A method for querying engineering cost data using electronic information

Abstract

The present invention belongs to the field of data query technology and provides a method for querying engineering cost data using electronic information technology. The method obtains the preliminary importance of each user's parameter by analyzing the differences in the browsing time of different clicked materials and the numerical differences of the same parameter. The preliminary importance is revised by analyzing the degree of parameter differences between the materials clicked and the materials collected by the user during the browsing process. Then, based on the revised importance of all parameters of all users, all users are divided into different job clusters. The similarity between the target user and each job cluster is analyzed to obtain the target user's reference users. Finally, materials are recommended for the target cluster based on the reference users. The present invention distinguishes users of different jobs by revising the importance, determines the job cluster to which the target user belongs, and filters the sample data relied on for collaborative filtering, thereby improving the accuracy of the recommendation results.

Description

Technical Field

[0001] The present invention relates to the technical field of data query, and in particular to a method for querying engineering cost data using electronic information. Background Art

[0002] Project Costs (PC) refers to the construction costs that constitute the estimated or actual expenditures of a project during the construction period, including various expenditures incurred throughout the entire process from project initiation to completion and delivery. It covers land costs, material costs, labor costs, machinery usage fees, management fees, profits, taxes and other related expenses. It is not only an important part of engineering economic activities, but also an important basis for controlling project investment, evaluating economic benefits and formulating construction plans.

[0003] To ensure accurate and reasonable project costs, it is necessary to query project cost data. This query can obtain the latest material prices, labor costs, machinery costs, and quota data, providing a reliable basis for budget preparation, cost control, and quotation. In the process of querying project cost data, in order to improve efficiency, accuracy, and traceability, electronic information technology is currently often used to query project cost data. Compared with traditional query methods, electronic information technology realizes real-time query and dynamic update of multi-dimensional information such as material prices, quota data, and construction technology through digital platforms and databases. It can quickly filter, classify, and compare data to meet the personalized needs of projects.

[0004] However, when users query engineering cost data through the engineering cost data platform, the engineering cost data platform will match the keywords searched by the users with the names of the materials to obtain basic recommended materials, and then use the collaborative filtering algorithm to continue to recommend to the users based on the interaction between the users and the basic recommended materials. However, the job responsibilities of users who use the engineering cost data platform to query engineering cost data are different, resulting in different material parameters that users pay attention to. Therefore, when using the collaborative filtering algorithm to recommend to users, inaccurate recommendation results will occur, resulting in reduced work efficiency. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides a method for querying engineering cost data using electronic information.

[0006] According to the present invention, a method for querying engineering cost data using electronic information is provided, the method comprising:

[0007] Obtain the parameters of each user's clicked materials and browsing time, as well as the collection of materials;

[0008] For each user, analyze the differences in browsing time of different click materials and the differences in the values ​​of the same parameter to obtain the preliminary importance of the parameter for each user;

[0009] Based on any one of the parameters of each user, analyzing the degree of difference between the clicked materials and the favorited materials during the browsing process of the user, revising the preliminary importance, and obtaining the revised importance of the parameter of each user;

[0010] Based on the modified importance of all the parameters of all the users, dividing all the users into different job clusters;

[0011] Analyze the similarity between the target user and each of the job clusters to obtain reference users of the target user;

[0012] Recommend materials to the target cluster based on the reference user.

[0013] In some embodiments of the present invention, based on any one of the parameters of each user, the difference between the clicked materials and the favorited materials during the user's browsing process is analyzed, and the preliminary importance is revised to obtain the revised importance of the parameter for each user, including:

[0014] Based on any one of the parameters of each user, analyzing the difference in parameter values ​​between the clicked material and the favorited material during the user's browsing process, and obtaining the expected skewness of the parameter in the clicked material for each user;

[0015] Clustering all the click materials of each user according to the browsing time of the click materials to obtain a click material cluster of each user;

[0016] For any parameter of each user, analyzing the differences in the expected skewness corresponding to different click material clusters to which the parameter belongs, and obtaining an important index of the parameter of each user;

[0017] The preliminary importance is revised according to the importance index to obtain the revised importance of the parameter of each user.

