[0015] Furthermore, the relative query ratio of the real estate registration files is calculated as follows:

[0016] During the historical period, the relative query ratio of real estate registration files is calculated by calculating the ratio of the number of queries to the total number of real estate registration files.

[0017] Furthermore, the process of screening out users with high-frequency and stable queries includes:

[0018] Divide the historical period into n small time periods. For each user, count the number of queries in each small time period to generate a query count sequence.

[0019] Compare the number of queries per user in a small time period with the set threshold. If the number of queries in a small time period is greater than the threshold, mark the small time period as a high-frequency time period.

[0020] Count the number of high-frequency time periods for each user in the historical period and calculate the ratio of this to the total number of small time periods to obtain the high-frequency time period ratio.

[0021] For each user, the high-frequency time period ratio and query number series standard deviation are compared with the threshold, and users with a high-frequency time period ratio greater than or equal to the threshold and a query time standard deviation greater than or equal to the threshold are screened as high-frequency stable query users.

[0022] Furthermore, the calculation process of the high-frequency query archive query efficiency includes:

[0023] Record the start and end time of each query to the real estate registration file, calculate the time taken for each query, then add up the time taken for all query operations within a certain period of time and divide it by the number of queries to get the average query time;

[0024] The query success rate is calculated by counting the ratio of the number of successful searches for the required files to the total number of searches within a certain period of time.

[0025] A successful query means that the user can accurately obtain the required file information within the specified time;

[0026] When a user enters a keyword or other search condition to conduct a search, the accuracy rate of the search results is obtained by calculating the ratio of the number of files related to the user's actual needs to the total number of search results.

[0027] The average query time, query success rate, and retrieval result accuracy are standardized, and the standardized data are multiplied by their respective weights and then added together to obtain the query efficiency.

[0028] Furthermore, the process of dividing the corresponding users in the user database into efficient users and inefficient users includes:

[0029] For each user with a high-efficiency profile, count the number of high-efficiency profiles and the total number of profiles queried by the user during the historical period, and calculate the proportion of high-efficiency profiles.

[0030] Compare the effective file ratio of each user with the set threshold;

[0031] If the percentage of efficient files is greater than or equal to the threshold, the user is an efficient user;

[0032] If the ratio of efficient files is less than the threshold, the user is an inefficient user.

[0033] Furthermore, the method for determining whether the difference between the query information of the efficient user and the inefficient user and the classification label of the archive is large is as follows:

[0034] Collect query information from efficient and inefficient users, extract keywords, and generate query information keyword set A and classification label keyword set B respectively;

[0035] Calculate the intersection and union of set A and set B respectively, and count the number of elements in the intersection and union;

[0036] The matching scores between the query information of efficient users and inefficient users and the archive classification labels are calculated according to the Jaccard coefficient formula;

[0037] Compare the matching scores between the query information of efficient users and inefficient users and the archive classification labels with a threshold;

[0038] If the matching score is less than or equal to the threshold, it means that the difference between the query information of the efficient user and the inefficient user and the classification label of the archive is large.

[0039] Furthermore, the method of identifying keywords that are both suitable for efficient users and inefficient users is as follows:

[0040] Label the extracted keywords to clarify whether each keyword belongs to an efficient or inefficient user, and organize the extracted keywords and corresponding user types into a data set;

[0041] The Naive Bayes model is used to identify keywords that are suitable for both efficient and inefficient users.

[0042] Furthermore, the process of using the Naive Bayes model to identify keywords that are suitable for both efficient users and inefficient users includes:

[0043] Calculate the prior probability and conditional probability of efficient users and inefficient users respectively;

[0044] Compare the product of the prior probability and conditional probability of efficient users with the product of the prior probability and conditional probability of inefficient users;

[0045] If the product of the prior probability and conditional probability of an efficient user is approximately equal to the product of the prior probability and conditional probability of an inefficient user, it means that the keyword is suitable for both efficient and inefficient users.

[0046] A real estate registration file sorting service system includes the following modules:

[0047] File query frequency classification module: counts the number of queries for each real estate registration file in the historical period and the total number of real estate files, calculates the relative query ratio of real estate registration files, and divides real estate registration files into high-frequency query files and low-frequency query files;

[0048] High-frequency and stable user screening module: For high-frequency query files, filter out users with high-frequency and stable queries and establish a corresponding user database;

[0049] High-efficiency archive and user classification module: Calculates the query efficiency of frequently queried archives, compares it with the threshold, and marks archives with query efficiency greater than or equal to the threshold as high-efficiency archives. For high-efficiency archives, the corresponding users in the user database are divided into high-efficiency users and low-efficiency users.

[0050] Classification label difference evaluation module: determines whether the difference between the query information of efficient users and inefficient users and the classification labels of the archives is large. If so, it determines whether the classification labels need to be changed based on the ratio of the actual number of efficient users to the ideal number of efficient users;

[0051] New classification label generation module: If the classification label needs to be changed, a query information database is established based on efficient users and inefficient users, keywords are extracted, and keywords that are consistent with both efficient users and inefficient users are identified as new classification labels.

