A matching method based on multi-dimensional weighting

Abstract

The application discloses a kind of matching method and system based on multi-dimensional weighting, it is related to data processing technical field.The method includes: constructing the multi-dimensional index system covering basic image, life habit, character trait and value;Using the utility evaluation operator set independently, the characteristic value is mapped into the satisfaction utility value containing negative value interval, realizes "one vote veto" mechanism;Based on two-way weighted logic, in combination with each dimension weight and bargaining right coefficient, calculate preliminary comprehensive fit degree;Introduce preference scarcity correction operator, based on the global integral of subject preference wide degree, the score is punished and adjusted, and the final ranking result is output.The application solves the problems of traditional matching dimension single, cannot reflect individual heterogeneity preference and score distortion, etc., and improves the accuracy of two-way willingness alignment.The application can be widely applied to marriage, love and friendship, talent recruitment and resource docking and other various two-way matching scenarios.

Description

Technical Field

[0001] This invention relates to the fields of data processing and social computing technology, specifically to a bidirectional matching method and system based on multidimensional dynamic weighting. Background Technology

[0002] With the improvement of social and economic levels and the development of cultural diversification, the mate selection demands of single people of marriageable age have shifted from the traditional "single economic dimension" to "all-dimensional compatibility". At present, the number of single people of marriageable age in China has exceeded 200 million. The decline in the birth rate and the aggravation of aging caused by late marriage of older youth have become major issues affecting the sustainable development of society. At present, existing marriage matching technologies are mainly divided into the following two categories: (1) Traditional offline matching mode: mainly relying on introductions from acquaintances or offline blind dates. Its current technology is characterized by manual screening using the existing scope of acquaintance networks. (2) Online matching mode based on Internet platforms: using mobile applications or social networking sites to automatically filter and recommend based on basic tags such as residence, education, age, and occupation filled in by users.

[0003] However, the above-mentioned background technologies have the following main drawbacks in actual operation: (1) The matching dimension is single and superficial. Existing matching mechanisms mostly rely on a few static and explicit objective attributes, which are difficult to touch on the more complex personality traits, emotional needs, interest structures and values ​​of individuals. (2) There is a lack of a self-setting mechanism for weights. Traditional algorithms often preset fixed matching weights, ignoring the significant heterogeneity of individuals in mate selection preferences. Different subjects pay great attention to the same feature (such as height or education), and a unified weighting logic is difficult to achieve accurate recommendations. (3) The accuracy of two-way willingness alignment is low. Due to the lack of quantitative measurement of two-way "bargaining power" and "satisfaction mapping", the recommendation results often present an awkward situation of "suitable conditions" but "incorrect subjective feeling", resulting in low matching efficiency and high search costs. Summary of the Invention

[0004] To address the problems in existing marriage and dating matching technologies, such as the single matching dimension, inability to reflect individual heterogeneous preferences, low accuracy of two-way intention alignment, and distortion of matching scores due to preference redundancy, this invention provides a two-way matching method based on multi-dimensional weighting, aiming to achieve more scientific and refined individual matching decisions.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: A bidirectional matching method based on multidimensional weighting is characterized by the following steps: (1) Constructing a multidimensional attribute indicator system: covering feature values ​​of multiple dimensions such as basic profile, living habits, personality traits and values. (2) Defining an individual utility mapping function: the matching subject independently sets a satisfaction function for different indicators, and transforms the feature values ​​of the opposite sex in each dimension into satisfaction values ​​in the range of [-1, 1]. (3) Implementing differentiated weight allocation: supporting the subject to set the original score of subjective preference weight according to the importance of each dimension, and obtaining the weight coefficient of each indicator through normalization processing. (4) Calculating the comprehensive matching score: based on the bidirectional weighting logic, a bargaining power coefficient is introduced to measure the comprehensive compatibility of individual pairs. Introducing preference scarcity correction: by constructing a correction operator based on exponential decay, the subject's "pickiness" is quantified. If the subject's preference is too broad (i.e., the integral value of the satisfaction function is too high), a punitive correction is made to its final matching score.

[0006] Compared with existing technologies, the beneficial effects of this invention are as follows: (1) It improves the accuracy and depth of matching: by introducing full-dimensional feature values ​​that cover deep features such as values, it effectively solves the problem of sensory misalignment of "conditions are suitable, but feelings are not right". (2) It realizes highly personalized mate selection modeling: both weight parameters and satisfaction functions can be set by the subject, fully respecting the heterogeneity of individual mate selection preferences. (3) It enhances the authenticity of matching scores: the preference scarcity correction mechanism effectively suppresses the redundant noise generated by "generalized preferences", making the recommendation results more in line with the supply and demand law of the real market. (4) It has broad application value in various scenarios: this model is not only applicable to the field of marriage and love, but also has significant application value in scenarios involving the intersection of multiple dimensions of conditions, such as talent recruitment, teacher-student matching, and business partner screening. Attached Figure Description

[0007] Figure 1 This is a schematic diagram illustrating the scenario of the technical problem to be solved by the present invention.

