User electronic portrait establishing method for huechys matching system

Abstract

The invention discloses a user electronic portrait establishing method for a huechys matching system, and relates to the technical field of information processing.The method comprises the steps that main features of each piece of identity information are extracted, and the identity information is classified through the main features; the matching value degree, the regional value degree and the user control degree of each kind of identity information are obtained through calculation, then weighted fusion is carried out, and the importance degree of each kind of identity information is calculated; grading all types of identity information by using the importance degree; calculating the information sensitivity of the real-time user, selecting the identity information of the real-time user according to the information sensitivity, and uploading the identity information to a huechys matching system; in the system, a real-time user electronic portrait is constructed by using uploaded real-time user identity information, matching is performed in a system personnel database by using the user electronic portrait, and adaptive personnel of the real-time user are searched for recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of information processing, and particularly to a method for establishing a user electronic portrait for a matchmaker matching system. Background Art

[0002] Before 2010, matchmaking services mainly relied on offline matchmaking agencies and blind date corners, depending on manual registration of basic information, with low matching efficiency and being easily affected by subjective biases. After that, with the development of information technology, various dating apps emerged one after another. Users began to actively fill in tags such as "annual income, housing situation", etc., but it only stayed at the listing of static information, and the matching accuracy rate was less than 30%. Currently, with the rapid development of artificial intelligence, the speed of information calculation and analysis has been greatly accelerated, and the matching of various information such as basic information and emotional information has been realized in the matchmaker matching system; the matching success rate has been greatly increased.

[0003] However, with the increase in the scope of information collection and processing and the significant increase in the matching success rate, problems regarding the security of user information have emerged one after another. There are great concerns about users entering personal information in the matchmaker matching system, and the system cannot be fully trusted; while reducing the upload and input of information in the system will result in a lack of necessary and effective information for matching. Therefore, how to balance the protection of personal privacy and the effectiveness of uploaded information is crucial. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for establishing a user electronic portrait for a matchmaker matching system to solve the problems raised in the prior art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] A method for establishing a user electronic portrait for a matchmaker matching system, the method comprising the following steps:

[0007] S100. Collect the identity information of all users in the matchmaker matching system, extract the main features of each identity information, and classify the identity information using the main features;

[0008] Further, the specific steps of classifying the identity information using the main features are:

[0009] S101. Collect the identity information of all users in the matchmaker matching system, construct a structured data set, preprocess the identity information in the data set, determine whether there are missing values in the data in the data set, calculate the average value of the adjacent data at the missing place as the filling value, and fill the missing values; normalize the preprocessed structured data;

[0010] S102. For each piece of collected identity information, extract the documents and words in the collected user identity information, and calculate the importance of each word. The formula is as follows:

[0011]

[0012] In the formula, G(i, d) represents the importance of the i-th word in document d, n (i，d) represents the number of the i-th word in document d, w d represents the total number of all words in document d, W represents the total number of documents of the collected identity information, D i represents the number of documents including the i-th word;

[0013] Calculate the importance of each word in turn. Compare the importance of all words in each identity information, and take the word with the greatest importance as the main feature of the identity information; obtain the main features of each piece of collected identity information in turn;

[0014] S103. Extract the main features of the identity information in pairs, and calculate the similarity of the extracted main features. The formula is as follows:

[0015]

[0016] In the formula, Sim(A, B) represents the similarity of the calculated main features A and B, A and B represent the main features of two pieces of extracted identity information, and A·B represents the dot product of A and B; calculate the similarity of the main features of all identity information in turn, set the similarity threshold, and classify those exceeding the similarity threshold as the same type of identity information to classify all the collected identity information.

