Answering and reviewing expert allocation method and related products

Abstract

The invention relates to the field of resource allocation, in particular to a reply and review expert allocation method and related products thereof. The reply and review expert allocation method comprises the following steps: acquiring the number of reply places and the number of reply sessions; obtaining a reply and review expert distribution tensor, a reply person distribution tensor, a reply schedule tensor and the similarity between the scientific research direction of the reply and review expert and the reply material of the reply person; calculating the matching degree between the reply and review experts allocated to each place and each session and the reply person; calculating the matching degree difference between the reply and review experts distributed in all places and all sessions and the reply person; calculating the matching degree difference between the reply and review experts distributed in the same place and the same session and the reply person; and according to the calculated matching degree and difference, selecting the optimal reply and review expert distribution tensor, reply person distribution tensor and reply schedule tensor. According to the invention, the matching degree of the reply review expert and the reply person is improved, so that the development of the reply work is more excellent and efficient.

Description

Technical Field

[0001] The present invention relates to the technical field of resource allocation, and in particular to a method for allocating defense review experts and related products. Background Art

[0002] Defense review is a key link in project review and paper review. The allocation of defense review experts is directly related to the quality, efficiency and fairness of defense review. Taking paper defense as an example, paper defense is an important link in reviewing the quality of graduation papers, and it is increasingly valued by tutors, students and school teaching management departments. In recent years, with the expansion of college enrollment, the allocation of graduate defense has become more and more complicated and difficult. In the traditional model, the grouping of review experts for college paper defense is carried out by the teaching management department in the form of manual matching, and the defense review experts and defense students are selected by manual matching. However, this method has obvious disadvantages: first, it is inefficient. It takes a lot of work to manually organize the research field information of the review experts and the paper information of the defense students, and it is also necessary to match them one by one, which will waste a lot of time and energy; on the other hand, the matching degree is not high. In the process of manual matching, it is inevitable to encounter the influence of personal emotional factors, and it is difficult to achieve appropriate matching when the amount of data is large. These uncertain factors will affect the final matching results. The traditional manual allocation method is time-consuming and labor-intensive, and it is difficult to ensure the matching degree of defense groups. The matching degree between the defense review experts and the research direction of the graduation thesis largely determines the quality of the defense. Therefore, it is necessary to build a defense review expert allocation grouping system. Summary of the invention

[0003] In order to solve the above technical problems, the embodiments of the present invention disclose a method for allocating defense review experts and related products.

[0004] In a first aspect, an embodiment of the present invention provides a method for allocating defense review experts, the method comprising:

[0005] Get the number of defense venues and defense sessions;

[0006] Get the defense review expert allocation tensor x mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp , the defense review experts assign tensor x mtp represents whether the defense review expert m is arranged in the defense team of the tth defense venue and the pth defense session; the respondent allocation tensor represents y stp Indicates whether the respondent s is assigned to the defense team of the tth defense venue and the pth defense session; the defense review schedule tensor indicates whether the tth defense venue and the pth defense session are scheduled for defense;

[0007] Obtain the similarity between the research direction of the defense review experts and the defense materials of the respondent;

[0008] According to the defense review experts, the tensor x is allocated mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp The matching degree between the defense review experts and the respondents assigned to each venue and each session is calculated with the obtained similarity;

[0009] Calculate the difference in matching between the defense review experts and the respondents in all venues and all sessions to obtain the first difference;

[0010] Calculate the difference in matching between the defense review experts and the respondents assigned in the same venue and the same session to obtain the second difference;

[0011] Obtain the average matching degree between each respondent and the defense review expert, and calculate the dispersion degree of the average matching degree; determine the weight of the first difference according to the dispersion degree of the average matching degree;

[0012] For each respondent, the matching degree between the respondent and each defense review expert is obtained to obtain the first matching degree sequence corresponding to the respondent; the respondent is clustered according to the similarity of the first matching degree sequence between the respondent to obtain the respondent cluster; the weight of the second difference is determined according to the clustering degree of each respondent cluster;

[0013] According to the weight of the first difference and the weight of the second difference, the matching degree of the defense review experts and the respondents assigned to all venues and all sessions, the first difference and the second difference are summed up, and the best defense review tensor, respondent allocation tensor, and defense schedule tensor are selected based on the summation results.

[0014] Preferably, the tensor x is allocated according to the defense review experts mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp The matching degree between the defense review experts and the respondents assigned to each venue and each session is calculated based on the obtained similarity, including:

[0015] The matching degree q between the defense review experts and the respondents assigned to each venue and each session is calculated according to the following formula: mstp :

[0016] q mstp =x mtp y stp c ms k tp

[0017] Among them, q mstpThat is, the matching degree between the defense expert m and the respondent s at the tth defense venue and the pth defense session, c ms It indicates the similarity between the research direction of defense review expert m and the defense materials of respondent s;

[0018] The matching degree Q between the defense review experts and the respondents assigned to all sessions in all venues is calculated according to the following formula:

[0019]

[0020] Among them, T represents the number of venues, P represents the number of sessions, M represents the number of defense review experts, and S represents the number of respondents.