[0052] The beneficial effects of the present invention are as follows:

[0053] The real estate registration archives are divided into high-frequency query archives and low-frequency query archives, and high-frequency stable query users corresponding to the high-frequency query archives are screened out. By calculating the query efficiency of the high-frequency query archives, the high-frequency query archives are divided into high-efficiency archives and low-efficiency archives, and the users corresponding to the high-efficiency archives are divided into high-efficiency users and low-efficiency users. By analyzing the differences between the query information of high-efficiency users and low-efficiency users and the archives and the number of high-efficiency users, it is determined whether to change the classification label of the archives, and keywords that are consistent with both high-efficiency users and low-efficiency users are identified as new classification labels. This solves the problem of low efficiency of user queries due to unreasonable label classification of archives in the process of real estate registration archive sorting, and improves the efficiency of user queries. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The present invention will be further described below with reference to the accompanying drawings.

[0055] Figure 1 This is a flowchart of the steps of a real estate registration file arrangement service method according to an embodiment of the present invention;

[0056] Figure 2 It is a flowchart of a real estate registration file arrangement service system described in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0058] Example 1

[0059] See also Figure 1 As shown, a real estate registration file arrangement service method according to an embodiment of the present invention includes the following steps:

[0060] Step 1: Count the number of queries for each real estate registration file in the historical period and the total number of real estate files, calculate the relative query ratio of the real estate registration files, and divide the real estate registration files into high-frequency query files and low-frequency query files;

[0061] In step 1, the relative query ratio refers to the ratio of the number of queries to the real estate registration files to the total number of real estate registration files in the historical period, and the calculation formula is:

[0062]

[0063] In step 1, the process of dividing the real estate registration files into high-frequency query files and low-frequency query files includes:

[0064] Compare the relative query ratio of real estate registration files with a threshold value, classify real estate registration files with a relative query ratio greater than or equal to the threshold value as high-frequency query files, and classify real estate registration files with a relative query ratio less than the threshold value as low-frequency query files;

[0065] The threshold setting of the relative query ratio includes: calculating the relative query ratio of all real estate registration files in the historical period, calculating the average value, and using the average value of the relative query ratio as the threshold;

[0066] For example, assume that file A is searched 30 times in a year. According to the relative query ratio calculation formula, the relative query ratio of file A is: 30 ÷ 1000 = 0.03. File B is searched 10 times in a year, and its relative query ratio is 10 ÷ 1000 = 0.01. File C is searched 50 times in a year, and its relative query ratio is 50 ÷ 1000 = 0.05. And so on. After counting the number of searches for 1000 files, their relative query ratios are calculated respectively.

[0067] Assume that after calculating the relative query ratios of 1,000 files, all relative query ratios are added together and then divided by 1,000, resulting in an average value of 0.02. 0.02 is used as the threshold value of the relative query ratio.

[0068] For file A, its relative query ratio is 0.03, 0.03>0.02, so file A is classified as a high-frequency query file;

[0069] For file B, its relative query ratio is 0.01, 0.01<0.02, so file B is classified as a low-frequency query file;

[0070] For file C, its relative query ratio is 0.05, 0.05>0.02, so file C is classified as a high-frequency query file;

[0071] Step 2: For high-frequency query files, filter out users with high-frequency and stable queries and establish a corresponding user database;

[0072] In step 2, the process of screening out users with high-frequency and stable queries is as follows:

[0073] Divide the historical period into n small time periods. For each user, count the number of queries in each small time period to generate a query count sequence.

[0074] Compare the number of queries per user in a small time period with the set threshold. If the number of queries in a small time period is greater than the threshold, mark the small time period as a high-frequency time period.

[0075] The setting of the query number threshold includes: calculating the average value of the query number sequence and using the average value as the query number threshold;

[0076] Count the number of high-frequency time periods for each user in the historical period and calculate the ratio of this to the total number of small time periods to obtain the high-frequency time period ratio.

[0077] Calculate the standard deviation of each user's query count sequence using the formula: , where n is the number of small time periods, The number of queries for each small time period, is the average number of queries;

[0078] For each user, the high-frequency time period ratio and query time standard deviation are compared with the threshold, and users with a high-frequency time period ratio greater than or equal to the threshold and a query time standard deviation greater than or equal to the threshold are selected as high-frequency stable query users;

[0079] Step 3: Calculate the query efficiency of high-frequency query files and compare it with the threshold. Mark the files with query efficiency greater than or equal to the threshold as high-efficiency files. For high-efficiency files, divide the corresponding users in the user database into high-efficiency users and low-efficiency users.

[0080] In step 3, the calculation process of the high-frequency query archive query efficiency includes:

[0081] Record the start and end time of each query to the real estate registration file, calculate the duration of each query, then add up the duration of all query operations within a certain period of time (such as a month or a quarter), and divide it by the number of queries to get the average query time;

[0082] The query success rate is calculated by counting the ratio of the number of successful searches for the required files to the total number of searches within a certain period of time. A successful search means that the user can accurately obtain the required file information within the specified time.

[0083] When a user enters a keyword or other search condition to conduct a search, the accuracy rate of the search results is obtained by calculating the ratio of the number of files related to the user's actual needs to the total number of search results.