[0017] Classify the collected identity information in the matching system so that subsequent operations can analyze different types of identity information, extract key identity information, and help the system find matching features more quickly;

[0018] S200. Extract the influence of each type of identity information on matching in the system, the behavior data of users in different regions, and the display quantity of all users, and calculate the matching value degree, regional value degree, and user control degree of each type of identity information respectively;

[0019] Furthermore, the specific steps for calculating the matching value degree, regional value degree, and user control degree of each type of identity information respectively are as follows:

[0020] S201. Collect the historical matching results and the types of identity information during matching in the system, and calculate the matching value degree of each type of identity information. The formula is as follows:

[0021]

[0022] In the formula, P j represents the matching value degree of the j-th type of identity information calculated, C j represents the number of successful matches when using the j-th type of identity information for matching, Z j represents the total number of times of using the j-th type of identity information for matching, m represents the total number of types of identity information selected during matching; calculate the matching value degrees of m types of identity information in sequence;

[0023] S202. Collect the user behavior data of all types of identity information in different regions, and calculate the regional value degree of each type of identity information in different regions. The formula is:

[0024]

[0025] In the formula, Du (j，u) represents the regional value degree of the j-th type of identity information in region u, V (j，u) represents the number of times the j-th type of identity information is browsed by users in region u, Cd (j，u) ’ represents the number of successful matches using the j-th type of identity information in region u; calculate the regional value degrees of m types of identity information in sequence;

[0026] S203. For the j-th type of identity information, collect the number of times the user is displayed and the number of successful matches of the user, and calculate the user control degree of the j-th type of identity information. The formula is:

[0027]

[0028] In the formula, Yu h represents the user control degree of the j-th type of identity information in user h, M h represents the number of successful matches of the j-th type of identity information in user h, E h represents the number of times the j-th type of identity information is displayed in user h. Calculate the user control degrees of the j-th type of identity information among all users in the same way, and then calculate the average value of the user control degrees as the final user control degree Yu j ; calculate the user control degrees of m types of identity information in sequence, and finally obtain three indicators of each type of identity information, including matching value degree, regional value degree and user control degree.

[0029] Calculate the matching value degree, regional value degree and user control degree of the identity information, and comprehensively analyze the importance of each type of identity information through calculations in three dimensions. In the matchmaker matching system, effective identity information can be accurately extracted, making the matching result more accurate;

[0030] S300. Perform weighted fusion on the matching value degree, regional value degree, and user control degree of each type of identity information, calculate the importance degree of each type of identity information; use the importance degree to classify all types of identity information;

[0031] Further, the specific steps for classifying all types of identity information using the importance degree are as follows:

[0032] S301. Standardize the three indicators of each type of identity information calculated, use the information entropy formula to calculate the information entropy of each indicator respectively, and calculate the weight value of each indicator using the information entropy of each indicator. The formula is:

[0033]

[0034] In the formula, β a represents the weight of indicator a, S a represents the information entropy of indicator a, and S g represents the information entropy of the gth indicator;

[0035] S302. Use the weight of each indicator calculated to perform weighted fusion on the three indicators to obtain the importance degree of each type of identity information. The formula is:

[0036] In j = β1×P j + β2×Du (j，u) + β3×Yu j ;

[0037] In the formula, In j represents the importance degree of the jth type of identity information, β1 represents the weight of the matching value degree, β2 represents the weight of the regional value degree, and β3 represents the weight of the user control degree;

[0038] S303. Sort the importance degrees of all identity information in ascending order, set the classification percentage, perform grade division according to the percentage, regard the first 20% of the identity information as low-level information, regard the identity information of 20% - 80% as medium-level information, and regard the last 20% of the identity information as high-level information.

[0039] Calculate the weights of the three indicators, intelligently calculate the importance degree of each type of identity information, use the importance degree to classify the identity information, clarify the influence of different identity information on the matching result in the matching, can save system computing power, and greatly improve the recommendation speed; use the identity information with a high level for matching, reduce the mis-matching rate, and strengthen the matching causal relationship;

[0040] S400. When performing real-time matching for users in the matchmaker matching system, collect the display times and display durations of the user's own identity information at different levels in historical network activities, calculate the information sensitivity of the real-time user, and use the information sensitivity to select the identity information of the real-time user and upload it to the matchmaker matching system.

