A campus book retrieval method based on Internet big data

Abstract

The present invention relates to the technical field of book retrieval, and specifically discloses a campus book retrieval method based on Internet big data, wherein the method comprises the following steps: S1: calculating date difference and screening out new books; S2: setting time nodes at preset time intervals T within a time interval, obtaining interactive data of new books, and calculating the heat value of new books at each time node; S3: generating coordinate points, obtaining fitting curves, calculating average heat values, and marking new books whose average heat values ​​are greater than or equal to the average heat threshold as target books; S4: calculating the heat value variation coefficient of the target book, and calculating the target PR value through a formula. The present invention stipulates the screening criteria for the target book, and also has anti-interference ability, and realizes the amplification of the target PR value of the target book through the two factors of heat value and date difference, thereby solving the problem of slow increase of the target PR value.

Description

Technical Field

[0001] The present invention relates to the technical field of book retrieval, and in particular to a campus book retrieval method based on Internet big data. Background Art

[0002] At present, the campus book search system mainly uses the PageRank algorithm. The application of the PageRank algorithm in book search is mainly reflected in improving the efficiency and accuracy of book search by analyzing the reference relationship between books, thereby helping users find high-value literature resources more quickly. If a book is cited by many other books, its PR value will be relatively high; conversely, if a book with a high PR value cites another book, the PR value of the cited book will also increase accordingly. This application not only optimizes the traditional book sorting method, but also provides technical support for personalized and intelligent book search.

[0003] The PR value of each book is initialized to 1 / N, where N is the total number of books of the same type. Books build a web graph through reference relationships, and each page is set with the same PR value. Iterative calculations are performed, and a new PR value can be obtained in each round of calculations. The transition probability matrix is ​​used to express the link relationship between books, and a new round of PR value calculations is performed by matrix multiplication. The algorithm converges through repeated iterative calculations until the PR value of each book does not change much or reaches the preset number of iterations. The final PR value reflects the importance and quality of each book. Books with higher PR values ​​have higher priority in the retrieval process.

[0004] In the prior art, the PR value needs to be calculated through multiple iterations until the PR value of each book does not change much or reaches a preset number of iterations. In multiple iterations, old books will continue to occupy high PR values ​​due to the accumulation of a large number of citations. New books are cited less frequently due to their short publication time, which leads to a slow increase in PR value. In this case, even if the new book is of high quality, it is difficult to be recognized quickly, which is not conducive to the retrieval of high-quality new books. Summary of the invention

[0005] The purpose of the present invention is to provide a campus book retrieval method based on Internet big data to solve the above technical problems.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A campus book retrieval method based on Internet big data includes the following steps:

[0008] S1: Determine the current date D t and the publication date of the book p , calculate the date difference DC=D t-D p , the date difference DC<D sta of books and marked them as new books, where D sta is the preset date difference;

[0009] S2: In the time interval [D p , D t ] at a preset time interval T (D p -D t ) / T time nodes, and obtain the interactive data of the new book at the time nodes, the interactive data includes the total reading volume, the comprehensive like volume, the total comment volume and the total collection volume, wherein the comprehensive like volume is equal to the like volume minus the dislike volume, and the popularity value is calculated by the following formula:

[0010]

[0011] Among them, Hj represents the popularity value of the new book at the jth time node, N j_view represents the total reading volume at the jth time node, N j_like represents the total number of likes at the jth time node, N j_reply represents the total number of comments at the jth time node, N j_favor represents the total collection at the jth time node, λ1, λ2, λ3 are all preset parameters, and 1<λ1<λ2<λ3;

[0012] S3: Generate coordinate points (t j , H j ), t j represents the time corresponding to the jth time node, the coordinate points are fitted to obtain the fitting curve H(t), and the average heat value is calculated. When the average heat value H ave When the average heat value is greater than or equal to a preset threshold value, the new book is marked as a target book;

[0013] S4: Calculate the heat value variation coefficient K=H of the target book ave / H s , and calculate the target PR value PR of the target book goal , , where H s represents the preset standard average heat value, γ represents the preset time coefficient, H now Represents the popularity value of the target book at the current time node, and µ is the preset unit coefficient.