[0018] In some embodiments of the present invention, based on any one of the parameters of each user, analyzing the difference in parameter values ​​between the clicked material and the favorited material during the user's browsing process to obtain each user's expected skewness of the parameter in the clicked material includes:

[0019] Based on any one of the parameters of each user, calculating the numerical mean of the parameter in all the user's collection materials;

[0020] The difference between the numerical value of the parameter in the user's click material and the numerical mean is calculated to obtain the expected skewness of each user for the parameter in the click material.

[0021] In some embodiments of the present invention, all click materials of each user are clustered according to the browsing time of the click materials to obtain a click material cluster of each user, including:

[0022] According to the browsing time of the click materials, DBSCAN clustering is performed on all the click materials of each user to obtain a click material cluster of each user, wherein each click material cluster represents the time required for the user to browse to a certain parameter.

[0023] In some embodiments of the present invention, after obtaining the click material cluster of each user, the following steps are further included:

[0024] According to the browsing time, the click material clusters are sorted in ascending order to obtain a click material cluster sequence for each user.

[0025] In some embodiments of the present invention, for any parameter of each user, analyzing the differences in the expected skewness corresponding to different click material clusters to which the parameter belongs, and obtaining an important index of the parameter of each user includes:

[0026] For any parameter of each user, based on the click material cluster sequence, the difference in the discrete degrees of the expected skewness corresponding to different click material clusters to which the parameter belongs is analyzed, and the possibility of each click material cluster of each user being used as the boundary of the parameter is obtained;

[0027] According to the demarcation possibility and in combination with the position ordinal number of the click material cluster in the click material cluster sequence, the importance index of the parameter of each user is obtained.

[0028] In some embodiments of the present invention, for any parameter of each user, based on the click material cluster sequence, the difference in the discrete degrees of the expected skewness corresponding to different click material clusters to which the parameter belongs is analyzed to obtain the possibility of each click material cluster of each user being used as the boundary of the parameter, including:

[0029] For any one of the parameters of each user, according to the position ordinal number of the click material cluster in the click material cluster sequence, the difference between the discrete degrees of the expected skewness corresponding to all the click material clusters before the position ordinal number and the discrete degrees of the expected skewness corresponding to all the click material clusters after the position ordinal number are analyzed, and combined with the mean of the corresponding expected skewness in all the click material clusters, the possibility of each click material cluster of each user as the demarcation of the parameter is obtained.

[0030] In some embodiments of the present invention, all users are divided into different job clusters based on the modified importance of all parameters of all users, including:

[0031] According to the modified importance, all the parameters of each user are clustered using DBSCAN to obtain a number of parameter clusters corresponding to each user;

[0032] Analyze the differences between the modified importance of all parameters in all parameter clusters of any two users to obtain the degree of job difference between the two users;

[0033] According to the degree of job differences, DBSCAN clustering is performed on all users to obtain several job clusters.

[0034] In some embodiments of the present invention, analyzing the similarity between the target user and each of the job clusters to obtain reference users for the target user includes:

[0035] Determine whether the target user collects the material;

[0036] If not, the initial importance of the parameter is used as a recommended reference index;

[0037] If yes, the revised importance of the parameter is used as a recommended reference index;

[0038] Calculating the mean of the modified importance of the parameters for all users in the job cluster, combining it with the recommendation reference index, traversing all the parameters of the target user, and obtaining the similarity between the target user and each of the job clusters;

[0039] All users in the job cluster that have the highest similarity to the target user are recorded as reference users of the target user.

[0040] In some embodiments of the present invention, recommending materials for the target cluster based on the reference user includes:

[0041] Scoring each material browsed by the reference user to obtain a reference score of each material by the reference user;

[0042] According to the reference score, a collaborative filtering algorithm is used to obtain the score of each material given by the target user, thereby obtaining a target score of each material given by the target user;

[0043] Materials are recommended to the target user in descending order of the target scores.