[0041] Further, the specific steps for using the information sensitivity to select the identity information of the real-time user and upload it to the matchmaker matching system are as follows:

[0042] S401. When performing real-time matching for users in the matchmaker matching system, collect the display times, display durations, and interaction times of the user's own different types of identity information in historical network activities, and calculate the interaction conversion rate of each type of identity information. The formula is: In the formula, J j represents the interaction conversion rate of the j-th type of identity information, H j represents the number of interactions after the display of the j-th type of identity information in the real-time user's history, sh j represents the number of displays of the j-th type of identity information in the real-time user's history, to prevent division by zero;

[0043] S402. Construct a decay function for the collected display durations to calculate the total time decay weight. The formula is: In the formula, Q j represents the total time decay weight of the real-time user for the j-th type of identity information, T now represents the real-time time, t r represents the display duration at the r-th display, λ represents the decay coefficient, and U represents the total number of displays of the j-th type of identity information in the real-time user's history;

[0044] S403. Calculate the sensitivity of the real-time user to the j-th type of identity information using the interaction conversion rate and the total decay weight. The formula is: In the formula, Su j represents the sensitivity of the real-time user to the j-th type of identity information; calculate the sensitivity of the real-time user to all identity information in turn and normalize it to Su';

[0045] S404. Assign values to the levels of identity information, with high-level information being 3, medium-level information being 2, and low-level information being 1; calculate the recommendation priority using the normalized sensitivity and level assignment of all identity information. The formula is: Priority = Level assignment × (1 - Sensitivity Su'); calculate the priorities of all identity information, arrange the priorities from smallest to largest, calculate the difference between adjacent priorities, extract the maximum value of the difference as the node threshold, and upload the identity information with a priority greater than the node threshold to the matchmaker matching system.

[0046] Calculate the sensitivity of real-time users to each type of identity information. When making matches, protect the users' identity information according to the sensitivity of real-time users, which greatly improves the users' experience. Also, combine sensitivity and importance grading, and comprehensively consider the objective importance of each type of identity information and the subjective display willingness of real-time users, which not only ensures the effectiveness of the uploaded information during matching but also takes into account the sensitivity of users to their own identity information and protects the users' identity information.

[0047] S500. Use the uploaded real-time user identity information in the system to construct a real-time user electronic portrait, and use the user electronic portrait to match in the system personnel database to find the suitable personnel for the real-time user for recommendation.

[0048] Further, the specific steps for finding the suitable personnel for the real-time user for recommendation are as follows:

[0049] S501. Use the uploaded identity information as portrait features to construct a real-time user electronic portrait, use the user electronic portrait to match in the system personnel database, calculate the similarity between the real-time user electronic portrait and each person in the system personnel database through cosine similarity, and finally select the person with the highest similarity as the suitable person for recommendation to the real-time user.

[0050] S600. Build a visualization platform to visually display the real-time matching process to the user.

[0051] Further, the specific steps for visually displaying the real-time matching process to the user are as follows:

[0052] S601. Build a visualization platform to display the identity information of the selected suitable person to the real-time user in the visualization platform, extract the system matching process, correspondently display the identity information uploaded by the real-time user himself and the identity information of the suitable person, and display the importance grading, sensitivity, and priority of the user's own identity information.

[0053] S700. The real-time user views the recommended person and the matching process through the visualization platform, the real-time user reviews and actively modifies the matching process, and the system constructs a personalized matching mechanism according to the modified content for re-matching and recommendation.

[0054] Further, the specific steps for the system to construct a personalized matching mechanism according to the modified content for re-matching and recommendation are as follows:

[0055] S701. The real-time user views the recommended person and the matching process through the visualization platform, the user subjectively adjusts all the proportional coefficients, grading percentages, and thresholds in the matching process, and after adjustment, the system constructs a personalized matching mechanism according to the modified content for re-matching and recommendation.

[0056] The visualization platform is used to display the matching process, enabling users to view the matching principle, enhancing the interpretability of the matching; allowing users to subjectively adjust the matching process, preventing the system from overfitting using algorithms and resulting in a lack of personalization in the results, ensuring the differentiation of system matching, and making the matching results more suitable for users.

[0057] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0058] 1. The present invention calculates the weights of three indicators, intelligently calculates the importance of each identity information, classifies the identity information using the importance, clarifies the influence of different identity information on the matching result in the matching, can save system computing power, and greatly improve the recommendation speed; uses the identity information with a higher level for matching, reduces the mis-matching rate, and strengthens the matching causal relationship.