[0084] The query efficiency is calculated based on the average query time, query success rate, and retrieval result accuracy. The process includes:

[0085] The data of average query time, query success rate and retrieval result accuracy are standardized. For average query time, the smaller the value, the better. The formula is used. Normalize to get the normalized value of average query time, where X is the original average query time, Max is the maximum value of the average query time, and Min is the minimum value of the average query time;

[0086] For query success rate and retrieval result accuracy, the larger the value, the better. The formula is used. Perform normalization to obtain the normalized value of query success rate and the normalized value of retrieval result accuracy, where X is the original query success rate or retrieval result accuracy, Max is the maximum value of the query success rate or retrieval result accuracy, and Min is the minimum value of the query success rate or retrieval result accuracy;

[0087] The standardized data is multiplied by their respective weights and then added together to obtain the comprehensive query efficiency. The formula is:

[0088] in, 、 、 are the weights of the standardized value of average query time, the standardized value of query success rate, and the standardized value of retrieval result accuracy;

[0089] Compare the calculated query efficiency of each file with the set threshold. If the query efficiency of the file is higher than the threshold, it is classified as an efficient file. If the query efficiency of the file is lower than the threshold, it is classified as an inefficient file.

[0090] The threshold setting of the archive query efficiency includes: adding the query efficiencies of all archives to obtain the total query efficiency, and dividing the total query efficiency by the number of archives to obtain the average archive query efficiency. The calculation formula is: , taking the average value of query efficiency as the threshold, where is the average value, is the query efficiency of the i-th file, n is the total number of files;

[0091] In step three, the process of dividing the users corresponding to the high-efficiency query profiles into high-efficiency users and low-efficiency users includes:

[0092] For each user with a high-efficiency profile, count the number of high-efficiency profiles N that the user has queried in the historical period. 高效 And the total number of queries N 总 , calculate the ratio of efficient files: , where R i is the ratio of efficient files;

[0093] Compare the percentage of efficient files of each user with the set threshold. If the percentage of efficient files is greater than or equal to the threshold, the user is an efficient user. If the percentage of efficient files is less than the threshold, the user is an inefficient user.

[0094] The threshold setting of the efficient file ratio includes: listing the efficient file ratio distribution of all users, taking the median of the efficient file ratio as the threshold of the efficient ratio;

[0095] For example, assume there are five frequently queried files, numbered A, B, C, D, and E. There are three users, numbered 1, 2, and 3, who have queried these files in the past quarter (historical period). The data parameters of the files include:

[0096] |File|Total number of queries|Number of successful queries|Total time consumed|Number of search results|Number of relevant search results|;

[0097] |A|10 times|8 times|30 minutes|20 items|16 items|;

[0098] |B|8 times|7 times|24 minutes|15 items|12 items|;

[0099] |C|12 times|10 times|48 minutes|25 items|20 items|;

[0100] |D|6 times|5 times|18 minutes|12 items|9 items|;

[0101] |E|9 times|8 times|27 minutes|18 items|15 items|;

[0102] Calculate the average query time:

[0103] File A: Average query time is 30÷10=3 minutes;

[0104] File B: Average query time is 24÷8=3 minutes;

[0105] File C: Average query time is 48÷12=minutes;

[0106] File D: Average query time is 18 ÷ 6 = 3 minutes;

[0107] File E: Average query time is 27÷9=3 minutes;

[0108] Calculate the query success rate:

[0109] File A: The query success rate is 8÷10=0.8;

[0110] File B: The query success rate is 7÷8=0.875;

[0111] File C: The query success rate is 10÷12≈0.833;

[0112] File D: The query success rate is 5÷6≈0.833;

[0113] File E: The query success rate is 8÷9≈0.889;

[0114] Calculate the accuracy of the search results:

[0115] File A: The search result accuracy is 16÷20=0.8;

[0116] File B: The search result accuracy is 12÷15=0.8;

[0117] File C: The search result accuracy is 20÷25=0.8;

[0118] File D: The accuracy of the search results is 9÷12=0.75;

[0119] File E: The accuracy of the search results is 15÷18≈0.833;

[0120] Assume that the maximum value of the average query time is Max=4 minutes, the minimum value is Min=3 minutes, the maximum value of the query success rate is Max=0.889, the minimum value is Min=0.8, the maximum value of the retrieval result accuracy is Max=0.833, the minimum value is Min=0.75, and the weight is set. , , ;

[0121] After calculation, the comprehensive query efficiency of file A, file B, file C, file D, and file E are 0.58, 0.712, 0.291, 0.511, and 1 respectively. The total query efficiency = 0.58 + 0.712 + 0.291 + 0.511 + = 3.094. The average query efficiency is 3.094 ÷ 5 = 0.6188, which is used as the threshold;

[0122] File B (0.712) and File E (1) above the threshold are efficient files, while File A (0.58), File C (0.291), and File D (0.511) are inefficient files;

[0123] Statistics of user queries:

[0124] User 1: Total query times N 总1 = 15 times, of which N is the number of queries to the efficient archives (archives B and E) 高效1 =8 times, the proportion of efficient files ;

[0125] User 2: Total query times N 总2 = 12 times, of which N is the number of queries to the efficient archives (archives B and E) 高效2 =4 times, the proportion of efficient files ;

[0126] User 3: Total query times N总3 = 18 times, of which N is the number of queries for efficient archives (archives B and E) 高效3 =10 times, the percentage of efficient files ;

[0127] Set the high-efficiency ratio threshold and divide users: the high-efficiency profile ratios of all users are 0.533, 0.333, and 0.556, sorted from smallest to largest as 0.333, 0.533, and 0.556, with a median of 0.533, which is used as the high-efficiency ratio threshold.