[0021] Preferably, the calculation of the matching difference between the defense review experts and the respondents assigned in all venues and all sessions to obtain the first difference includes:

[0022] According to the matching degree between the defense review experts and the respondents assigned to each venue and each session, the difference between the maximum and minimum matching degrees is calculated to obtain the first difference.

[0023] Preferably, the calculation of the matching difference between the defense review experts and the respondents assigned in all venues and all sessions to obtain the first difference includes:

[0024] Calculate the average matching degree between the defense review experts and the respondents in the same venue and session to obtain the first average matching degree corresponding to the venue and session;

[0025] Calculate the mean of the first average matching degree to obtain a second average matching degree;

[0026] A first difference is obtained according to the difference between the first average matching degree and the second average matching degree.

[0027] Preferably, the calculation of the matching difference between the defense review experts and the respondent assigned in the same venue and the same session to obtain the second difference includes:

[0028] Obtain the matching degree between each defense review expert and each respondent assigned in the same venue and the same session;

[0029] Calculate the difference between the maximum and minimum values ​​of the matching degree in the same venue and the same session to obtain the third difference corresponding to the venue and the session;

[0030] The second difference is obtained based on the third differences corresponding to all venues and all sessions.

[0031] Preferably, the respondent allocates the tensor y stp The following conditions should be met:

[0032] when When it is an integer:

[0033]

[0034] when When not an integer:

[0035]

[0036] in, Indicates rounding up and rounding down respectively.

[0037] Preferably, the defense review expert assigns the tensor x mtp The following conditions should be met:

[0038] when hour:

[0039]

[0040] Otherwise when hour:

[0041]

[0042] Among them, α1 represents the upper limit of the defense review experts in each defense session.

[0043] In a second aspect, an embodiment of the present invention further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a method for allocating defense review experts when executing the computer program.

[0044] In a third aspect, an embodiment of the present invention further provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed, the steps of a method for allocating defense review experts are implemented.

[0045] The technical solution provided by the embodiment of the present invention has at least the following advantages compared with the prior art:

[0046] The method for allocating defense review experts of the present invention obtains the number of defense venues and the number of defense sessions; obtains the defense review expert allocation tensor, the respondent allocation tensor, the defense schedule tensor, and the similarity between the defense review expert's scientific research direction and the respondent's defense materials; calculates the matching degree of the defense review experts and the respondent assigned to each venue and each session; calculates the difference in matching degree between the defense review experts and the respondent assigned to different venues and different sessions; calculates the difference in matching degree between the defense review experts and the respondent assigned to the same venue and the same session; selects the best defense review expert allocation tensor, respondent allocation tensor, and defense schedule tensor according to the calculated matching degree and difference, so as to obtain the best allocation plan, which can not only greatly reduce the human cost of defense grouping, but also improve the matching degree between the defense review experts and the respondent to a greater extent, improve the scheduling efficiency and quality of human resources, and make the defense work more high-quality, efficient and fair.

[0047] The prior art does not consider the matching degree between the defense review experts and the respondent's materials, and only considers time conflicts and manpower conflicts to allocate human resources. The present invention introduces the matching degree between the defense review experts and the respondent, the difference in matching degree between the defense review experts and the respondent assigned in different places and different sessions, and the average matching degree between the defense review experts and the respondent in the same place and the same session to evaluate the effect of various allocation schemes, improve the matching degree between the defense object and the review object, and the review group with high matching degree can analyze and evaluate the respondent's topic more comprehensively and accurately, and improve the efficiency, rigor and fairness of the defense. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The figure is an exemplary flow chart of a method for allocating defense review experts according to the first embodiment of the present invention. DETAILED DESCRIPTION

[0049] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0050] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0051] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the specific embodiments below. Those skilled in the art should understand that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure. A method for allocating defense review experts provided in an embodiment of the present invention.

[0052] Figure 1 FIG. 1 is a flowchart of a method for allocating defense review experts according to a first embodiment of the present invention. Figure 1 As shown, the method includes step S11, step S12, step S13, and step S14.