[0044] As can be seen from the above embodiments, the method for querying engineering cost data using electronic information provided by the embodiments of the present invention has the following beneficial effects:

[0045] The present invention obtains the preliminary importance of each user's parameter by analyzing the differences in the browsing time of users for different clicked materials and the numerical differences of the same parameter; since the browsing time has little correlation with unimportant parameters during the calculation of the preliminary importance, the preliminary importance is corrected according to the degree of difference in parameters between the clicked materials and the favorite materials during the user's browsing process to obtain the corrected importance; as for the problem that users pay attention to different parameters due to different positions, the present invention divides all users into different position clusters through the corrected importance of parameters, analyzes the similarity between the target user and each position cluster to obtain the reference user of the target user, and then screens the sample data relied on by collaborative filtering, which can effectively improve the accuracy of the recommendation results.

[0046] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 A schematic diagram of the basic flow of a method for querying engineering cost data using electronic information provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0049] To further illustrate the technical means and effectiveness of the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail a method for querying construction cost data using electronic information technology, including its specific implementation, structure, features, and effectiveness. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. Terms such as "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a circuit structure, article, or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such article or device. In the absence of further limitations, the phrase "comprising a ..." to define an element does not preclude the presence of other identical elements in the article or device comprising the element.

[0051] A method for querying engineering cost data using electronic information provided by this embodiment will be described in detail below with reference to the accompanying drawings.

[0052] See also Figure 1 , which shows the basic process of a method for querying engineering cost data using electronic information provided by an embodiment of the present invention.

[0053] like Figure 1 As shown, an embodiment of the present invention provides a method for querying engineering cost data using electronic information, which specifically includes the following steps:

[0054] S100: Obtain the parameters of each user's clicked material and browsing time, and obtain the collection material.

[0055] Based on the keywords entered by users on the Engineering Cost Data Platform, materials containing the keywords are retrieved from the platform and recorded as basic recommended materials. Since basic recommended materials with high click-through rates often meet the needs of most users, in order to more quickly and accurately determine user preferences for these materials, basic recommended materials are sorted from highest to lowest click-through rate and displayed accordingly.

[0056] At this point, users will click on materials that may meet their criteria to view their details. Therefore, all parameters of all materials clicked by each user since the basic recommended materials were obtained, as well as the viewing time of each clicked material, are obtained. Users may also collect materials that meet their criteria. Therefore, the user's collected materials and all parameters contained in each collected material are also obtained.

[0057] S200: For each user, analyze the differences in browsing time of different click materials and the differences in the values ​​of the same parameter to obtain the preliminary importance of the parameter for each user.

[0058] When users click on basic recommended materials to view them, the engineering cost data platform mainly obtains the parameters of the materials that users are interested in from the browsing data of basic recommended materials, and then makes more comprehensive and accurate recommendations to users based on the parameters of the materials that users are interested in. Because when users browse the details page of basic recommended materials, they often view the important parameters first and then the secondary parameters in the order of importance. When the important parameters do not meet the requirements, the user will close the material details page and continue to select. Therefore, when the browsing time of a certain material's details page is short, the important parameters of the material may not meet the user's needs; when the browsing time of a certain material's details page is long, the important parameters of the material are more likely to meet the user's needs; that is, for any parameter browsed by each user, the greater the difference in the parameter between the material with a longer browsing time and the material with a shorter browsing time, the more likely the parameter is to be an important parameter.

[0059] Based on the above analysis, in the embodiment of the present invention, for each user, the differences in browsing time of different click materials and the differences in the values ​​of the same parameter are analyzed to obtain the preliminary importance of each user's parameter. User's The preliminary importance of each parameter is calculated as:

[0060]

[0061] Where, Indicates the User's The initial importance of the parameters; Indicates the All click materials of users include The number of click materials of each parameter; Indicates the User's The browsing time of each clicked material; Indicates the User's The browsing time of each clicked material; Indicates the User's Click material The value of the parameter; Indicates the User's Click material The value of the parameter; represents the linear normalization function.

[0062] In the above formula, the absolute value of the difference in the user's browsing time for different click materials and the absolute value of the numerical difference of the same parameter in different click materials are calculated. When the absolute values ​​of the two differences are large, it means that the browsing time of the parameter in different click materials is relatively different and the numerical difference is relatively large, which means that the parameter is likely to be an important parameter.

[0063] S300: Based on any parameter of each user, analyze the difference between the materials clicked and the materials collected by the user during the browsing process, revise the preliminary importance, and obtain the revised importance of the parameter of each user.