[0128] User 1's efficiency ratio is equal to the threshold, making it an efficient user. User 2's efficiency ratio is less than the threshold, making it an inefficient user. User 3's efficiency ratio is greater than the threshold, making it an efficient user.

[0129] Step 4: Determine whether the difference between the query information of efficient users and inefficient users and the classification labels of the archives is large. If so, determine whether the classification labels need to be changed based on the ratio of the actual number of efficient users to the ideal number of efficient users.

[0130] In step 4, the process of determining whether the difference between the query information of the efficient user and the inefficient user and the classification label of the archive is large includes:

[0131] Collect query information from efficient and inefficient users, where the query information includes keywords, phrases, or sentences entered by users during querying, and obtain classification labels for archives;

[0132] Using lexical analysis tools in natural language processing technology, we process the query information of efficient and inefficient users and the classification labels of the archives, extract keywords, and generate query information keyword set A and classification label keyword set B respectively;

[0133] Calculate the intersection of set A and set B, that is, the set consisting of elements belonging to both A and B, denoted as , count the number of elements in the intersection, recorded as ;

[0134] Calculate the union of set A and set B, that is, the set consisting of elements belonging to A or B, denoted as , count and collect the number of elements, recorded as ;

[0135] According to the formula of Jaccard similarity coefficient: , calculate the matching scores between the query information of efficient users and inefficient users and the archive classification labels;

[0136] The Jaccard similarity coefficient is an indicator used to measure the similarity between two sets. Its value range is [0,1]. The closer the coefficient is to 1, the higher the similarity between the two sets, and the closer it is to 0, the lower the similarity.

[0137] Compare the matching scores between the query information of efficient and inefficient users and the archive classification labels with the threshold. If the matching scores are less than or equal to the threshold, it means that the difference between the query information of efficient and inefficient users and the archive classification labels is large.

[0138] For example, assume there are three efficient users (User A, User B, User C) and four inefficient users (User D, User E, User F, User G). There are eight real estate registration files in the system, namely File 1, File 2, File 3, File 4, File 5, File 6, File 7, and File 8. The query information and file classification labels are:

[0139] Efficient user query information:

[0140] User A: "Query the latest records of residential property registration in Haidian District"

[0141] User B: "Looking for commercial real estate mortgage registration records in Chaoyang District"

[0142] User C: "Requesting industrial land change registration files processed after 2020"

[0143] Inefficient user query information:

[0144] User D: "I want to check the files related to the house"

[0145] User E: "Looking for files related to real estate registration"

[0146] User F: "Is there any file about mortgage?"

[0147] User G: "Check previous registration files"

[0148] Archive Category Tags:

[0149] File 1: "Haidian District Residential Property Registration - Processing in 2018"

[0150] File 2: "Chaoyang District Commercial Real Estate Mortgage Registration - Processed in 2016"

[0151] File 3: "Xicheng District Industrial Land Change Registration - Processing in 2015"

[0152] File 4: "Fengtai District Residential Property Transfer Registration - Processing in 2019"

[0153] File 5: "Dongcheng District Commercial Real Estate Lease Registration - Processing in 2021"

[0154] File 6: "Tongzhou District Industrial Land Transfer Registration - Processing in 2017"

[0155] File 7: "Daxing District Residential Property Rights Cancellation Registration - Processed in 2014"

[0156] File 8: "Shijingshan District Commercial Real Estate Mortgage Registration - Processing in 2022"

[0157] Perform lexical analysis on the query information of all efficient and inefficient users, extract keywords, and obtain the set A = {Haidian District, residential, property registration, latest, commercial real estate, mortgage registration, Chaoyang District, after 2020, industrial land, change registration, house, real estate registration, mortgage, previous, registration};

[0158] Perform lexical analysis on the classification tags of all archives, extract keywords, and obtain the set B = {Haidian District, residential, property registration, 2018, Chaoyang District, commercial real estate, mortgage registration, 2016, Xicheng District, industrial land, change registration, 2015, Fengtai District, residential property transfer registration, 2019, Dongcheng District, commercial real estate lease registration, 2021, Tongzhou District, industrial land transfer registration, 2017, Daxing District, residential property cancellation registration, 2014, Shijingshan District, commercial real estate mortgage registration, 2022, registration};

[0159] Compute the intersection of set A and set B ,get ={Haidian District, Residential, Property Registration, Commercial Real Estate, Mortgage Registration, Chaoyang District, Industrial Land, Change Registration, Registration}, count the number of elements in the intersection =9, calculate the union of set A and set B , count and collect the number of elements =32;

[0160] According to the Jaccard similarity coefficient formula , calculated ;

[0161] Assuming that the threshold is set to 0.5, the calculated matching score of 0.28 is compared with the threshold of 0.5. Since 0.28≤0.5, it means that there is a large difference between the query information of efficient users and inefficient users and the classification labels of the archives;

[0162] In step 4, the process of determining whether the classification label needs to be changed includes:

[0163] Calculate the ratio of the actual number of efficient users to the ideal number of efficient users to obtain the actual efficient user ratio, and compare the actual efficient user ratio with the threshold;

[0164] If the actual percentage of efficient users is greater than or equal to the threshold, the classification label of the file needs to be changed;

[0165] If the actual percentage of efficient users is less than the threshold, there is no need to change the classification label of the file;

[0166] The ideal number of efficient users can be calculated by analyzing the behavioral data of user query files over a period of time and calculating the average number of efficient users under the existing classification labels.