[0008] Figure 2 This is a schematic diagram of the mapping function under the age item in this invention, where (a) is the mapping function of female j to male, and (b) is the mapping function of male i to female.

[0009] Figure 3 This is a schematic diagram of the satisfaction function of the distance dimension in this invention, where: (a) is the satisfaction function of the household registration distance for male i and female j; (b) is the satisfaction function of the current work / study distance for male i and female j.

[0010] Figure 4 This is a schematic diagram of the rational / emotional satisfaction functions of male i and female j in this invention.

[0011] Figure 5These are schematic diagrams of two typical forms of the single-peak satisfaction function in this invention, where (a) is a triangular single-peak satisfaction function and (b) is a single-peak curve satisfaction function. Detailed Implementation

[0012] 1. Sample definition and attribute dimension construction In this embodiment, the matching indicators cover the following four core dimensions (a total of n items): (1) Basic profile: including birth year, appearance, height, weight, education, place of origin and current location, etc. (2) Lifestyle habits: including duration of dating experience, cooking experience, exercise time, cleanliness, gaming time, household chores allocation ratio, consumption ratio and smoking and drinking habits, etc. (3) Personality and psychological traits: including fast / slow personality, delicate / rough tendency, emotional / rational degree, preference for being alone / living in a group, sensitive / sudden feeling type and enterprising / content type, etc. (4) Values ​​and outlook on life: including savings ratio, family pattern preference, exploratory / conservative orientation, dominance / cooperation relationship pattern and pet preference, etc.

[0013] 2. Two-way matching mechanism and formula calculation: Let M (male) and F (female) represent the sets of male and female samples, respectively. Define the matching degree of any pair of individuals of opposite sexes (i, j) (where i ∈ M, j ∈ F) as Score. i,j This variable is used to measure the degree of attribute matching (mutual satisfaction) between samples, and its calculation logic is as follows:

[0014] In this equation, the first half on the right side represents male i's satisfaction with female j; the second half represents female j's satisfaction with male i. α=1 indicates that both men and women have completely equal "bargaining power" in the relationship, but this can be dynamically adjusted based on actual circumstances. k is the index of the evaluation dimensions (e.g., age, appearance, height, weight, education, etc., totaling n items); W i, k W j,k : represent the weighting coefficients of male i and female j for the k-th indicator, respectively; f i,k f j,k : Represent the utility evaluation operators (i.e., mapping functions / satisfaction functions for the requirements of the opposite sex) set by male i and female j for the k-th indicator, respectively. The maximum value of these two functions is 1; C i,k C j,k : represent the characteristic values ​​of male i and female j on the k-th indicator, respectively. W i,k W j,k Represented as:

[0015] Among them, w i,k w j,kThe original score of the subjective preference weight set by male i and female j for the k-th indicator, whose value ranges from [0, 100]. And ϵ is the regularization term, whose value is a极小 positive real number (such as 10 -6 ), aiming to improve the numerical stability of the model and prevent the denominator from being singular during the calculation process. W i,k and W j,k have values ∈ [0, 1). Score i,j . has a value ∈ [0, 1).

[0016] 3. Calculation method for each item 3.1 Quantization processing of multi-source heterogeneous data and satisfaction mapping For different types of indicators, the system constructs utility evaluation operators f i,k and f j,k using different quantization logics.

[0017] In a specific embodiment of the present invention, taking the birth year indicator as an example. As Figure 2 (a) shows, let the eigenvalue C i,1 of male sample i be 1995, and the eigenvalue C j,1 of female sample j be 1999. Female sample j independently sets the mapping function f j,1 : (1) Complete matching interval: When the eigenvalue of male sample i is within the core expected interval preset by female sample j, the utility evaluation operator outputs f = 1.0, indicating complete satisfaction; (2) Marginal utility interval: When the eigenvalue is outside the core interval but does not exceed the maximum tolerance limit, 0 < f < 1.0 is output, indicating partial satisfaction and decreasing utility; (3) One-vote veto mechanism: If the eigenvalue exceeds the tolerance limit (such as a too large age difference), a penalty mechanism of f < 0 is triggered, that is, the negative utility of this dimension will directly lead to the failure of the comprehensive matching determination of this pair of samples.