[0064] When users do not express clear preferences, only the preliminary importance obtained in the above-mentioned way can be used to roughly judge the parameters that users are concerned about. Since users will first view important parameters, they are likely to be viewed and judged. However, some unimportant parameters may not be viewed or viewed very often from beginning to end. Therefore, the correlation between browsing time and unimportant parameters is not high, that is, the preliminary importance may be biased for unimportant parameters, so it is necessary to revise the preliminary importance. When users collect materials that meet the selection criteria, the preliminary importance can be revised based on the difference in parameters between the clicked material and the collected material.

[0065] Based on the above analysis, in an embodiment of the present invention, based on any parameter of each user, the difference between the materials clicked and the materials saved by the user during the browsing process is analyzed, and the preliminary importance is revised to obtain the revised importance of each user's parameter. Further including:

[0066] First, based on any parameter of each user, we analyze the difference in parameter values ​​between the materials clicked and the materials saved during browsing to obtain each user's expected skewness for the parameter in the clicked materials. A specific implementation involves calculating the mean value of the parameter in all of the user's saved materials based on any parameter of each user; then calculating the difference between the parameter value in the user's clicked materials and the mean value to obtain each user's expected skewness for the parameter in the clicked materials.

[0067] Since users review parameters in descending order of importance, for any given material, if a parameter is relatively important to the user and the expected skewness for that parameter is large, it will be initially excluded, resulting in a shorter viewing time for that material. If the expected skewness for that parameter is small, the parameter meets the user's requirements, and the user will continue to review less important parameters, resulting in a longer viewing time for that material. For any given material, if a parameter is moderately important to the user and the expected skewness for that parameter is large, it will be excluded at a relatively mid-viewing time. Before this time, the user will filter materials based on more important parameters and has not yet determined whether the parameter meets the requirements, resulting in a more dispersed distribution of the expected skewness for that parameter. After this time, the parameter has already been determined to meet the requirements, resulting in a smaller expected skewness. Therefore, all clicked materials for each user are clustered according to their viewing time, yielding a cluster of clicked materials for each user. The specific implementation method is: according to the browsing time of the click material, DBSCAN clustering is performed on all the click materials of each user to obtain the click material cluster of each user, wherein each click material cluster represents the time required for the user to browse to a certain parameter.

[0068] The click material clusters of each user are obtained, and then the process further includes: sorting the click material clusters from small to large according to the browsing time to obtain the click material cluster sequence of each user.

[0069] Then, for any parameter of each user, the differences in expected skewness corresponding to different click material clusters to which the parameter belongs are analyzed to obtain the importance index of each user's parameter. For any parameter of each user, if the discreteness of the expected skewness of the parameter in all click material clusters before a certain click material cluster is stronger, and the discreteness of the expected skewness of the parameter in all click material clusters after the click material cluster is significantly reduced, and the expected skewness of the click material cluster is significantly higher, then the click material cluster is more likely to be the click material cluster at the time position required to browse to the parameter. Therefore, the specific implementation method is as follows:

[0070] For any parameter of each user, based on the click material cluster sequence, the difference in the discrete degree of expected skewness corresponding to different click material clusters to which the parameter belongs is analyzed, and the demarcation possibility of each click material cluster of each user as a parameter is obtained. More specifically: for any parameter of each user, according to the position ordinal number of the click material cluster in the click material cluster sequence, the difference between the discrete degree of expected skewness corresponding to all click material clusters before the position ordinal number and the discrete degree of expected skewness corresponding to all click material clusters after the position ordinal number is analyzed, and the mean of the corresponding expected skewness in all click material clusters is combined to obtain the demarcation possibility of each click material cluster of each user as a parameter. Construct the first Click material clusters as the first User's The calculation formula for the demarcation probability of the parameters is:

[0071]

[0072] Where, Indicates the Click material clusters as the first User's The demarcation possibility of parameters; Indicates the The first of all click materials in the click material cluster the mean of the expected skewness of the parameters; Indicates the The first of all click materials in the click material cluster the variance of the expected skewness of the parameters; Indicates the The first of all click materials in the click material cluster the variance of the expected skewness of the parameters; Indicates the The number of click material clusters of each user; Indicates the first Click material clusters before Click material clusters; Indicates the first Click material clusters; represents the linear normalization function.