[0167] Step 5: If the classification label needs to be changed, establish a query information database based on high-efficiency users and low-efficiency users, extract keywords, and identify keywords that match both high-efficiency users and low-efficiency users as new classification labels;

[0168] In step 5, the query information database contains the following data:

[0169] Record the user's identity (such as user number, name, etc.) and user type (efficient user or inefficient user);

[0170] Completely record the original query information such as keywords, phrases or sentences entered by users when querying real estate registration files;

[0171] Record the number, name and other identification information of the real estate registration file involved in the user's query operation;

[0172] In step 5, the method of extracting keywords is:

[0173] Use lexical analysis tools in natural language processing technology to process user query information and extract representative keywords;

[0174] Keywords refer to representative words or phrases that are extracted after processing the query information of efficient and inefficient users and the classification labels of archives and can reflect the main content and core meaning of the query information or classification labels;

[0175] In step 5, the process of identifying keywords that are suitable for both efficient users and inefficient users includes:

[0176] Label the extracted keywords to clarify whether each keyword belongs to an efficient or inefficient user, and organize the extracted keywords and corresponding user types into a data set;

[0177] Using the Naive Bayes model to identify keywords that are suitable for both efficient and inefficient users, the process includes:

[0178] Calculate the prior probability. The prior probability refers to the probability of a certain user type appearing without any query information. The calculation method is to count the number of efficient users and inefficient users in the statistical data set, and then divide them by the total number of users. The calculation formula is , ;

[0179] Among them, P(high) and P(low) are the prior probabilities of efficient users and inefficient users respectively, and N 高 、N 低 are the number of efficient users and inefficient users respectively, and N is the total number of efficient users and inefficient users;

[0180] Calculate the conditional probability. Conditional probability refers to the probability of a keyword appearing when a certain user type is known. The calculation formula is , ;

[0181] in, 、 are the conditional probabilities of efficient users and inefficient users respectively, For keywords, N 高 、N 低 are the number of efficient users and inefficient users respectively, To include keywords in efficient user query records The number of records, To include keywords in the query records of inefficient users Number of records;

[0182] According to Bayes' theorem, for a keyword ,Compare ×P (high) and × the size of P(low);

[0183] like ×P(High)> ×P(low), keyword tends to be more associated with efficient users;

[0184] like ×P(height)< ×P(low), keyword tends to be more associated with inefficient users;

[0185] like ×P(height)≈ ×P(low), keyword It meets the needs of both efficient and inefficient users;

[0186] Repeat the prediction step for all keywords in the dataset and find the keywords that the model judges to be both efficient and inefficient users as new classification labels.

[0187] The technical solutions and benefits of the embodiments of the present application are:

[0188] Count the number of queries for each real estate registration file in the historical period and the total number of real estate files, calculate the relative query ratio of the real estate registration files, divide the real estate registration files into high-frequency query files and low-frequency query files, for high-frequency query files, screen out users with high-frequency and stable queries, establish a corresponding user database, calculate the query efficiency of high-frequency query files, compare it with the threshold, and mark the files with query efficiency greater than or equal to the threshold as high-efficiency files. For high-efficiency files, divide the corresponding users in the user database into high-efficiency users and low-efficiency users, and judge whether the difference between the query information of high-efficiency users and low-efficiency users and the classification label of the file is large. If large, judge whether the classification label needs to be changed based on the ratio of the actual number of high-efficiency users to the ideal number of high-efficiency users. If the classification label needs to be changed, establish a query information based on the high-efficiency users and the low-efficiency users. Information database, extract keywords, identify keywords that are consistent with both efficient users and inefficient users as new classification labels, this application divides the real estate registration archives into high-frequency query archives and low-frequency query archives, and screens out high-frequency stable query users corresponding to the high-frequency query archives, and divides the high-frequency query archives into high-efficiency archives and low-efficiency archives by calculating the query efficiency of the high-frequency query archives, and divides the users corresponding to the high-efficiency archives into high-efficiency users and low-efficiency users. By analyzing the differences between the query information of high-efficiency users and inefficient users and the archives and the number of efficient users, it is determined whether to change the classification labels of the archives, and keywords that are consistent with both efficient users and inefficient users are identified as new classification labels, thereby solving the problem of low efficiency of user queries due to unreasonable label classification of archives in the process of real estate registration archive sorting, and improving the efficiency of user queries.

[0189] Example 2

[0190] See also Figure 2 As shown, a real estate registration file arrangement service system according to an embodiment of the present invention includes:

[0191] File query frequency classification module: counts the number of queries for each real estate registration file in the historical period and the total number of real estate files, calculates the relative query ratio of real estate registration files, and divides real estate registration files into high-frequency query files and low-frequency query files;

[0192] The relative query ratio refers to the ratio of the number of queries to the real estate registration files to the total number of real estate registration files within the historical period. The calculation formula is:

[0193]

[0194] The process of dividing the real estate registration files into high-frequency query files and low-frequency query files includes:

[0195] Compare the relative query ratio of real estate registration files with a threshold value, classify real estate registration files with a relative query ratio greater than or equal to the threshold value as high-frequency query files, and classify real estate registration files with a relative query ratio less than the threshold value as low-frequency query files;

[0196] The threshold setting of the relative query ratio includes: calculating the relative query ratio of all real estate registration files in the historical period, calculating the average value, and using the average value of the relative query ratio as the threshold;

[0197] High-frequency and stable user screening module: For high-frequency query files, filter out users with high-frequency and stable queries and establish a corresponding user database;

[0198] The process of screening out users with high-frequency and stable queries is as follows:

[0199] Divide the historical period into n small time periods. For each user, count the number of queries in each small time period to generate a query count sequence.