[0059] 2. The present invention calculates the sensitivity of the real-time user to each identity information. When performing matching, it protects the user's identity information according to the sensitivity of the real-time user, greatly improving the user experience; and combines the sensitivity and importance classification, comprehensively considering the objective importance of each identity information and the subjective display willingness of the real-time user, not only ensuring the success rate of the matching, but also considering the user's sensitivity to their own identity information and protecting the user's identity information.

[0060] 3. The present invention uses the visualization platform to display the matching process, enabling users to view the matching principle, enhancing the interpretability of the matching; allowing users to subjectively adjust the matching process, preventing the system from overfitting using algorithms and resulting in a lack of personalization in the results, ensuring the differentiation of system matching, and making the matching results more suitable for users. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a schematic diagram of the steps of a method for establishing a user electronic portrait for a matchmaker matching system according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0063] Embodiment: As Figure 1 shown, the present invention provides a technical solution,

[0064] A method for establishing a user electronic portrait for a matchmaker matching system, the method comprising the following steps:

[0065] S100. Collect the identity information of all users in the matchmaker matching system, extract the main features of each piece of identity information, and classify the identity information using the main features;

[0066] The specific steps for classifying the identity information using the main features are as follows:

[0067] S101. Collect the identity information of all users in the matchmaker matching system, construct a structured data set, preprocess the identity information in the data set, judge whether there are missing values in the data in the data set, calculate the average value of the adjacent data at the missing place as the filling value, and fill the missing values; normalize the preprocessed structured data;

[0068] S102. For each piece of collected identity information, extract the documents and words in the collected user identity information, calculate the importance of each word, and the formula is:

[0069]

[0070] In the formula, G(i, d) represents the importance of the i-th word in document d, n (i，d) represents the number of the i-th word in document d, w d represents the total number of all words in document d, W represents the total number of documents of the collected identity information, D i represents the number of documents including the i-th word;

[0071] Calculate the importance of each word in turn, compare the importance of all words in each piece of identity information, and take the word with the greatest importance as the main feature of the identity information; obtain the main features of each piece of collected identity information in turn;

[0072] S103. Extract the main features of the identity information in pairs, calculate the similarity of the extracted main features, and the formula is:

[0073]

[0074] In the formula, Sim(A, B) represents the calculated similarity of the main features A and B, A and B represent the main features of the two pieces of extracted identity information, and A·B represents the dot product of A and B; calculate the similarity of the main features of all identity information in turn, set the similarity threshold, classify those exceeding the similarity threshold into the same type of identity information, and classify all the collected identity information.

[0075] Classify the collected identity information in the matching system so that subsequent operations can analyze different types of identity information, extract key identity information, and help the system find matching features more quickly;

[0076] S200. Extract the impact of each type of identity information on matching in the system, the behavior data of users in different regions, and the display quantity of all users, and calculate the matching value degree, regional value degree, and user control degree of each type of identity information respectively;

[0077] The specific steps for calculating the matching value degree, regional value degree, and user control degree of each type of identity information respectively are as follows:

[0078] S201. Collect the historical matching results and the types of identity information during matching in the system, and calculate the matching value degree of each type of identity information. The formula is:

[0079]

[0080] In the formula, P j represents the matching value degree of the j-th type of identity information to be calculated, C j represents the number of successful matches when using the j-th type of identity information for matching, Z j represents the total number of times of matching using the j-th type of identity information, and m represents the total number of types of identity information selected during matching; calculate the matching value degrees of m types of identity information in sequence;

[0081] S202. Collect the user behavior data of all types of identity information in different regions, and calculate the regional value degree of each type of identity information in different regions. The formula is:

[0082]

[0083] In the formula, Du (j，u) represents the regional value degree of the j-th type of identity information in region u, V (j，u) represents the number of times the j-th type of identity information is browsed by users in region u, Cd (j，u) ’ represents the number of successful matches using the j-th type of identity information in region u; calculate the regional value degrees of m types of identity information in sequence;

[0084] S203. For the j-th type of identity information, collect the number of times the user is displayed and the number of successful matches of the user, and calculate the user control degree of the j-th type of identity information. The formula is:

[0085]

[0086] In the formula, Yu h represents the user control degree of the j-th type of identity information in user h, M h represents the number of successful matches of the j-th type of identity information in user h, E hDenote the number of times the j-th type of identity information is displayed in user h. Calculate the user control degree of the j-th type of identity information in all users in the same way. Then calculate the average value of the user control degree as the final user control degree Yu of the j-th type of identity information j ; Calculate the user control degrees of m types of identity information in sequence. Finally, obtain three indicators for each type of identity information, including the matching value degree, regional value degree, and user control degree.