[0073] For the The first of all click materials in the click material cluster The mean of the expected skewness of the parameters The larger the value, the more The judgment of the parameters completes the selection of the click material, indicating that the click material cluster is the first User's The greater the possibility of demarcation of the parameters; , indicating the first The degree of dispersion of the expected skewness of all click material clusters before the first click material cluster is the same as that of the first click material cluster. The difference between the discrete degrees of the expected skewness corresponding to all click material clusters after the first click material cluster. The larger the value, the greater the Click material clusters before The expected skewness corresponding to the parameter is more discrete, and the After clicking on the material cluster The expected skewness corresponding to the parameter is less discrete, indicating that the Click material clusters as the first User's The greater the possibility of demarcation of the parameters.

[0074] For any parameter, there is only one demarcation click material cluster corresponding to the parameter, so the demarcation probability of the real demarcation click material cluster is greater. If the browsing time of all click materials in the demarcation click material cluster is longer, it means that the parameter is less important. Therefore, according to the demarcation probability, combined with the position ordinal number of the click material cluster in the click material cluster sequence, the importance index of each user's parameter is obtained. User's The calculation formula for the important index of each parameter is:

[0075]

[0076] Where, Indicates the User's The important index of the parameter; Indicates the The number of click material clusters of each user; Indicates the Click material clusters as the first User's The demarcation possibility of parameters; Indicates the first The ordinal number of the click material cluster; Represents the activation function, which is used to Convert to probability values ​​between [0,1] and ensure that the sum of the probabilities of all elements is 1; represents the linear normalization function.

[0077] By normalized boundary probability The ordinal numbers of the click material clusters are weighted, that is, the larger the ordinal number of the click material cluster, the greater the corresponding demarcation possibility, which means that the parameter is less important, and the importance index of the parameter for the user is smaller.

[0078] Finally, since the preliminary importance is relatively accurate in calculating important parameters, but needs to be corrected for unimportant parameters, the importance index is used to correct the preliminary importance to improve the accuracy of parameter importance judgment. Therefore, according to the importance index, the preliminary importance is corrected to obtain the corrected importance of each user's parameter. User's The calculation formula for the modified importance of a parameter is:

[0079]

[0080] Where, Indicates the User's The revised importance of each parameter; Indicates the User's The initial importance of the parameters; Indicates the User's The importance index of the parameter.

[0081] When the importance of parameter correction is high, that is, If it is relatively high, it means that the parameter is more important, and the initial importance shall prevail. If it is smaller, it should be Based on adjustments.

[0082] S400: Based on the modified importance of all parameters of all users, all users are divided into different job clusters.

[0083] Since different positions have different focuses, for example, cost estimators focus on standard unit prices and historical costs; designers focus on material specifications and brand recommendations; construction managers focus on material suppliers and delivery cycles. Therefore, for any position, the position pays much more attention to certain parameters than other parameters, and the parameters that are focused on vary greatly between positions.

[0084] Based on the above analysis, in an embodiment of the present invention, all users are divided into different job clusters based on the modified importance of all parameters of all users. Further including:

[0085] First, all parameters of each user are clustered using DBSCAN according to the corrected importance, and several parameter clusters corresponding to each user are obtained.

[0086] Then, each parameter cluster represents a set of parameters that the user has a certain degree of attention to. For any two users, the more similar the parameter types in the parameter clusters of the two users are, and the more similar the modified importance of all parameters in the parameter clusters are, the more likely the two users are to belong to the same position, and the smaller the degree of difference between the positions of the two users is. Therefore, the difference between the modified importance of all parameters in all parameter clusters of any two users is analyzed to obtain the degree of difference between the positions of any two users. Construct the first User and The calculation formula for the degree of job difference of a user is:

[0087]

[0088] Where, Indicates the User and The degree of job differentiation among users; Indicates the User's parameter clusters and the User's The number of identical parameters in each parameter cluster; Indicates the User's The average value of the corrected importance of all parameters in a parameter cluster; Indicates the User's The average value of the corrected importance of all parameters in a parameter cluster; Indicates the The number of parameter clusters for each user; Indicates the The number of parameter clusters for each user; Expressed as a natural constant An exponential function with base ; represents the linear normalization function.

