Ranking aggregation method and system based on link prediction, storage medium and equipment

By constructing and standardizing the competition graph and adjusting weights using link prediction technology, the problems of insufficient accuracy and high computational complexity in existing ranking aggregation methods are solved, resulting in more accurate and comprehensive ranking results.

CN119025951APending Publication Date: 2024-11-26BEIJING NORMAL UNIV AT ZHUHAI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411017959.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Existing ranking aggregation methods suffer from insufficient precision and high computational complexity when dealing with complex ranking relationships, resulting in ranking results that are not comprehensive or accurate enough.

Method used

By constructing a competition matrix and competition graph among the evaluated objects, and after standardization, link prediction technology is used to adjust the edge weights to obtain an enhanced competition graph. Finally, the aggregate ranking is determined based on the enhanced competition graph.

Benefits of technology

It improves the accuracy and comprehensiveness of ranking aggregation, providing reliable aggregated ranking results even with limited information, thus alleviating the need for a large number of ranking lists.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025951A_ABST
    Figure CN119025951A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses a ranking aggregation method based on link prediction. The method comprises the following steps: acquiring a ranking list of a plurality of evaluated objects; constructing a competition matrix among the evaluated objects according to the ranking list, and obtaining a corresponding competition map; standardizing the competition map to obtain a standardized competition map; according to link prediction, adjusting a connecting edge weight in the standardized competition graph, and obtaining an enhanced competition graph; and determining an aggregation ranking according to the enhanced competition map. According to the method, the information of a plurality of ranking lists and the complex relation between objects can be synthesized, and a more comprehensive and accurate ranking result can be provided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data quantitative analysis and processing, and in particular to a ranking aggregation method, system, storage medium and device based on link prediction. Background Art

[0002] Ranking issues, as a core element of decision support, are widely permeated in various fields of social economy, from education evaluation, product recommendation to scientific research evaluation, all of which reflect its importance. The decision-making process in these fields often relies on ranking information from multiple sources. This information may be based on different evaluation criteria, attributes or personal preferences of judges, forming a diverse ranking list. In order to obtain a comprehensive, comprehensive and accurate ranking result, the effective integration of these diverse ranking lists, namely "ranking aggregation", has become a research hotspot.

[0003] Existing ranking aggregation methods are mainly divided into two categories: heuristic methods and optimization methods. Heuristic methods, such as Borda counting method (BM), Markov chain method (MC) and competitive graph method (CG), have been widely used in practical applications due to their high computational efficiency and simple implementation. However, these methods may face the problem of insufficient precision when dealing with complex ranking relationships. In contrast, although optimization methods have higher accuracy potential in theory, they usually involve NP-hard problems. When faced with large-scale rated objects, the computational complexity increases sharply, making it difficult to implement effectively in practical applications.

[0004] In the prior art, since evaluators can only evaluate based on their limited experience or knowledge, the generated ranking lists often only cover a small part of the entire set of evaluation objects, and there is a lack of sufficient overlap between these lists for direct comparison. The incompleteness and insufficiency of this ranking information poses a huge challenge to ranking aggregation, making it difficult for traditional methods to generate reliable and objective aggregate ranking results. Summary of the invention

[0005] Based on this, it is necessary to propose a ranking aggregation method based on link prediction to address the above problems.

[0006] A ranking aggregation method based on link prediction, the method comprising the following steps:

[0007] Get a ranking list of several rated objects;

[0008] According to the ranking list, a competition matrix between the evaluated objects is constructed, and a corresponding competition graph is obtained;

[0009] Standardizing the competition graph to obtain a standardized competition graph;

[0010] Adjust the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph;

[0011] An aggregate ranking is determined based on the enhanced competition graph.