[0014] As a further solution of the present invention: in the step S3, another method for screening the target books is to calculate the fitting curve [D p ,(D t -Dp ) / 2) and ((D t -D p ) / 2,D t The average heat value H front and H rear , if H front >H rear Then cancel the mark of the target book and stop the subsequent steps.

[0015] As a further solution of the present invention: in step S3, another method for screening target books specifically includes:

[0016] The heat value increase Δh is calculated by a formula, wherein the formula specifically includes:

[0017]

[0018] When the heat value increase Δh is greater than Δh sta , the corresponding new book is marked as the target book, where J represents the number of time nodes, Δh sta It is the preset standard heat value increment.

[0019] As a further solution of the present invention: in the step S1, when the date difference DC of a book is less than the preset minimum date difference D min , the subsequent steps will not be performed on the book.

[0020] As a further solution of the present invention: in the step S3, a heat floating value δ is set, and the heat value that does not belong to the heat value interval [H ave -δ,H ave +δ] are eliminated.

[0021] As a further solution of the present invention: in step S2, for the comprehensive likes N j_like Calculation of the total number of likes N j_like =AB, where A represents the number of likes and B represents the number of dislikes. A like means that the reader recognizes the quality of the book, and a dislike means that the book does not recognize the quality of the book. If the number of dislikes B is greater than the number of likes A, let λ2=-λ2.

[0022] As a further solution of the present invention: in the step S4, for the setting of the time coefficient γ, the time coefficient γ=β*DC, wherein β is a preset first time adjustment coefficient.

[0023] As a further solution of the present invention: in step S4, when the time coefficient γ=γ sta season , calculate the standard date difference X, where γ sta is the preset time coefficient standard value, when Dt -D p <X, let γ=γ sta .

[0024] Beneficial effects of the present invention: In the present invention, firstly, in order to screen out new books, the date difference of the books is calculated, and the new books are selected by the date difference, which represents the time when the books exist. In this way, books with low PR values ​​due to short publishing time can be selected;

[0025] Then, the popularity value is calculated using the interaction data of the new book. The interaction data includes the total reading volume, the total number of likes, the total number of comments, and the total number of collections. The interaction data can reflect the popularity of the book. The higher the value in the interaction data, the more popular the book is. The books that are popular among readers are of higher quality. Therefore, the popularity value is calculated through the interaction data, and the quality of the book is expressed in the form of the size of the heat value. The higher the quality of the book, the greater the heat value.

[0026] Then the average popularity of the books is calculated, and only new books with an average popularity value greater than the average popularity threshold can be marked as target books. The average popularity value can reflect the overall quality of the book, while the average popularity threshold is the dividing line between high-quality and low-quality books. Being marked as a target book means that the book meets the standards of high-quality new books.

[0027] When calculating the target PR value of the target book, it is mainly affected by two factors: the popularity value change coefficient and the date difference. The popularity value change coefficient is used to amplify the target PR value. This is the initial amplification process. Through a series of calculations of the date difference, it is used to adjust the impact of the date on the target popularity value. The combination of the two speeds up the increase of the target PR value.

[0028] In summary, the present invention not only clarifies the specific screening criteria for the target books, but also has a certain anti-interference ability and enhanced stability, and amplifies the target PR value of the target book through the two factors of heat value and date difference, thereby solving the problem of slow increase of the target PR value. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention will be further described below in conjunction with the accompanying drawings.

[0030] Figure 1 The present invention is a flow chart of a campus book retrieval method based on Internet big data. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] See also Figure 1 As shown, the present invention is a campus book retrieval method based on Internet big data, comprising the following steps:

[0033] S1: Determine the current date D t and the publication date of the book p , calculate the date difference DC=D t -D p , the date difference DC<D sta of books and marked them as new books, where D sta is the preset date difference;

[0034] S2: In the time interval [D p , D t ] at a preset time interval T (D p -D t ) / T time nodes, and obtain the interactive data of the new book at the time nodes, the interactive data includes the total reading volume, the comprehensive like volume, the total comment volume and the total collection volume, wherein the comprehensive like volume is equal to the like volume minus the dislike volume, and the popularity value is calculated by the following formula:

[0035]

[0036] Among them, Hj represents the popularity value of the new book at the jth time node, N j_view represents the total reading volume at the jth time node, N j_like represents the total number of likes at the jth time node, N j_reply represents the total number of comments at the jth time node, N j_favor represents the total collection at the jth time node, λ1, λ2, λ3 are all preset parameters, and 1<λ1<λ2<λ3;

[0037] S3: Generate coordinate points (t j , H j ), t j represents the time corresponding to the jth time node, the coordinate points are fitted to obtain the fitting curve H(t), and the average heat value is calculated. When the average heat value H ave When the average heat value is greater than or equal to a preset threshold value, the new book is marked as a target book;

[0038] S4: Calculate the heat value variation coefficient K=H of the target book ave / H s , and calculate the target PR value PR of the target book goal , , where H s represents the preset standard average heat value, γ represents the preset time coefficient, H now Represents the popularity value of the target book at the current time node, and µ is the preset unit coefficient.

[0039] First, in order to filter out new books, calculate the date difference of the books and select new books by the date difference. The date difference represents the time of existence of the book. In this way, books with low PR value can be selected due to short publication time.

[0040] Then, the popularity value is calculated using the interaction data of the new book. The interaction data includes the total reading volume, the total number of likes, the total number of comments, and the total number of collections. The interaction data can reflect the popularity of the book. The higher the value in the interaction data, the more popular the book is. The books that are popular among readers are of higher quality. Therefore, the popularity value is calculated through the interaction data, and the quality of the book is expressed in the form of the size of the heat value. The higher the quality of the book, the greater the heat value.

[0041] Then the average popularity of the books is calculated, and only new books with an average popularity value greater than the average popularity threshold can be marked as target books. The average popularity value can reflect the overall quality of the book, while the average popularity threshold is the dividing line between high-quality and low-quality books. Being marked as a target book means that the book meets the standards of high-quality new books.

[0042] When calculating the target PR value of the target book, it is mainly affected by two factors: the popularity value change coefficient and the date difference. The popularity value change coefficient is used to amplify the target PR value. This is the initial amplification process. Through a series of calculations of the date difference, it is used to adjust the impact of the date on the target popularity value. The combination of the two speeds up the increase of the target PR value.

[0043] In addition, it should be noted that due to different evaluation criteria for different types of books, the target books finally screened out according to different evaluation criteria may also be different. In order to make the results more accurate, the present invention only operates on books of the same type. The purpose is to avoid errors in subsequent steps due to different evaluation criteria. The present invention does not make any requirements for the classification of books, and the specific classification is based on the classification rules of the library.

[0044] In a preferred embodiment of the present invention, a campus book retrieval method based on Internet big data is characterized in that in the step S3, another method for screening target books is to calculate the fitting curve [D p ,(D t -D p ) / 2) and ((D t -D p ) / 2,D t The average heat value H front and H rear , if H front >H rear Then cancel the mark of the target book and stop the subsequent steps.

[0045] For the method in this preferred embodiment, the target book that has shown a decreasing trend in popularity value is unmarked, indicating that the target book is already in the process of decreasing popularity value. At this time, it is meaningless to process the target book, so the subsequent steps are stopped.

[0046] In addition, it is also necessary to note that there is no unique explanation for the publication date of the target book. The main purpose is to clarify the publication date of serial books, which are books of the type that are released in batches. When selecting this type of serial book, the first publication date of the serial book is used as the publication date of the book. In the present invention, if the latest publication date is used as the publication date of the book, the date difference DC will be too small, resulting in a small target PR value for the target book when it is subsequently calculated. Therefore, the first publication date of the serial book is used as the publication date of the book, and all serial books are regarded as a whole. This is not only conducive to the subsequent calculation of the popularity value of the serial book, but also can reduce the amount of calculation when generating the fitting curve.