[0200] Compare the number of queries per user in a small time period with the set threshold. If the number of queries in a small time period is greater than the threshold, mark the small time period as a high-frequency time period.

[0201] The setting of the query number threshold includes: calculating the average value of the query number sequence and using the average value as the query number threshold;

[0202] Count the number of high-frequency time periods for each user in the historical period and calculate the ratio of this to the total number of small time periods to obtain the high-frequency time period ratio.

[0203] Calculate the standard deviation of each user's query count sequence using the formula: , where n is the number of small time periods, The number of queries for each small time period, is the average number of queries;

[0204] For each user, the high-frequency time period ratio and query time standard deviation are compared with the threshold, and users with a high-frequency time period ratio greater than or equal to the threshold and a query time standard deviation greater than or equal to the threshold are selected as high-frequency stable query users;

[0205] High-efficiency archive and user classification module: Calculates the query efficiency of frequently queried archives, compares it with the threshold, and marks archives with query efficiency greater than or equal to the threshold as high-efficiency archives. For high-efficiency archives, the corresponding users in the user database are divided into high-efficiency users and low-efficiency users.

[0206] The calculation process of the high-frequency query archive query efficiency includes:

[0207] Record the start and end time of each query to the real estate registration file, calculate the duration of each query, then add up the duration of all query operations within a certain period of time (such as a month or a quarter), and divide it by the number of queries to get the average query time;

[0208] The query success rate is calculated by counting the ratio of the number of successful searches for the required files to the total number of searches within a certain period of time. A successful search means that the user can accurately obtain the required file information within the specified time.

[0209] When a user enters a keyword or other search condition to conduct a search, the accuracy rate of the search results is obtained by calculating the ratio of the number of files related to the user's actual needs to the total number of search results.

[0210] The query efficiency is calculated based on the average query time, query success rate, and retrieval result accuracy. The process includes:

[0211] The data of average query time, query success rate and retrieval result accuracy are standardized. For average query time, the smaller the value, the better. The formula is used. Normalize to get the normalized value of average query time, where X is the original average query time, Max is the maximum value of the average query time, and Min is the minimum value of the average query time;

[0212] For query success rate and retrieval result accuracy, the larger the value, the better. The formula is used. Perform normalization to obtain the normalized value of query success rate and the normalized value of retrieval result accuracy, where X is the original query success rate or retrieval result accuracy, Max is the maximum value of the query success rate or retrieval result accuracy, and Min is the minimum value of the query success rate or retrieval result accuracy;

[0213] The standardized data is multiplied by their respective weights and then added together to obtain the comprehensive query efficiency. The formula is:

[0214] in, 、 、 are the weights of the standardized value of average query time, the standardized value of query success rate, and the standardized value of retrieval result accuracy;

[0215] Compare the calculated query efficiency of each file with the set threshold. If the query efficiency of the file is higher than the threshold, it is classified as an efficient file. If the query efficiency of the file is lower than the threshold, it is classified as an inefficient file.

[0216] The threshold setting of the archive query efficiency includes: adding the query efficiencies of all archives to obtain the total query efficiency, and dividing the total query efficiency by the number of archives to obtain the average archive query efficiency. The calculation formula is: , taking the average value of query efficiency as the threshold, where is the average value, is the query efficiency of the i-th file, n is the total number of files;

[0217] The process of dividing users corresponding to high-efficiency query files into high-efficiency users and low-efficiency users includes:

[0218] For each user with a high-efficiency profile, count the number of high-efficiency profiles N that the user has queried in the historical period. 高效 And the total number of queries N 总 , calculate the ratio of efficient files: , where R i is the ratio of efficient files;

[0219] Compare the percentage of efficient files of each user with the set threshold. If the percentage of efficient files is greater than or equal to the threshold, the user is an efficient user. If the percentage of efficient files is less than the threshold, the user is an inefficient user.

[0220] The threshold setting of the efficient file ratio includes: listing the efficient file ratio distribution of all users, taking the median of the efficient file ratio as the threshold of the efficient ratio;

[0221] Classification label difference evaluation module: determines whether the difference between the query information of efficient users and inefficient users and the classification labels of the archives is large. If so, it determines whether the classification labels need to be changed based on the ratio of the actual number of efficient users to the ideal number of efficient users;

[0222] The process of determining whether the difference between the query information of the efficient user and the inefficient user and the classification label of the archive is large includes:

[0223] Collect query information from efficient and inefficient users, where the query information includes keywords, phrases, or sentences entered by users during querying, and obtain classification labels for archives;

[0224] Using lexical analysis tools in natural language processing technology, we process the query information of efficient and inefficient users and the classification labels of the archives, extract keywords, and generate query information keyword set A and classification label keyword set B respectively;

[0225] Calculate the intersection of set A and set B, that is, the set consisting of elements belonging to both A and B, denoted as , count the number of elements in the intersection, recorded as ;

[0226] Calculate the union of set A and set B, that is, the set consisting of elements belonging to A or B, denoted as , count and collect the number of elements, recorded as ;

[0227] According to the formula of Jaccard similarity coefficient: , calculate the matching scores between the query information of efficient users and inefficient users and the archive classification labels;

[0228] The Jaccard similarity coefficient is an indicator used to measure the similarity between two sets. Its value range is [0,1]. The closer the coefficient is to 1, the higher the similarity between the two sets, and the closer it is to 0, the lower the similarity.