[0012] In the above solution, obtaining a ranking list of several evaluated objects specifically includes:

[0013] According to m objects to be evaluated and n judges, a ranking list R of the objects to be evaluated is determined. i =[r i1 ,r i2 …r im ], the ranking list R i The length is L i =|{r ij |r ij >0,1≤j≤m}|, where 1≤L i ≤m;

[0014] The set of m evaluated objects is preset as: O = {o1, o2…o m}, the set of n voter is: V = {v1, v2…v n};

[0015] If the judges v i No evaluation was done on the object j For ranking, then r ij =0;

[0016] If the ranking list R i The first evaluated object in s With the second evaluated object o t If the ranking is the same, then r is =r it .

[0017] In the above scheme, constructing a competition matrix between the evaluated objects according to the ranking list and obtaining a corresponding competition graph specifically includes:

[0018] Preset the judges v i The preference matrix is

[0019] In the ranking list R i In the case where the first evaluated object o s The ranking is lower than the second rated object o t , then: r it >r is >0, and in the preference matrix

[0020] In the ranking list R i In the case where the first evaluated object o s The ranking is greater than the second rated object o t , then: rr is , and in the preference matrix

[0021] In the ranking list R i In the case where the first evaluated object o s The ranking is equal to the second rated object o t , then: rr it >0, and in the preference matrix

[0022] According to the preference matrix of the n voter judges, the competition matrix between the objects to be evaluated is determined as follows: A = (a st ) n×n ,in, a st For the object of evaluation s Ranked lower than the rated object o t The total number of times.

[0023] In the above scheme, constructing a competition matrix between the evaluated objects according to the ranking list and obtaining a corresponding competition graph specifically includes:

[0024] The competition graph G=(V,E) between the evaluated objects is preset, where V is a node set, The ranking relationship between the rated objects;

[0025] All nodes in the competition graph correspond one-to-one to the evaluated objects;

[0026] If a st >0, then there is a line in the competition graph consisting of the evaluated object o s Start from the node and go to the object to be evaluated o t The edge of the node st , and the edge e st The weight is a st .

[0027] In the above solution, the step of normalizing the competition graph to obtain a standardized competition graph specifically includes:

[0028] The edge e st The weight a st Adjust to in, For the object of evaluation s Defeat the object being evaluated t The probability of

[0029] Determine the standardized competition map based on the adjustment results

[0030] In the above solution, adjusting the edge weights in the standardized competition graph according to the link prediction to obtain the enhanced competition graph specifically includes:

[0031] When there is no node pair with direct edges in the competition graph (o i ,o k ) There are common neighbor nodes o j ,and When, according to Determine the node pair (o i ,o k )’s potential edge ik Partial weight of

[0032] Obtain all common neighbor nodes in the standardized competition graph, add up the weights corresponding to the common neighbor nodes, and obtain the potential edge e ik The overall weight of:

[0033]

[0034] The overall weight is standardized according to the following formula to determine the potential edge e ik The standardized weights are:

[0035]

[0036] Potential edges are mined and weights are defined for nodes that meet preset requirements to obtain an enhanced competition graph.

[0037] In the above solution, determining the aggregate ranking according to the enhanced competition graph specifically includes:

[0038] Counting the weighted out-degree and weighted in-degree of each node in the enhanced competition graph;

[0039] Determine the score of each node according to the weighted out-degree and the weighted in-degree;

[0040] All nodes are arranged in ascending or descending order according to the score of each node to obtain an aggregate ranking.

[0041] The present application proposes a ranking aggregation system based on link prediction, the system comprising: a ranking list acquisition unit, a competition graph acquisition unit, a normalization unit and a ranking determination unit;

[0042] The ranking list acquisition unit is used to obtain a ranking list of several evaluated objects;

[0043] The competition graph acquisition unit is used to construct a competition matrix between the evaluated objects according to the ranking list and acquire a corresponding competition graph; adjust the edge weights in the standardized competition graph according to the link prediction to acquire an enhanced competition graph;

[0044] The standardization unit is used to standardize the competition graph to obtain a standardized competition graph;

[0045] The ranking determination unit is used to determine the aggregate ranking according to the enhanced competition graph.