[0047] In a preferred embodiment of the present invention, another method for screening target books specifically includes:

[0048] The heat value increase Δh is calculated by a formula, wherein the formula specifically includes:

[0049]

[0050] When the heat value increase Δh is greater than Δh sta , the corresponding new book is marked as the target book, where J represents the number of time nodes, Δh sta It is the preset standard heat value increment.

[0051] It should be noted that in this preferred embodiment, the focus of target book screening is on the increase in popularity Δh. This method of screening target books pays more attention to the potential of books. For books whose popularity increase Δh does not meet the requirements, it may be because the date difference DC of the book is small and thus temporarily does not meet the screening requirements of the target books. However, overall, the popularity value of the book shows a steady upward trend. The popularity value of each time node of the book is steadily increasing, and it can be considered that the popularity value of the book will continue to rise in the future. The book will be able to meet the requirements of the target book in the future, so the book is marked as a target book.

[0052] In a preferred embodiment of the present invention, when the date difference DC of a book is less than the preset minimum date difference D min , the subsequent steps will not be performed on the book.

[0053] It is understandable that if the date difference DC of a book is too small, the book has too few time nodes set, resulting in too little interaction data. Too little interaction data is not enough to support subsequent steps. Assuming that only two time nodes can be set within the date difference DC of the book, since only two sets of interaction data are obtained, there are only two coordinates of the coordinate points calculated through the interaction data. At this time, it is impossible to generate a fitting curve; or increase the time nodes to obtain more interaction data. Increasing time nodes means shortening the time interval T. This idea seems reasonable, but there is another problem. When the time interval T decreases, the interaction data of multiple time nodes may be very close. At this time, even if the number of coordinate points increases, it is difficult to reflect the trend of the subsequent heat value of the book through the fitting curve, and the accuracy of the fitting curve is also low. At this time, there is no practical significance for performing subsequent steps on the book. Therefore, when the date difference DC is too small, it is necessary to wait until the date difference of the book meets the preset minimum date difference before performing subsequent operations on the book.

[0054] In a preferred embodiment of the present invention, a heat floating value δ is set, and coordinate points whose heat values ​​do not belong to the heat value interval [Have-δ, Have+δ] are eliminated.

[0055] It should be understood that not all calculated coordinate points have reference value. Some coordinate points with excessive deviations should be eliminated. After eliminating the coordinate points, the fitting curve can better describe the change process of the popularity value of the target book. In actual situations, a book may be repeatedly liked or disliked by readers within a certain period of time due to improper operation or subjective factors, resulting in the popularity value of the book being too high or too low. At the same time, in order to avoid a short-term surge in popularity caused by following the trend, in order to reduce the impact of these factors, before generating the fitting curve, these coordinate points with excessive deviations from the average value are eliminated.

[0056] In a preferred embodiment of the present invention, for the comprehensive likes N j_like Calculation of the total number of likes N j_like =AB, where A represents the number of likes and B represents the number of dislikes. If the number of dislikes B is greater than the number of likes A, set the value of λ2 equal to -λ2.

[0057] It should be noted that in this preferred scheme, likes represent readers' recognition of the quality of the book, and dislikes represent readers' disapproval of the quality of the book. The comprehensive likes are calculated by combining the values ​​of the two. Both the likes and dislikes will affect the heat value. Therefore, when the dislikes in the interactive data are greater than the likes, it means that most readers do not recognize the quality of the book. At this time, the comprehensive likes will make the heat value relatively reduced, so the value of λ2 is set equal to -λ2.

[0058] In a preferred embodiment of the present invention, for setting the time coefficient γ, the time coefficient γ=β*DC, wherein β is a preset first time adjustment coefficient.

[0059] It can be understood from this preferred solution that the value of the time coefficient γ is inversely proportional to the date difference DC. As DC increases, the value of γ decreases and the target PR value increases. The purpose of this preferred solution is to increase the influence of the date difference DC on the target PR value, reduce the amplifying function of the date difference on the PR value of the target book, and avoid the final target PR value rising too fast.

[0060] In a preferred embodiment of the present invention, when the time coefficient γ=γ sta season , calculate the value of X, where γ sta is the preset time coefficient standard value, when D t -D p <X, let γ=γ sta .