[0053] In step S11, the number of defense venues and the number of defense sessions are obtained; the defense review expert allocation tensor x is obtained. mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp , the defense review experts assign tensor x mtp represents whether the defense review expert m is arranged in the defense team of the tth defense venue and the pth defense session; the respondent allocation tensor represents y stp Indicates whether the respondent s is assigned to the defense group of the t-th defense venue and the p-th defense session; the defense review schedule tensor indicates whether the t-th defense venue and the p-th defense session are scheduled for defense; obtains the similarity between the scientific research direction of the defense review expert and the respondent's defense materials.

[0054] Obtaining the similarity between the research direction of the defense review experts and the defense materials of the respondent includes: obtaining the research information of the defense review experts; accurately extracting keywords and establishing a research direction model; and calculating the similarity matching degree between the defense review experts and the respondent.

[0055] Taking thesis defense as an example, a specific implementation method is listed below. Step 1: Acquisition of scientific research information of review experts, using data crawling technology. Select databases such as China Knowledge Network CNKI and Web of Science as data sources for obtaining academic papers, and index the published academic papers through information such as the name and affiliated unit of the review experts. Select Python language to crawl the academic paper data of the review experts, parse and extract the required scientific research information and store it in the database. Preferably, the above-mentioned academic paper data source is only for example, and the corresponding data source can be selected according to the specific review expert paper situation. Selecting Python language for data crawling is only for example, and the corresponding development language can be selected according to the specific system development implementation, which does not affect the system method. Step 2: Extraction of scientific research information keywords and establishment of review expert scientific research direction model. Use a pre-trained natural language processing model to extract mixed keyword groups of single words and double words from the target text, and calculate the weight value corresponding to each keyword. The establishment of the review expert scientific research direction model is to first use a pre-trained natural language processing model to extract article keywords and corresponding weights for each paper of the review expert, and then merge the keyword groups of all papers of the expert for keywords to calculate new keyword weights. In the same way, keywords in the thesis of the defending students are extracted, and the weight of each keyword is calculated. The extraction of mixed keyword groups of single words and double words refers to two keyword extraction strategies. In the single word mode, the keyword "agricultural product supply chain" will be divided into two keywords "agricultural product" and "supply chain", and the combined keyword "agricultural product supply chain" is retained in the double word mode. Mixed keyword group extraction can retain the keyword groups divided in both modes at the same time, and express the text content to a greater extent. Construct the scientific research direction model of the corresponding experts based on keywords and weights. Optionally, for Chinese academic papers, use Python's Jieba library to divide the abstract text of Chinese academic papers into words. Use spaces to separate the words and construct an English-like language structure. Use the spaCy library to initialize and load the pre-trained natural language processing model, use the KeyBERT library based on the SCKKRS algorithm, extract the mixed keyword groups of single words and double words for the target text, and calculate the weight value corresponding to each keyword. Step 3: Calculate the similarity between the keyword group of the student's graduation thesis and the expert's scientific research direction model. Use the cosine theorem formula of the vector space model to calculate the similarity between the two. Preferably, the algorithm used for similarity calculation is the cosine theorem formula of the vector space model. The algorithm converts the text into a space vector, and then uses the cosine theorem formula to calculate the angle between the vectors, and uses this to determine the similarity of the text. Generally speaking, the smaller the angle, the more similar the text. The cosine similarity value range is [0,1].The cosine similarity can be normalized for subsequent matching calculations using the method (cosine similarity + 1) / 2. The result is the matching between the defense review expert and the respondent, that is, the similarity c between the defense review expert m and the defense materials of the respondent s. ms .

[0056] In step S12, the tensor x is assigned according to the defense review experts. mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp The matching degree between the defense review experts and the respondents assigned to each venue and each session is calculated based on the obtained similarity.

[0057] Specifically, the matching degree q between the defense review experts and the respondents assigned to each venue and each session is calculated according to the following formula: mstp :

[0058] q mstp =x mtp y stp c ms k tp

[0059] Among them, q mstp That is, the matching degree between the defense expert m and the respondent s at the tth defense venue and the pth defense session, c ms It indicates the similarity between the research direction of defense review expert m and the defense materials of respondent s.

[0060] The matching degree Q between the defense review experts and the respondents assigned to all sessions in all venues is calculated according to the following formula:

[0061]

[0062] Wherein, T represents the number of venues, P represents the number of sessions, M represents the number of defense review experts, and S represents the number of respondents. The obtained summation result is used for summation in step S14.

[0063] In step S13, the matching difference between the defense review experts and the respondents assigned in all venues and all sessions is calculated to obtain a first difference; the matching difference between the defense review experts and the respondents assigned in the same venue and the same session is calculated to obtain a second difference.

[0064] In step S14, based on the calculated matching degree, the first difference and the second difference, the best defense review expert allocation tensor, respondent allocation tensor and defense schedule tensor are selected.