[0087] Calculate the matching value degree, regional value degree, and user control degree of identity information. Through the calculations in three dimensions, comprehensively analyze the importance of each type of identity information, and can accurately extract effective identity information in the matchmaker matching system, increasing the effectiveness of matching information;

[0088] S300. Perform weighted fusion on the matching value degree, regional value degree, and user control degree of each type of identity information to calculate the importance of each type of identity information; Use the importance to classify all types of identity information;

[0089] The specific steps for classifying all types of identity information using the importance are as follows:

[0090] S301. Standardize the three indicators of each type of identity information calculated. Use the information entropy formula to calculate the information entropy of each indicator respectively, and calculate the weight value of each indicator using the information entropy of each indicator. The formula is:

[0091]

[0092] In the formula, β a represents the weight of indicator a, S a represents the information entropy of indicator a, S g represents the information entropy of the g-th indicator;

[0093] S302. Use the weight of each indicator calculated to perform weighted fusion on the three indicators to obtain the importance of each type of identity information. The formula is:

[0094] In j =β1×P j +β2×Du (j，u) +β3×Yu j ;

[0095] In the formula, In j represents the importance of the j-th type of identity information, β1 represents the weight of the matching value degree, β2 represents the weight of the regional value degree, and β3 represents the weight of the user control degree;

[0096] S303. Sort the importance levels of all identity information in ascending order, set the classification percentages, and perform level classification according to the percentages. Consider the top 20% of the identity information as low-level information, the 20%-80% of the identity information as medium-level information, and the last 20% of the identity information as high-level information.

[0097] Calculate the weights of the three indicators, intelligently calculate the importance level of each piece of identity information, use the importance level to classify the identity information, clarify the impact of different identity information on the matching result in the matching process, which can save system computing power and greatly improve the recommendation speed; use the identity information with a higher level for matching, reduce the false matching rate, and strengthen the matching causal relationship.

[0098] S400. When performing real-time matching on users in the matchmaker matching system, collect the display times and display durations of the user's different-level identity information in their historical network activities, calculate the information sensitivity of the real-time user, and use the information sensitivity to select the identity information of the real-time user to upload to the matchmaker matching system.

[0099] The specific steps of using the information sensitivity to select the identity information of the real-time user to upload to the matchmaker matching system are as follows:

[0100] S401. When performing real-time matching on users in the matchmaker matching system, collect the display times, display durations, and interaction times of the user's different types of identity information in their historical network activities, calculate the interaction conversion rate of each piece of identity information. The formula is: In the formula, J j represents the interaction conversion rate of the j-th piece of identity information, H j represents the number of interactions after the display of the j-th piece of identity information in the history of the real-time user, sh j represents the number of displays of the j-th piece of identity information in the history of the real-time user, Prevent division by zero;

[0101] S402. Construct an attenuation function for the collected display durations to calculate the total time attenuation weight. The formula is: In the formula, Q j represents the total time attenuation weight of the real-time user for the j-th piece of identity information, T now represents the real-time, t r represents the display duration at the r-th display, λ represents the attenuation coefficient, and U represents the total number of displays of the j-th piece of identity information in the history of the real-time user;

[0102] S403. Calculate the sensitivity of the real-time user to the j-th piece of identity information using the interaction conversion rate and the total attenuation weight. The formula is: In the formula, Su jIndicates the sensitivity of the real-time user to the j-th type of identity information; calculate the sensitivity of the real-time user to all identity information in sequence and normalize it to Su';

[0103] S404. Assign grades to the classification of identity information, with high-level information being 3, medium-level information being 2, and low-level information being 1; use the normalized sensitivity and grade assignment of all identity information to calculate the recommendation priority. The formula is: Priority = Grade Assignment × (1 - Sensitivity Su'); calculate the priorities of all identity information, arrange the priorities from smallest to largest, calculate the difference between adjacent priorities, extract the maximum value of the difference as the node threshold, and upload the identity information with a priority greater than the node threshold to the matchmaker matching system.