[0229] Compare the matching scores between the query information of efficient and inefficient users and the archive classification labels with the threshold. If the matching scores are less than or equal to the threshold, it means that the difference between the query information of efficient and inefficient users and the archive classification labels is large.

[0230] The process of determining whether the classification label needs to be changed includes:

[0231] Calculate the ratio of the actual number of efficient users to the ideal number of efficient users to obtain the actual efficient user ratio, and compare the actual efficient user ratio with the threshold;

[0232] If the actual percentage of efficient users is greater than or equal to the threshold, the classification label of the file needs to be changed;

[0233] If the actual percentage of efficient users is less than the threshold, there is no need to change the classification label of the file;

[0234] The ideal number of efficient users can be calculated by analyzing the behavioral data of user query files over a period of time and calculating the average number of efficient users under the existing classification labels.

[0235] New category label generation module: If a category label needs to be changed, a query information database is established based on high-efficiency users and low-efficiency users, keywords are extracted, and keywords that match both high-efficiency and low-efficiency users are identified as new category labels;

[0236] The data contained in the query information database includes:

[0237] Record the user's identity (such as user number, name, etc.) and user type (efficient user or inefficient user);

[0238] Completely record the original query information such as keywords, phrases or sentences entered by users when querying real estate registration files;

[0239] Record the number, name and other identification information of the real estate registration file involved in the user's query operation;

[0240] The method of extracting keywords is:

[0241] Use lexical analysis tools in natural language processing technology to process user query information and extract representative keywords;

[0242] Keywords refer to representative words or phrases that are extracted after processing the query information of efficient and inefficient users and the classification labels of archives and can reflect the main content and core meaning of the query information or classification labels;

[0243] In step 5, the process of identifying keywords that are suitable for both efficient users and inefficient users includes:

[0244] Label the extracted keywords to clarify whether each keyword belongs to an efficient or inefficient user, and organize the extracted keywords and corresponding user types into a data set;

[0245] Using the Naive Bayes model to identify keywords that are suitable for both efficient and inefficient users, the process includes:

[0246] Calculate the prior probability. The prior probability refers to the probability of a certain user type appearing without any query information. The calculation method is to count the number of efficient users and inefficient users in the statistical data set, and then divide them by the total number of users. The calculation formula is , ;

[0247] Among them, P(high) and P(low) are the prior probabilities of efficient users and inefficient users respectively, and N 高 、N 低 are the number of efficient users and inefficient users respectively, and N is the total number of efficient users and inefficient users;

[0248] Calculate the conditional probability. Conditional probability refers to the probability of a keyword appearing when a certain user type is known. The calculation formula is , ;

[0249] in, 、 are the conditional probabilities of efficient users and inefficient users respectively, For keywords, N 高 、N 低 are the number of efficient users and inefficient users respectively, To include keywords in efficient user query records The number of records, To include keywords in the query records of inefficient users Number of records;

[0250] According to Bayes' theorem, for a keyword ,Compare ×P (high) and × the size of P(low);

[0251] like ×P(High)> ×P(low), keyword tends to be more associated with efficient users;

[0252] like ×P(height)< ×P(low), keyword tends to be more associated with inefficient users;

[0253] like ×P(height)≈ ×P(low), keyword It meets the needs of both efficient and inefficient users;

[0254] Repeat the prediction step for all keywords in the dataset and find the keywords that the model judges to be both efficient and inefficient users as new classification labels.

[0255] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A real estate registration file arrangement service method, characterized by: include: Step 1: Count the number of queries for each real estate registration file in the historical period and the total number of real estate files, calculate the relative query ratio of the real estate registration files, and divide the real estate registration files into high-frequency query files and low-frequency query files; Step 2: For high-frequency query files, filter out users with high-frequency and stable queries and establish a corresponding user database; Step 3: Calculate the query efficiency of high-frequency query files and compare it with the threshold. Mark the files with query efficiency greater than or equal to the threshold as high-efficiency files. For high-efficiency files, divide the corresponding users in the user database into high-efficiency users and low-efficiency users. Step 4: Determine whether the difference between the query information of efficient users and inefficient users and the classification labels of the archives is large. If so, determine whether the classification labels need to be changed based on the ratio of the actual number of efficient users to the ideal number of efficient users. Step 5: If the classification label needs to be changed, establish a query information database based on efficient users and inefficient users, extract keywords, and identify keywords that are consistent with both efficient users and inefficient users as new classification labels.