[0089] For the User's parameter clusters and the User's The number of identical parameters in the parameter clusters The larger the value, the more similar the parameter types of the parameter clusters of the two users are, which means that the two users are more likely to belong to the same position and the smaller the degree of difference between the positions of the two users is; Indicates the User's parameter clusters and the User's The smaller the value, the more similar the modified importance of all parameters in the parameter clusters of the two users are, which means that the two users are more likely to belong to the same position and the smaller the degree of difference in the positions of the two users is.

[0090] Finally, according to the degree of job difference, that is, taking the job difference as the metric distance, DBSCAN clustering is performed on all users to obtain several job clusters, each of which represents a job.

[0091] S500: Analyze the similarity between the target user and each job cluster to obtain reference users of the target user.

[0092] When recommending materials to a search target user, we infer their job title based on their initial browsing data, and then recommend materials to them based on the preferences of users in the same job title. When determining the target user's likely job title, we primarily consider the difference in importance of each parameter between the target user and users in each job cluster. The smaller the difference, the more likely they are to be in the same job title.

[0093] Based on the above analysis, in an embodiment of the present invention, the similarity between the target user and each job cluster is analyzed to obtain reference users of the target user. Further steps include:

[0094] First, determine whether the target user collects materials; if not, use the initial importance of the parameter as the recommendation reference index; if yes, use the revised importance of the parameter as the recommendation reference index.

[0095] Then, we calculate the mean of the corrected importance of parameters for all users in the job cluster, combine it with the recommendation reference index, and traverse all parameters of the target user to obtain the similarity between the target user and each job cluster. The similarity calculation formula between the target user and each job cluster is constructed as follows:

[0096]

[0097] Where, Indicates the target user and Similarity of job clusters; Indicates the total number of types of parameters for each click material; Indicates that the target user is Recommended reference index of each parameter (initial importance or revised importance); Indicates the All users in the job cluster The mean of the corrected importance of the parameters; Expressed as a natural constant An exponential function with base .

[0098] Indicates the target user’s The recommended reference index (initial importance or revised importance) of the parameter is compared with the reference index of all users in the job cluster about the first The smaller the difference, the more likely the target user is to belong to the job cluster, that is, the target user and the users in the job cluster are more likely to belong to the same job.

[0099] Finally, all users in the job cluster with the highest similarity to the target user are recorded as the reference users of the target user.

[0100] S600: Recommending materials for the target cluster based on the reference user.

[0101] Recommending materials to the target cluster based on the reference user. This process further includes: first, using the reference user's browsing time, clicks, favorites, and interactive behavior to rate each material viewed by the reference user, obtaining the reference user's reference rating for each material; then, using a collaborative filtering algorithm based on the reference rating, obtaining the target user's rating for each material, obtaining the target user's target rating for each material; and finally, recommending materials to the target user in descending order of target ratings, i.e., ranking the highest-rated material first for recommendation.

[0102] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0103] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for querying engineering cost data using electronic information, characterized in that: The method comprises: Obtain the parameters of each user's clicked materials and browsing time, as well as the collection of materials; For each user, analyze the differences in browsing time of different click materials and the differences in the values ​​of the same parameter to obtain the preliminary importance of the parameter for each user; Based on any one of the parameters of each user, analyzing the degree of difference between the clicked materials and the favorited materials during the browsing process of the user, revising the preliminary importance, and obtaining the revised importance of the parameter of each user; Based on the modified importance of all the parameters of all the users, dividing all the users into different job clusters; Analyze the similarity between the target user and each of the job clusters to obtain reference users of the target user; Recommending materials to target users based on the reference users; The method of analyzing the difference between the clicked materials and the favorited materials during the browsing process of each user based on any one of the parameters of each user, revising the preliminary importance, and obtaining the revised importance of the parameter of each user includes: Based on any one of the parameters of each user, analyzing the difference in parameter values ​​between the clicked material and the favorited material during the user's browsing process, and obtaining the expected skewness of the parameter in the clicked material for each user; Clustering all the click materials of each user according to the browsing time of the click materials to obtain a click material cluster of each user; For any parameter of each user, analyzing the differences in the expected skewness corresponding to different click material clusters to which the parameter belongs, and obtaining an important index of the parameter of each user; The preliminary importance is revised according to the importance index to obtain the revised importance of the parameter of each user.