[0104] Calculate the sensitivity of the real-time user to each type of identity information. When making a match, protect the user's identity information according to the sensitivity of the real-time user, which greatly improves the user experience; and combine the sensitivity and importance classification, comprehensively consider the objective importance of each type of identity information and the subjective display willingness of the real-time user, which not only ensures the effectiveness of the uploaded information during the match but also considers the user's sensitivity to their own identity information and protects the user's identity information.

[0105] S500. Use the uploaded real-time user identity information in the system to construct a real-time user electronic portrait, and use the user electronic portrait to match in the system personnel database to find the suitable personnel for the real-time user for recommendation;

[0106] The specific steps to find the suitable personnel for the real-time user for recommendation are as follows:

[0107] S501. Use the uploaded identity information as portrait features to construct a real-time user electronic portrait, use the user electronic portrait to match in the system personnel database, calculate the similarity between the real-time user electronic portrait and each person in the system personnel database through cosine similarity, and finally select the person with the highest similarity as the suitable person for recommendation to the real-time user.

[0108] S600. Construct a visualization platform to visually display the real-time matching process to the user;

[0109] The specific steps to visually display the real-time matching process to the user are as follows:

[0110] S601. Construct a visualization platform to display the identity information of the selected suitable person to the real-time user in the visualization platform, extract the system matching process, display the identity information uploaded by the real-time user himself and the identity information of the suitable person in correspondence, and display the importance classification, sensitivity, and priority of the user's own identity information.

[0111] S700. Real-time users view the recommended personnel and the matching process through the visualization platform. The real-time users review and actively modify the matching process, and the system constructs a personalized matching mechanism according to the modified content to re-match and recommend.

[0112] The specific steps for the system to construct a personalized matching mechanism according to the modified content to re-match and recommend are as follows:

[0113] S701. Real-time users view the recommended personnel and the matching process through the visualization platform. The users subjectively adjust all the proportional coefficients, grading percentages, and thresholds in the matching process. After the adjustment, the system constructs a personalized matching mechanism according to the modified content to re-match and recommend.

[0114] Using the visualization platform to display the matching process enables users to view the matching principle, increasing the interpretability of the matching; allowing users to subjectively adjust the matching process avoids the system overfitting with algorithms and resulting in a lack of personalization in the results, ensuring the differentiation of the system matching and making the matching results more suitable for users.

[0115] Example: Now a matchmaking system is launched in a certain place. The identity information is classified into basic information, family information, occupation information, emotional information, personal hobbies, and future plans; the importance of each type of identity information is calculated by calculating the matching value, regional value, and user control degree, and the high-level information is obtained as: basic information, emotional information, and occupation information; the medium-level information is: family information and future plans; the low-level information is personal hobbies;

[0116] First, a certain user logs in to the system and needs to find a suitable person. Calculate the sensitivity of the real-time user. Taking the user's age as an example to calculate the sensitivity, set the attenuation coefficient to 0.2, the real-time time to 7 days, set a total of four displays, and the display times are 0 days ago, 1 day ago, 3 days ago, and 6 days ago. Calculate the total time attenuation weight to be 2.6687; and there are two interaction records in the four displays, and the interaction conversion rate is obtained as 0.4; finally, the sensitivity of "age" is calculated to be 1.9; normalized to 0.75;

[0117] Among them, "age" belongs to the basic information, which is high-level information. Set the high-level information to be assigned a value of 3, and calculate the priority to be 0.75; calculate the rest of the identity information in the same way; comprehensively consider that "age" needs to be uploaded.

[0118] It is judged that the identity information uploaded by the real-time user includes age and occupation; the suitable personnel are matched and displayed; the user views the matching process through the visualization platform, removes the age, and requests not to use the uploaded age information, and the system re-matches.