[0046] The present application proposes a readable storage medium storing a computer program. When the computer program is executed by a processor, the processor performs the following steps:

[0047] Get a ranking list of several rated objects;

[0048] According to the ranking list, a competition matrix between the evaluated objects is constructed, and a corresponding competition graph is obtained;

[0049] Standardizing the competition graph to obtain a standardized competition graph;

[0050] Adjust the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph;

[0051] An aggregate ranking is determined based on the enhanced competition graph.

[0052] The present application proposes a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor in the following steps:

[0053] Get a ranking list of several rated objects;

[0054] According to the ranking list, a competition matrix between the evaluated objects is constructed, and a corresponding competition graph is obtained;

[0055] Standardizing the competition graph to obtain a standardized competition graph;

[0056] Adjust the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph;

[0057] An aggregate ranking is determined based on the enhanced competition graph.

[0058] The embodiment of the present invention has the following beneficial effects: first, a ranking list of several evaluated objects is obtained; based on the ranking list, a competition matrix between the evaluated objects is constructed, and a corresponding competition graph is obtained; the competition graph is standardized to obtain a standardized competition graph; the edge weights in the standardized competition graph are adjusted according to link prediction to obtain an enhanced competition graph; and the aggregate ranking is determined according to the enhanced competition graph. The present invention can integrate the information of multiple ranking lists and the complex relationships between objects to provide more comprehensive and accurate ranking results, mine potential ranking information through link prediction, and improve the effect of ranking aggregation under limited information. Link prediction can alleviate the need for a large number of ranking lists in ranking aggregation, so that the ranking system can obtain relatively accurate and reliable aggregate ranking results even in the cold start phase. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 creative work.

[0060] in:

[0061] Figure 1 A schematic diagram of a ranking aggregation method based on link prediction in one embodiment;

[0062] Figure 2 is a standard competition diagram in one embodiment;

[0063] Figure 3 Based on Figure 2 Enhanced contention graph using link prediction techniques. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0065] In the following description, a large number of specific details are given in order to provide a more thorough understanding of the present invention; however, it is obvious to those skilled in the art that the present invention can be implemented without one or more of these details; in other examples, in order to avoid confusion with the present invention, some technical features known in the art are not described, and it should be understood that the present invention can be implemented in different forms and should not be interpreted as limited to the embodiments presented here; on the contrary, providing these embodiments will make the disclosure thorough and complete and fully convey the scope of the present invention to those skilled in the art.

[0066] The purpose of the terms used herein is only to describe specific embodiments and is not intended to be limiting of the present invention. When used herein, the singular forms "a", "an" and "said / the" are also intended to include the plural forms, unless the context clearly indicates otherwise, and it should also be understood that the terms "consisting of" and / or "comprising", when used in this specification, determine the presence of the features, integers, steps, operations, elements and / or parts, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, parts and / or groups. When used herein, the term "and / or" includes any and all combinations of the relevant listed items.

[0067] In the prior art, due to the diversity of data sources, different ranking systems may generate rankings based on different standards or algorithms, resulting in significant differences in the positions of the same object in different ranking lists, which increases the difficulty of user decision-making. Many ranking lists are limited by the difficulty and cost of data collection and cannot fully cover all relevant objects, resulting in incomplete ranking results and easy omission of important information. In response to the ranking aggregation problem of fragmented and limited ranking lists, this patent discloses a ranking aggregation method based on link prediction, which uses the transitive principle of ranking to apply link prediction technology on the network, mine potential ranking information, and enhance the ranking aggregation effect.

[0068] In order to thoroughly understand the present invention, a detailed structure will be presented in the following description to illustrate the technical solution proposed by the present invention; optional embodiments of the present invention are described in detail as follows, but in addition to these detailed descriptions, the present invention may also have other implementation methods.