2. The real estate registration file arrangement service method according to claim 1, characterized in that: The method of dividing the real estate registration files into high-frequency query files and low-frequency query files is as follows: Comparing the relative query ratio of real estate registration files with a threshold value; Classify real estate registration files with a relative query ratio greater than or equal to a threshold as high-frequency query files; Real estate registration files with a relative query ratio less than a threshold are classified as low-frequency query files.

3. The real estate registration file arrangement service method according to claim 2, characterized in that: The relative query ratio of the real estate registration files is calculated as follows: During the historical period, the relative query ratio of real estate registration files is calculated by calculating the ratio of the number of queries to the total number of real estate registration files.

4. The real estate registration file arrangement service method according to claim 1, characterized in that: The process of screening out users with high-frequency and stable queries includes: Divide the historical period into n small time periods. For each user, count the number of queries in each small time period to generate a query count sequence. Compare the number of queries per user in a small time period with the set threshold. If the number of queries in a small time period is greater than the threshold, mark the small time period as a high-frequency time period. Count the number of high-frequency time periods for each user in the historical period and calculate the ratio of this to the total number of small time periods to obtain the high-frequency time period ratio. For each user, the high-frequency time period ratio and query number series standard deviation are compared with the threshold, and users with a high-frequency time period ratio greater than or equal to the threshold and a query time standard deviation greater than or equal to the threshold are screened as high-frequency stable query users.

5. The real estate registration file arrangement service method according to claim 1, characterized in that: The calculation process of the high-frequency query archive query efficiency includes: Record the start and end time of each query to the real estate registration file, calculate the time taken for each query, then add up the time taken for all query operations within a certain period of time and divide it by the number of queries to get the average query time; The query success rate is calculated by counting the ratio of the number of successful searches for the required files to the total number of searches within a certain period of time. A successful query means that the user can accurately obtain the required file information within the specified time; When a user enters a keyword or other search condition to conduct a search, the accuracy rate of the search results is obtained by calculating the ratio of the number of files related to the user's actual needs to the total number of search results. The average query time, query success rate, and retrieval result accuracy are standardized, and the standardized data are multiplied by their respective weights and then added together to obtain the query efficiency.

6. The real estate registration file arrangement service method according to claim 1, characterized in that: The process of dividing the corresponding users in the user database into efficient users and inefficient users includes: For each user with a high-efficiency profile, count the number of high-efficiency profiles and the total number of profiles queried by the user during the historical period, and calculate the proportion of high-efficiency profiles. Compare the effective file ratio of each user with the set threshold; If the percentage of efficient files is greater than or equal to the threshold, the user is an efficient user; If the ratio of efficient files is less than the threshold, the user is an inefficient user.

7. The real estate registration file arrangement service method according to claim 1, characterized in that: The method for determining whether the difference between the query information of efficient users and inefficient users and the classification labels of the archives is large is as follows: Collect query information from efficient and inefficient users, extract keywords, and generate query information keyword set A and classification label keyword set B respectively; Calculate the intersection and union of set A and set B respectively, and count the number of elements in the intersection and union; The matching scores between the query information of efficient users and inefficient users and the archive classification labels are calculated according to the Jaccard coefficient formula; Compare the matching scores between the query information of efficient users and inefficient users and the archive classification labels with a threshold; If the matching score is less than or equal to the threshold, it means that the difference between the query information of the efficient user and the inefficient user and the classification label of the archive is large.

8. The real estate registration file arrangement service method according to claim 1, characterized in that: The method of identifying keywords that are both suitable for efficient users and inefficient users is as follows: Label the extracted keywords to clarify whether each keyword belongs to an efficient or inefficient user, and organize the extracted keywords and corresponding user types into a data set; The Naive Bayes model is used to identify keywords that are suitable for both efficient and inefficient users.

9. The real estate registration file arrangement service method according to claim 8, characterized in that: The process of using the Naive Bayes model to identify keywords that are suitable for both efficient users and inefficient users includes: Calculate the prior probability and conditional probability of efficient users and inefficient users respectively; Compare the product of the prior probability and conditional probability of efficient users with the product of the prior probability and conditional probability of inefficient users; If the product of the prior probability and conditional probability of an efficient user is approximately equal to the product of the prior probability and conditional probability of an inefficient user, it means that the keyword is suitable for both efficient and inefficient users.

10. A real estate registration file arrangement service system, characterized by: File query frequency classification module: counts the number of queries for each real estate registration file in the historical period and the total number of real estate files, calculates the relative query ratio of real estate registration files, and divides real estate registration files into high-frequency query files and low-frequency query files; High-frequency and stable user screening module: For high-frequency query files, filter out users with high-frequency and stable queries and establish a corresponding user database; High-efficiency archive and user classification module: Calculates the query efficiency of frequently queried archives, compares it with the threshold, and marks archives with query efficiency greater than or equal to the threshold as high-efficiency archives. For high-efficiency archives, the corresponding users in the user database are divided into high-efficiency users and low-efficiency users. Classification label difference evaluation module: determines whether the difference between the query information of efficient users and inefficient users and the classification labels of the archives is large. If so, it determines whether the classification labels need to be changed based on the ratio of the actual number of efficient users to the ideal number of efficient users; New classification label generation module: If the classification label needs to be changed, a query information database is established based on efficient users and inefficient users, keywords are extracted, and keywords that are consistent with both efficient users and inefficient users are identified as new classification labels.

Images