[0119] It is obvious to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A method for establishing a user electronic portrait for a matchmaker matching system, characterized in that: The method includes the following steps: S100. Collect the identity information of all users in the matchmaker matching system, extract the main features of each identity information, and classify the identity information using the main features; S200. Extract the influence of each identity information on matching in the system, the behavior data of users in different regions, and the display quantity of all users, and calculate the matching value degree, regional value degree, and user control degree of each identity information respectively; S300. Perform weighted fusion on the matching value degree, regional value degree, and user control degree of each identity information, and calculate the importance degree of each identity information; classify all types of identity information using the importance degree; S400. When performing real-time matching on users in the matchmaker matching system, collect the display times and display time of the user's own different-level identity information in historical network activities, calculate the information sensitivity of the real-time user, and select the identity information of the real-time user to upload to the matchmaker matching system using the information sensitivity; S500. Use the uploaded real-time user identity information to construct a real-time user electronic portrait in the system, and perform matching in the system personnel database using the user electronic portrait to find the suitable personnel for the real-time user for recommendation; S600. Construct a visualization platform to visually display the real-time matching process to users; S700. The real-time user views the recommended personnel and the matching process through the visualization platform, the real-time user reviews and actively modifies the matching process, and the system constructs a personalized matching mechanism according to the modified content for re-matching and recommendation.

2. The method for establishing a user electronic portrait for a matchmaker matching system according to claim 1, wherein: The specific steps of classifying the identity information using the main features in S100 are as follows: S101. Collect the identity information of all users in the matchmaker matching system, construct a structured data set, preprocess the identity information in the data set, judge whether there are missing values in the data in the data set, calculate the average value of the adjacent data at the missing place as the filling value, and fill the missing values; Normalize the preprocessed structured data; S102. For each collected identity information, extract the documents and words in the collected user identity information, and calculate the importance degree of each word. The formula is: In the formula, G(i, d) represents the importance of the i-th word in document d, and n (i，d) represents the number of the i-th word in document d, and w d represents the total number of all words in document d, W represents the total number of documents of the collected identity information, and D i represents the number of documents including the i-th word; Calculate the importance degree of each word in turn, compare the importance degrees of all words in each identity information, and take the word with the largest importance degree as the main feature of the identity information; obtain the main features of each collected identity information in turn; S103. Extract the main features of the identity information in pairs, and calculate the similarity of the extracted main features. The formula is: In the formula, Sim(A, B) represents the calculated similarity of the main features A and B, A and B represent the main features of the two extracted identity information, and A·B represents the dot product of A and B; calculate the similarity of the main features of all identity information in turn, set the similarity threshold, and classify the identity information whose similarity exceeds the similarity threshold into the same type of identity information, and classify all the collected identity information.

3. The method for establishing a user electronic portrait for a matchmaker matching system according to claim 2, wherein: The specific steps of calculating the matching value degree, regional value degree, and user control degree of each identity information respectively in S200 are as follows: S201. Collect the historical matching results and the types of identity information during matching in the system, and calculate the matching value degree of each identity information. The formula is: In the formula, P j represents the matching value degree of the j-th type of identity information calculated, C j represents the number of successful matches when using the j-th type of identity information for matching, Z j represents the total number of times of matching using the j-th type of identity information, and m represents the total number of types of identity information selected during matching; the matching value degrees of m types of identity information are calculated in sequence; S202. Collect the user behavior data of all types of identity information in different regions, and calculate the regional value of each type of identity information in different regions. The formula is: In the formula, Du (j，u) represents the regional value degree of the j-th type of identity information in region u, V (j，u) represents the number of times the j-th type of identity information is viewed by users in region u, Cd (j，u) ’ represents the number of successful matches using the j-th type of identity information in region u; calculate the regional value degrees of m types of identity information in sequence; S203. For the j-th type of identity information, collect the number of times the user is displayed and the number of times the user matches successfully, and calculate the user control degree of the j-th type of identity information. The formula is: In the formula, Yu h represents the user control degree of the j-th type of identity information in user h, M h represents the number of successful matches of the j-th type of identity information in user h, E h represents the number of times the j-th type of identity information is displayed in user h. Calculate the user control degree of the j-th type of identity information in all users in the same way, and then calculate the average value of the user control degree as the final user control degree Yu of the j-th type of identity information j ; Calculate the user control degrees of m types of identity information in sequence, and finally obtain three indicators for each type of identity information, including the matching value degree, the regional value degree, and the user control degree.