[0069] like Figure 1 As shown, in one embodiment, a ranking aggregation method based on link prediction is provided, and the ranking aggregation method based on link prediction includes steps S101 to S106, which are described in detail as follows:

[0070] S101, obtaining a ranking list of several evaluated objects;

[0071] By taking multiple ranking lists as input, this method is able to integrate information from multiple sources, increasing the comprehensiveness and diversity of perspectives of the rankings. This helps reduce the bias or error that may exist in a single ranking list.

[0072] In some embodiments, obtaining a ranking list of several evaluated objects specifically includes:

[0073] Determine the ranking list R of the evaluated objects based on m evaluated objects and n judges voter i =[r i1 ,r i2 …r im ], ranking list R i The length is L i =|{r ij |r ij >0,1≤j≤m}|, where 1≤L i ≤m;

[0074] The set of m objects to be evaluated is assumed to be: O = {o1, o2…o m}、The set of n voter judges is: V = {v1,v2…v n};

[0075] If the judges v i No evaluation was done on the object j For ranking, then r ij =0;

[0076] If the ranked list R i The first evaluated object in s With the second evaluated object o t If the ranking is the same, then r is =r it .

[0077] In practical applications, a judge will often only rank some of the objects being evaluated, which means that R i It is just an incomplete ranking list containing some of the rated objects. i =[r i1 ,r i2 …r im ] indicates judges i Given a ranked list, where r ij For the object of evaluation j In R i Ranking in .

[0078] S102. Construct a competition matrix among the evaluated objects according to the ranking list, and obtain a corresponding competition graph;

[0079] Constructing a competition matrix and competition diagram can intuitively display the relative strength of the relationships between the evaluated objects, providing a strong data basis for subsequent ranking aggregation. This method helps capture the complex relationships between objects, especially those subtle differences that are difficult to reflect in traditional single rankings.

[0080] In some embodiments, a competition matrix between the evaluated objects is constructed based on the ranking list, and a corresponding competition graph is obtained, which specifically includes:

[0081] Default judges i The preference matrix is

[0082] In the ranking list R i If the first evaluated object o s The ranking is lower than the second rated object o t , then: r it >r is >0, and in the preference matrix

[0083] In the ranking list R i In the case where the first evaluated object o s The ranking is greater than the second rated object o t , then: rr is , and in the preference matrix

[0084] In the ranking list R i In the case where the first evaluated object o s The ranking is equal to the second rated object o t , then: rr it >0, and in the preference matrix

[0085] According to the preference matrix of n voters, the competition matrix between the objects to be evaluated is determined as follows: A = (a st ) n×n ,in, a st For the object of evaluation s Ranked lower than the rated object o t The total number of times (the object of evaluation s Defeat the object being evaluated t total number of times).

[0086] Preferably, because the object of evaluation is j There will be no record of winning or losing with itself, so the elements on the main diagonal position in the competition matrix are all 0.

[0087] In some embodiments, a competition matrix between the evaluated objects is constructed based on the ranking list, and a corresponding competition graph is obtained, which specifically includes:

[0088] The competition graph G = (V, E) between the evaluated objects is preset, where V is a node set, The ranking relationship between the rated objects;

[0089] All nodes in the competition graph correspond one-to-one to the evaluated objects;

[0090] If a st >0, then there is a line in the competition graph consisting of the evaluated object o s Start from the node and go to the object to be evaluated o t The edge of the node st , and the edge e st The weight is a st .

[0091] S103, standardizing the competition graph to obtain a standardized competition graph;

[0092] Standardizing the competition graph can eliminate possible dimensional differences between different ranking lists, allowing data from different sources to be compared and aggregated under the same framework, which helps improve the accuracy and interpretability of the aggregate ranking.