[0018] Similarly, as Figure 2 (b) shows, male sample i can also set its corresponding satisfaction mapping function f i,1 . Through the above bidirectional mapping logic, the system can accurately quantify the attribute matching degree between individuals.

[0019] 3.2 Spatial distance measurement In another specific embodiment of the present invention, taking the spatial distance indicator as an example. As Figure 3 (a) and 3(b) show the satisfaction functions of sample male i and female j for their native places and current workplaces. The system first obtains the location information of the native places and current workplaces of male sample i and female sample j. Using the geographic information system, the above addresses are converted into longitude and latitude coordinates, and the great circle distance d between the two on the earth's surface is calculated based on the Haversine formula.

[0020] Based on the measured spatial distance d, the subject independently sets its distance satisfaction mapping function: (1) Geographical fit interval: When the distance d between the two parties is within the ideal range preset by the subject (such as a native place distance of 80 km or a working distance of 40 km), the utility evaluation operator outputs f = 1.0, representing a completely satisfied state under this spatial constraint. (2) Distance utility attenuation interval: When the distance exceeds the ideal range but is still within the acceptable migration or commuting limit, 0 < f < 1.0 is output. For example, as the distance increases, the satisfaction of female sample j with the household registration distance of male sample i may decay to 0.2, while the satisfaction with the current working distance remains at 0.6. (3) Spatial remote veto mechanism: If the distance d exceeds the maximum tolerance limit set by the subject (such as too far away), the mapping function triggers a penalty value of f < 0. The negative utility of this dimension represents extremely high geographical fit costs or psychological barriers to marrying far away, directly leading to the failure of the matching determination for this individual pair.

[0021] Through this spatial distance mapping logic, the system can convert abstract geographical displacements into standardized utility values, thereby accurately measuring the compatibility of both parties in terms of geographical attributes.

[0022] In another specific embodiment of the present invention, taking the spatial distance index as an example. As Figure 3 (a) and 3(b) show the satisfaction functions of sample male i and female j with respect to their native places and current workplaces. The system first obtains the location information of the native places and current workplaces of male sample i and female sample j. Using the geographic information system, the above addresses are converted into longitude and latitude coordinates, and based on the Haversine formula, the great circle distance d between the two parties on the earth's surface is calculated.

[0023] Based on the measured spatial distance d, the subject independently sets its distance satisfaction mapping function: (1) Geographical fit interval: When the distance d between the two parties is within the ideal range preset by the subject (such as a native place distance of 80 km or a working distance of 40 km), the utility evaluation operator outputs f = 1.0, representing a completely satisfied state under this spatial constraint. (2) Distance utility attenuation interval: When the distance exceeds the ideal range but is still within the acceptable migration or commuting limit, 0 < f < 1.0 is output. For example, as the distance increases, the satisfaction of female sample j with the household registration distance of male sample i may decay to 0.2, while the satisfaction with the current working distance remains at 0.6. (3) Spatial remote veto mechanism: If the distance d exceeds the maximum tolerance limit set by the subject (such as too far away), the mapping function triggers a penalty value of f < 0. The negative utility of this dimension represents extremely high geographical fit costs or psychological barriers to marrying far away, directly leading to the failure of the matching determination for this individual pair.

[0024] Through this spatial distance mapping logic, the system can convert abstract geographical displacements into standardized utility values, thereby accurately measuring the compatibility of both parties in terms of geographical attributes.

[0025] 3.3 Bias Variables (Taking rational / emotional traits as an example) For psychological tendency indicators with continuous distribution characteristics, the system measures them through axial positioning and expectation mapping. The value range is defined as [0,1], where 0 represents complete rationality, 0.5 represents a balance between rationality and emotion, and 1 represents complete emotion.

[0026] The subject constructs a mapping relationship based on its own attribute positioning and expectations of the opposite sex: (1) Attribute positioning mapping: If male sample i's self-assessment positioning in the rational-emotional dimension is 0.3 (leaning towards rationality), while female sample j's attribute positioning is 0.6 (leaning towards emotion). (2) Personality compatibility measurement: such as Figure 4 As shown, female sample j autonomously sets its satisfaction mapping function for the opposite sex's personality. When male sample i's positioning of 0.3 falls into the high fit interval set by female sample j, the calculated satisfaction score is 0.8; similarly, the calculated satisfaction of male sample i with female sample j is 0.9. (3) Extreme mutual exclusion avoidance: If the personality positioning of both parties is on the extreme opposite axis (such as 0 and 1) and exceeds the maximum preference tolerance set by the subject, then a negative utility value is output, indicating that there may be serious compatibility obstacles between the two parties in long-term interaction.