[0061] It can be understood that when the date difference DC is less than the time coefficient γ, if the value obtained by calculation is less than 1, then the target PR value calculated at this time may eventually result in a smaller result, resulting in the target PR value result being inaccurate. In order to reduce the impact of this situation on the target PR value, in this preferred solution, by presetting the time coefficient standard value γ sta , find the value of X, and then when D t -D p <X, let γ=γ sta , X is obtained by calculation, and then γ is dynamically adjusted according to the value of , so that the calculation result of is 1, avoiding affecting the final calculation result.

[0062] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A campus book retrieval method based on Internet big data, characterized in that: The following steps are involved: S1: Determine the current date D t and the publication date of the book p , calculate the date difference DC=D t -D p , the date difference DC<D sta of books and marked them as new books, where D sta is the preset date difference; S2: In the time interval [D p , D t ] at a preset time interval T (D p -D t ) / T time nodes, and obtain the interactive data of the new book at the time nodes, the interactive data includes the total reading volume, the comprehensive like volume, the total comment volume and the total collection volume, wherein the comprehensive like volume is equal to the like volume minus the dislike volume, and the popularity value is calculated by the following formula: Among them, Hj represents the popularity value of the new book at the jth time node, N j_view represents the total reading volume at the jth time node, N j_like represents the total number of likes at the jth time node, N j_reply represents the total number of comments at the jth time node, N j_favor represents the total collection at the jth time node, λ1, λ2, λ3 are all preset parameters, and 1<λ1<λ2<λ3; S3: Generate coordinate points (t j , H j ), t j represents the time corresponding to the jth time node, the coordinate points are fitted to obtain the fitting curve H(t), and the average heat value is calculated. When the average heat value H ave When the average heat value is greater than or equal to a preset threshold, the new book is marked as a target book; S4: Calculate the heat value variation coefficient K=H of the target book ave / H s , and calculate the target PR value PR of the target book goal , , where H s represents the preset standard average heat value, γ represents the preset time coefficient, H now Represents the popularity value of the target book at the current time node, and µ is the preset unit coefficient.

2. A campus book retrieval method based on Internet big data according to claim 1, characterized in that: In step S3, another method for screening target books is to calculate the fitting curve [D p ,(D t -D p ) / 2) and ((D t -D p ) / 2,D t The average heat value H front and H rear , if H front >H rear Then cancel the mark of the target book and stop the subsequent steps.

3. A campus book retrieval method based on Internet big data according to claim 1, characterized in that: In step S3, another method for screening target books specifically includes: The heat value increase Δh is calculated by a formula, wherein the formula specifically includes: When the heat value increase Δh is greater than Δh sta , the corresponding new book is marked as the target book, where J represents the number of time nodes, Δh sta It is the preset standard heat value increment.

4. A campus book retrieval method based on Internet big data according to claim 1, characterized in that: In step S1, when the date difference DC of a book is less than the preset minimum date difference D min , the subsequent steps will not be performed on the book.

5. The campus book retrieval method based on Internet big data according to claim 1 is characterized in that: In step S3, a heat floating value δ is set to set the heat value not belonging to the heat value interval [H ave -δ,H ave +δ] are eliminated.

6. The campus book retrieval method based on Internet big data according to claim 1 is characterized in that: In step S2, for the comprehensive number of likes N j_like Calculation of the total number of likes N j_like =AB, where A represents the number of likes and B represents the number of dislikes. A like means that the reader recognizes the quality of the book, and a dislike means that the book does not recognize the quality of the book. If the number of dislikes B is greater than the number of likes A, let λ2=-λ2.

7. The campus book retrieval method based on Internet big data according to claim 1 is characterized in that: In the step S4, the time coefficient γ is set as follows: the time coefficient γ=β*DC, wherein β is a preset first time adjustment coefficient.

8. A campus book retrieval method based on Internet big data according to claim 7, characterized in that: In step S4, when the time coefficient γ=γ sta season , calculate the standard date difference X, where γ sta is the preset time coefficient standard value, when D t -D p <X, let γ=γ sta .

Images