[0093] In some embodiments, the competition graph is standardized to obtain a standardized competition graph, specifically including:

[0094] The edge st The weight a st Adjust to in, For the object of evaluation s Defeat the object being evaluated t The probability of

[0095] Determine the standardized competition map based on the adjustment results

[0096] S104, adjusting the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph;

[0097] In order to improve the effectiveness of ranking aggregation under poor information conditions, based on the ranking aggregation analysis framework based on network representation (competition graph), this patent introduces link prediction in network science to mine and supplement potential effective ranking information. i With o k There is no direct record of wins and losses between them, but o i Many victories over j , o jAlso defeated o many times k , o i I should be able to win many times. k , which is a generally accepted property in ranking theory - transitivity. Based on this unique property, this patent introduces link prediction technology in the standardized competition graph of the rated objects to mine these potential ranking information. By adjusting the edge weights through link prediction, the prediction of the potential relationship between the rated objects can be introduced, thereby enhancing the accuracy and information content of the competition graph. This method helps to discover hidden relationships that are not fully reflected in the existing ranking list.

[0098] In some embodiments, adjusting the edge weights in the standardized competition graph according to the link prediction to obtain the enhanced competition graph specifically includes:

[0099] When there is no node pair with direct edges in the competition graph (o i ,o k ) There are common neighbor nodes o j ,and When, according to Determine the node pair (o i ,o k )’s potential edge ik Partial weight of

[0100] Get all common neighbor nodes in the standardized competition graph, add up the weights corresponding to the common neighbor nodes, and get the potential edge e ik The overall weight of:

[0101]

[0102] The overall weight is standardized according to the following formula to determine the potential edge e ik The standardized weights are:

[0103]

[0104] Potential edges are mined and weights are defined for nodes that meet preset requirements to obtain an enhanced competition graph.

[0105] S105. Determine an aggregate ranking according to the enhanced competition graph.

[0106] The aggregate ranking determined based on the enhanced competition graph can comprehensively consider the information of multiple ranking lists and the complex relationships between objects, providing a more comprehensive and accurate ranking result. This method helps users better understand and compare different rated objects.

[0107] Preferably, the competition graph can be used as ranking information input and applied to certain graph theory-based or classical ranking aggregation methods, such as the Borda counting method (BM), the Markov chain method (MC), and the competition graph method (CG). Similarly, the enhanced competition graph obtained after processing by the link prediction technology proposed in this patent can also be used as ranking information input, and then the above-mentioned method is used to perform ranking aggregation on the competition graph.

[0108] In some embodiments, determining the aggregate ranking according to the enhanced competition graph specifically includes:

[0109] Statistically enhance the weighted out-degree and weighted in-degree of each node in the competition graph;

[0110] Determine the score of each node based on the weighted out-degree and weighted in-degree;

[0111] All nodes are sorted in ascending or descending order according to the score of each node to obtain the aggregate ranking.

[0112] Among them, the weighted out-degree is the sum of the weights of the edges from the node to other nodes, and the weighted in-degree is the sum of the weights of the edges from the node to other nodes. Preferably, the ratio of the weighted out-degree to the weighted in-degree or other suitable indicators can be used to calculate the score of each node, and this score can reflect the performance of the object in the overall competition.

[0113] In summary, the present application scheme mines potential ranking information through link prediction, improving the effect of ranking aggregation under limited information; link prediction can alleviate the demand for a large number of ranking lists in ranking aggregation, so that the ranking system can obtain relatively accurate and reliable aggregate ranking results even in the cold start phase.

[0114] The above steps S101 to S105 are described below using a specific embodiment:

[0115] In a ranking aggregation problem with 6 objects to be evaluated and 3 judges, all three judges only gave the rankings of some of the objects to be evaluated (as shown in the following table):

[0116]

[0117] In the above table, the ranking after aggregation should be o1>o2>o3>o4>o5>o6. The aggregation result obtained using the Borda counting method (BM) is: o3>o1>o2=o4=o5>o6, the aggregation result obtained using the competition graph method (CG) is: o1>o3>o2=o4>o5>o6, and the aggregation result obtained using the Markov chain method (MC) is: o1>o3>o2>o5>o4>o6.