[0027] The matching degree of any pair of opposite-sex individuals (i, j) (where i ∈ M, j ∈ F) can be calculated using the above method as Score. ij A high matching value reflects a high degree of psychological expectation achievement and consistency of value recognition between individuals; conversely, a low matching value indicates significant mutual exclusion in attribute characteristics between the two parties.

[0028] By hierarchically processing ordinal variables and axially mapping symmetric variables, this invention can transform qualitative sociological characteristics into standardized utility values, thereby significantly improving the scientific rigor and real-world applicability of the matching results.

[0029] 4. Picking function 4.1 Full-featured version: Two-way weighted preference and scarcity correction model In the above model, if male i's satisfaction ratings across multiple attributes are all close to 1 (i.e., exhibiting "undifferentiated preference"), it will lead to a large number of female individuals assigning him extremely high matching scores. This clearly deviates from the principle of "preference scarcity" in the real marriage and dating market. Therefore, this study introduces a preference scarcity correction factor f. i with f j Define f i f j Let be the measure of the integral of the satisfaction function of male i and female j with respect to all potential traits, used to characterize the "pickiness" of individual preferences:

[0030] in, and ... a 1 represents the basic adjustment coefficient, used to control the corrected reference strength. Parameter a 2 represents the sensitivity coefficient. The larger this value, the more severely the model penalizes the tendency to be "uncritical" (larger integral value), and the more sensitively it can capture the decline in matching quality caused by the broadening of subject preferences. The corrected comprehensive matching score is defined as: Right now: .

[0031] Through this corrective logic, if male i The preference function is close to 1 across the entire range (i.e., lacks selectivity), and its correction factor... f i It will approach 0, thus significantly reducing its final matching score. This mechanism effectively simulates the logic of "the decline in matching effectiveness due to a lack of clear preferences" in reality, making the model more empirically explanatory.

[0032] When a2=0, it is the simplified version of the picky function: .

Claims

1. A matching method based on multidimensional weighting, characterized in that, The steps include: Step S100: Construct a multi-dimensional attribute index system and obtain the feature values ​​C of the first subject sample i and the second subject sample j in n dimensions. i,k With C j,k Step S200: Construct the utility evaluation mapping operator f i,k with f j,k The mapping operator is set autonomously by the subject according to personalized preferences, and is used to transform the feature values ​​of the opposite-sex subject into utility values ​​representing satisfaction; Step S300: Set the original weight scores of each dimension index and perform normalization processing to obtain the weight coefficient W. i,k With W j,k Step S400: Based on bidirectional weighted logic, calculate the attribute matching degree between the first subject sample i and the second subject sample j to obtain a preliminary comprehensive fit score. i,j Step S500: Introduce a preference scarcity correction operator to adjust the score based on the subject's preference breadth. i,j Perform punitive adjustments, output the final matching score, and rank and recommend the results.

2. The matching method based on multidimensional weighting according to claim 1, characterized in that: The output value range of the utility evaluation mapping operator in step S200 includes a negative value range. When the feature value exceeds the tolerance limit set by the subject, a negative utility value is output to trigger a veto mechanism.

3. The matching method based on multidimensional weighting according to claim 1, characterized in that: The multidimensional attribute indicator system in step S100 includes continuous numerical variables, ordinal variables, spatial distance variables, and binary or multivariate tendency variables.

4. The matching method based on multidimensional weighting according to claim 3, characterized in that: For the spatial distance variable, the system uses the Havesing formula to calculate the great circle distance d between samples based on latitude and longitude, and substitutes it into the corresponding utility evaluation mapping operator to calculate the satisfaction.

5. The matching method based on multidimensional weighting according to claim 3, characterized in that: For the aforementioned tendency variable, the system calculates the compatibility of the two parties in terms of personality traits or values ​​based on the attribute positioning coordinates of the subject on a preset axis and the expected distribution function.

6. The matching method based on multidimensional weighting according to claim 1, characterized in that: In step S400, the preliminary comprehensive compatibility score is obtained. ij The calculation formula is: Where α is the bargaining power coefficient, and its value ranges from [0, 1].

7. The matching method based on multidimensional weighting according to claim 1, characterized in that: The calculation logic of the preference scarcity correction operator in step S500 is as follows: First, the utility evaluation mapping operator set by the subject is integrated over the entire domain to measure its preference breadth. Then, the correction factor is calculated through the exponential decay function to suppress the artificially high matching score caused by excessive lack of pickiness.

8. A matching method based on multidimensional weighting according to any one of claims 1 to 7, characterized in that: This method can be applied not only to dating and matchmaking, but also to talent recruitment, screening of entrepreneurial partners, teacher-student pairing, and business resource matching.

Images