2. The method for querying engineering cost data using electronic information according to claim 1, characterized in that: Based on any one of the parameters of each user, the difference in parameter values ​​between the clicked material and the favorited material during the user's browsing process is analyzed to obtain the expected skewness of the parameter in the clicked material for each user, including: Based on any one of the parameters of each user, calculating the numerical mean of the parameter in all the user's collection materials; The difference between the numerical value of the parameter in the user's click material and the numerical mean is calculated to obtain the expected skewness of each user for the parameter in the click material.

3. The method for querying engineering cost data using electronic information according to claim 1, characterized in that: According to the browsing time of the click materials, all the click materials of each user are clustered to obtain the click material cluster of each user, including: According to the browsing time of the click materials, DBSCAN clustering is performed on all the click materials of each user to obtain a click material cluster of each user, wherein each click material cluster represents the time required for the user to browse to a certain parameter.

4. The method for querying engineering cost data using electronic information according to claim 3 is characterized in that: Get the click material cluster of each user, and then also include: According to the browsing time, the click material clusters are sorted in ascending order to obtain a click material cluster sequence for each user.

5. The method for querying engineering cost data using electronic information according to claim 4 is characterized in that: For any parameter of each user, the differences in the expected skewness corresponding to different click material clusters to which the parameter belongs are analyzed to obtain the important index of the parameter of each user, including: For any parameter of each user, based on the click material cluster sequence, the difference in the discrete degrees of the expected skewness corresponding to different click material clusters to which the parameter belongs is analyzed, and the possibility of each click material cluster of each user being used as the boundary of the parameter is obtained; According to the demarcation possibility and in combination with the position ordinal number of the click material cluster in the click material cluster sequence, the importance index of the parameter of each user is obtained.

6. The method for querying engineering cost data using electronic information according to claim 5, characterized in that: For any parameter of each user, based on the click material cluster sequence, the difference in the discrete degrees of the expected skewness corresponding to different click material clusters to which the parameter belongs is analyzed, and the possibility of each click material cluster of each user being used as the boundary of the parameter is obtained, including: For any one of the parameters of each user, according to the position ordinal number of the click material cluster in the click material cluster sequence, the difference between the discrete degrees of the expected skewness corresponding to all the click material clusters before the position ordinal number and the discrete degrees of the expected skewness corresponding to all the click material clusters after the position ordinal number are analyzed, and combined with the mean of the corresponding expected skewness in all the click material clusters, the possibility of each click material cluster of each user as the demarcation of the parameter is obtained.

7. The method for querying engineering cost data using electronic information according to claim 1, characterized in that: Based on the modified importance of all the parameters of all the users, all the users are divided into different job clusters, including: According to the modified importance, all the parameters of each user are clustered using DBSCAN to obtain a number of parameter clusters corresponding to each user; Analyze the differences between the modified importance of all parameters in all parameter clusters of any two users to obtain the degree of job difference between the two users; According to the degree of job differences, DBSCAN clustering is performed on all users to obtain several job clusters.

8. The method for querying engineering cost data using electronic information according to claim 1, characterized in that: Analyze the similarity between the target user and each of the job clusters to obtain reference users for the target user, including: Determine whether the target user collects the material; If not, the initial importance of the parameter is used as a recommended reference index; If yes, the revised importance of the parameter is used as a recommended reference index; Calculating the mean of the modified importance of the parameters for all users in the job cluster, combining it with the recommendation reference index, traversing all the parameters of the target user, and obtaining the similarity between the target user and each of the job clusters; All users in the job cluster that have the highest similarity to the target user are recorded as reference users of the target user.

9. The method for querying engineering cost data using electronic information according to claim 1, characterized in that: Recommend materials to target users based on the reference users, including: Scoring each material browsed by the reference user to obtain a reference score of each material by the reference user; According to the reference score, a collaborative filtering algorithm is used to obtain the score of each material given by the target user, thereby obtaining a target score of each material given by the target user; Materials are recommended to the target user in descending order of the target scores.

Images