[0118] Next, we use the ranking aggregation method based on link prediction proposed in this patent to aggregate the ranking. First, we standardize the competition matrix to obtain the standardized competition matrix

[0119]

[0120] Then in Figure 2 Link prediction is performed on the standardized competition graph shown in FIG. , and the predicted edge weights are standardized to obtain an enhanced competition graph, such as Figure 3 shown.

[0121] The corresponding enhanced competition matrix is

[0122]

[0123] Based on the enhanced competition graph, the three methods mentioned above are used for ranking aggregation, and the ranking aggregation results are consistent, which are o1>o2>o3>o4>o5>o6.

[0124] Specifically, the enhanced competition graph replaces the original competition graph and is used as the input data of the above three ranking aggregation methods:

[0125] Borda counting method: sum each row of the enhanced competition matrix to obtain the out-degree of each evaluated object, and sort the elements in the vector from large to small to obtain the aggregate ranking;

[0126] Competition graph method: Sum each row and column of the enhanced competition matrix to obtain the out-degree and in-degree of each evaluated object, and use the ratio of out-degree to in-degree as an indicator to sort from large to small to obtain the aggregate ranking.

[0127] Markov chain method: Take the enhanced competition graph as input, use the PageRank algorithm to calculate the score of each rated object, and sort them from large to small according to the score to obtain the aggregate ranking.

[0128] The present application proposes a ranking aggregation system based on link prediction, the system comprising: a ranking list acquisition unit, a competition graph acquisition unit, a normalization unit and a ranking determination unit;

[0129] A ranking list acquisition unit, used to obtain a ranking list of several evaluated objects;

[0130] A competition graph acquisition unit is used to construct a competition matrix between the evaluated objects according to the ranking list and obtain the corresponding competition graph; adjust the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph;

[0131] A standardization unit, used for standardizing the competition graph to obtain a standardized competition graph;

[0132] The ranking determination unit is used to determine the aggregate ranking according to the enhanced competition graph.

[0133] The present application proposes a readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the following steps:

[0134] Get a ranking list of several rated objects;

[0135] Based on the ranking list, construct a competition matrix between the evaluated objects and obtain the corresponding competition graph;

[0136] The competition graph is standardized to obtain a standardized competition graph;

[0137] Adjust the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph;

[0138] Determine the aggregate ranking based on the enhanced competition graph.

[0139] The present application proposes a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor in the following steps:

[0140] Get a ranking list of several rated objects;

[0141] Based on the ranking list, construct a competition matrix between the evaluated objects and obtain the corresponding competition graph;

[0142] The competition graph is standardized to obtain a standardized competition graph;

[0143] Adjust the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph;

[0144] Determine the aggregate ranking based on the enhanced competition graph.

[0145] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in the present application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0146] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0147] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent of the present application. It should be pointed out that, for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. The above disclosure is only a preferred embodiment of the present invention, and of course it cannot be used to limit the scope of the rights of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope covered by the present invention.

Claims

1. A ranking aggregation method based on link prediction, characterized in that: The method comprises: Get a ranking list of several rated objects; According to the ranking list, a competition matrix between the evaluated objects is constructed, and a corresponding competition graph is obtained; Standardizing the competition graph to obtain a standardized competition graph; Adjust the edge weights in the standardized competition graph according to the link prediction to obtain an enhanced competition graph; An aggregate ranking is determined based on the enhanced competition graph.