4. A method for establishing a user electronic portrait for a matchmaker matching system according to claim 3, characterized in that: The specific steps of using importance to classify all types of identity information in S300 are as follows: S301. Standardize the three indicators of each type of identity information calculated, calculate the information entropy of each indicator using the information entropy formula respectively, and calculate the weight value of each indicator using the information entropy of each indicator. The formula is: In the formula, β a represents the weight of index a, and S a represents the information entropy of index a, and S g represents the information entropy of the g-th index; S302. Use the weights of each calculated indicator to perform weighted fusion on the three indicators to obtain the importance of each type of identity information. The formula is: In j = β1 × P j + β2 × Du (j，u) × β3 × Yu j ; In the formula j represents the importance of the j-th type of identity information, β1 represents the weight of the matching value degree, β2 represents the weight of the regional value degree, and β3 represents the weight of the user control degree; S303. Sort the importance of all identity information in ascending order, set the classification percentage, and perform level classification according to the percentage. The first 20% of the identity information is used as low-level information, the 20%-80% of the identity information is used as medium-level information, and the last 20% of the identity information is used as high-level information.

5. A method for establishing a user electronic portrait for a matchmaker matching system according to claim 4, wherein: The specific steps of using information sensitivity to select the identity information of real-time users and upload it to the matchmaker matching system in S400 are as follows: S401. When performing real-time matching on users in the matchmaker matching system, collect the display times, display durations, and interaction times of the user's own different types of identity information in historical network activities, and calculate the interaction conversion rate of each type of identity information. The formula is: In the formula, J j represents the interaction conversion rate of the j-th type of identity information, and H j represents the number of interactions after the display of the j-th type of identity information in the real-time user history, and sh j represents the number of displays of the j-th type of identity information in the real-time user history. Prevent division by zero; S402. Construct a decay function for the collected display time to calculate the total time decay weight. The formula is as follows: In the formula, Q j represents the total time decay weight of the j-th type of identity information for the real-time user, T now represents the real-time, t r represents the display time at the r-th display, λ represents the decay coefficient, and U represents the total number of displays of the j-th type of identity information in the real-time user's history; S403. Calculate the sensitivity of the real-time user to the j-th type of identity information using the interaction conversion rate and the total decay weight. The formula is as follows: In the formula, Su j represents the sensitivity of the real-time user to the j-th type of identity information. Calculate the sensitivities of the real-time user to all identity information in sequence and normalize them to Su'. S404. Assign values to the classification of identity information. High-level information is 3, medium-level information is 2, and low-level information is 1; calculate the recommendation priority using the normalized sensitivity and level assignment of all identity information. The formula is: Priority = level assignment × (1 - sensitivity Su'); Calculate the priority of all identity information, arrange the priority in ascending order, calculate the difference between adjacent priorities, extract the maximum value of the difference as the node threshold, and upload the identity information with a priority greater than the node threshold to the matchmaker matching system.

6. A method for establishing a user electronic portrait for a matchmaker matching system according to claim 5, characterized in that: The specific steps of finding and recommending suitable personnel for real-time users in S500 are as follows: S501. Use the uploaded identity information as portrait features to construct an electronic portrait of the real-time user, match it in the system personnel database, calculate the similarity between the electronic portrait of the real-time user and each person in the system personnel database through cosine similarity, and finally select the person with the highest similarity as the suitable person to recommend to the real-time user.

7. A method for establishing a user electronic portrait for a matchmaker matching system according to claim 6, characterized in that: The specific steps of visually displaying the real-time matching process to the user in S600 are as follows: S601. Build a visualization platform, display the identity information of the selected suitable person to the real-time user in the visualization platform, extract the system matching process, display the identity information uploaded by the real-time user himself and the identity information of the suitable person correspondingly, and display the importance classification, sensitivity, and priority of the user's own identity information.

8. A method for establishing a user electronic portrait for a matchmaker matching system according to claim 7, characterized in that: The specific steps of the system constructing a personalized matching mechanism to re-match and recommend according to the modified content in S700 are as follows: S701. The real-time user views the recommended person and the matching process through the visualization platform, and the user subjectively adjusts all the proportional coefficients, classification percentages, and thresholds in the matching process. After adjustment, the system constructs a personalized matching mechanism according to the modified content to re-match and recommend.

Images