2. The ranking aggregation method based on link prediction according to claim 1, characterized in that: The step of obtaining a ranking list of several evaluated objects specifically includes: According to m objects to be evaluated and n judges, a ranking list R of the objects to be evaluated is determined. i =[r i1 ,r i2 …r im ], the ranking list R i The length is L i =|{r ij |r ij >0,1≤j≤m}|, where 1≤L i ≤m; The set of m evaluated objects is preset as: O = {o1, o2…o m }, the set of n voter is: V = {v1, v2…v n }; If the judges v i No evaluation was done on the object j For ranking, then r ij =0; If the ranking list R i The first evaluated object in s With the second evaluated object o t If the ranking is the same, then r is =r it .

3. The ranking aggregation method based on link prediction according to claim 2 is characterized in that: The step of constructing a competition matrix among the evaluated objects according to the ranking list and obtaining a corresponding competition graph specifically includes: Preset the judges v i The preference matrix is In the ranking list R i In the case where the first evaluated object o s The ranking is lower than the second rated object o t , then: r it >r is >0, and in the preference matrix In the ranking list R i In the case where the first evaluated object o s The ranking is greater than the second rated object o t , then: rr is , and in the preference matrix In the ranking list R i In the case where the first evaluated object o s The ranking is equal to the second rated object o t , then: rr it >0, and in the preference matrix According to the preference matrices of the n voters, the competition matrix between the evaluated objects is determined: A=(a st ) n×n ,in, a st For the object of evaluation s Ranked lower than the rated object o t The total number of times.

4. The ranking aggregation method based on link prediction according to claim 3 is characterized in that: The step of constructing a competition matrix among the evaluated objects according to the ranking list and obtaining a corresponding competition graph specifically includes: The competition graph G=(V,E) between the evaluated objects is preset, where V is a node set, The ranking relationship between the rated objects; All nodes in the competition graph correspond one-to-one to the evaluated objects; If a st >0, then there is a line in the competition graph consisting of the evaluated object o s Start from the node and go to the object to be evaluated o t The edge of the node st , and the edge e st The weight is a st .

5. The ranking aggregation method based on link prediction according to claim 4 is characterized in that: The step of normalizing the competition graph to obtain a standardized competition graph specifically includes: The edge e st The weight a st Adjust to in, For the object of evaluation s Defeat the object being evaluated t probability; Determine the standardized competition map based on the adjustment results 6. The ranking aggregation method based on link prediction according to claim 4, characterized in that: The step of adjusting the edge weights in the standardized competition graph according to the link prediction to obtain the enhanced competition graph specifically includes: When there is no node pair with direct edges in the competition graph (o i ,o k ) There are common neighbor nodes o j ,and When, according to Determine the node pair (o i ,o k )’s potential edge ik The partial weight of Obtain all common neighbor nodes in the standardized competition graph, add up the weights corresponding to the common neighbor nodes, and obtain the potential edge e ik The overall weight of: The overall weight is standardized according to the following formula to determine the potential edge e ik The standardized weights are: Potential edges are mined and weights are defined for nodes that meet preset requirements to obtain an enhanced competition graph.

7. The ranking aggregation method based on link prediction according to claim 6, characterized in that: Determining the aggregate ranking according to the enhanced competition graph specifically includes: Counting the weighted out-degree and weighted in-degree of each node in the enhanced competition graph; Determine the score of each node according to the weighted out-degree and the weighted in-degree; All nodes are arranged in ascending or descending order according to the score of each node to obtain an aggregate ranking.

8. A ranking aggregation system based on link prediction, characterized in that: The system comprises: a ranking list acquisition unit, a competition graph acquisition unit, a normalization unit and a ranking determination unit; The ranking list acquisition unit is used to obtain a ranking list of several evaluated objects; The competition graph acquisition unit is used to construct a competition matrix between the evaluated objects according to the ranking list and acquire a corresponding competition graph; adjust the edge weights in the standardized competition graph according to the link prediction to acquire an enhanced competition graph; The standardization unit is used to standardize the competition graph to obtain a standardized competition graph; The ranking determination unit is used to determine an aggregate ranking according to the enhanced competition graph.

9. A readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to execute the steps of the method according to any one of claims 1 to 7.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 7.