[0065] Optionally, based on the above embodiment, step S14 includes:

[0066] Step S141, obtaining the average matching degree between each respondent and the defense review experts, and calculating the dispersion degree of the average matching degree.

[0067] Without considering the venue and the number of sessions, for each respondent, the matching degree between the respondent and each defense review expert (the similarity between the defense materials and the expert's scientific research direction) is calculated, and then the average of the matching degrees is calculated to obtain the average matching degree corresponding to the respondent. It should be noted that the matching degree between each respondent and each defense review expert here has nothing to do with the defense review expert allocation tensor and the respondent allocation tensor, and is not used to calculate the objective function value and fitness, but is used to measure the matching degree between each respondent and each defense review expert.

[0068] Get the average matching degree corresponding to all respondents, and calculate the dispersion degree of all average matching degrees. Optionally, the dispersion degree is characterized by mean square error.

[0069] Step S142: determining the weight of the first difference according to the dispersion degree of the average matching degree.

[0070] The higher the dispersion of the average matching degree, the higher the probability that the matching degree of the defense review experts to the respondent's materials tends to be dispersed. In this case, the weight of the first difference should be lower to ensure the quality of the defense review. Conversely, the lower the dispersion of the average matching degree, the higher the weight of the first difference should be to take into account the fairness of the defense review.

[0071] Optionally, the weight of the first difference is: w1 = α1 - β1*(1 - σ x ). Wherein, w1 is the weight of the first difference, α1 is the reference weight, β1 is the adjustable weight coefficient, σ is the constant base, and x is the discrete degree of the average matching degree. Optionally, the initial values ​​of α1, β1, and σ are 1, 0.35, and 0.75, respectively.

[0072] Step S143, for each respondent, obtain the matching degree between the respondent and each defense review expert, and obtain the first matching degree sequence corresponding to the respondent.

[0073] The first matching degree sequence stores the matching degree between the respondent and each defense review expert.

[0074] Step S144, clustering the respondents according to the similarity of the first matching degree sequence between the respondents to obtain the respondent clusters.

[0075] Calculate the similarity of the first matching sequence corresponding to each pair of respondents. Optionally, use Euclidean distance to calculate the similarity.

[0076] The respondents are clustered according to the similarity of their corresponding sequences. Optionally, the k-nearest neighbor clustering method is used for clustering. After clustering, the respondent clusters can be obtained.

[0077] Step S145, determining the weight of the second difference according to the degree of aggregation of each respondent cluster.

[0078] The higher the degree of aggregation of each respondent cluster, the higher the probability that the matching degree between the defense review experts and the respondent in each defense session is close, and the higher the weight of the second difference is, so as to ensure the fairness of the defense review. Conversely, the lower the weight of the second difference is, the higher the quality of the defense review is. Optionally, the silhouette coefficient is used to characterize the degree of aggregation of the respondent cluster.

[0079] The value range of the silhouette coefficient is [-1,1]. The degree of clustering of the respondent cluster is normalized, and then the normalized value is subtracted by 1. Optionally, the weight of the second difference is: w2 = α2 + β2 * y. Among them, w1 is the weight of the first difference, α2 is the reference weight, β2 is the adjustable weight coefficient, and y is the value obtained by subtracting 1 from the normalized degree of clustering of the respondent cluster. Optionally, the initial values ​​of α2 and β2 are 1 and 0.35 respectively.

[0080] Step S146, summing the calculated matching degree, the first difference and the second difference according to the weight of the first difference and the weight of the second difference, and selecting the best defense review tensor, respondent allocation tensor and defense schedule tensor according to the summation result.

[0081] The present invention aims to achieve a balance between maximizing the degree of matching, minimizing the first difference, and minimizing the second difference. Therefore, when performing weighted summation, it is necessary to reverse normalize the degree of matching, and normalize the first difference and the second difference. The reverse normalization here is (maximum value-current value) / (maximum value-minimum value), that is, the maximum value corresponds to 0 and the minimum value corresponds to 1. Then, according to the weight of the first difference, the weight of the second difference, and the weight of the sum of the degree of matching Q, the first difference, the second difference, and the degree of matching are weighted and summed. Optionally, the weight of the sum of the degree of matching Q is 1. According to the summation result, the tensor corresponding to the minimum summation result is selected as the best defense review expert allocation tensor, the respondent allocation tensor, and the defense schedule tensor. The number of each tensor obtained in step one is multiple, that is, it includes multiple groups of different defense review expert allocation tensors, respondent allocation tensors, and defense schedule tensors. It should be noted that the weight of the first difference, the weight of the second difference, and the weight of the sum of the degree of matching Q are adjusted according to the specific implementation data volume. The weight examples given in this embodiment are given based on the fact that the value ranges of the first difference, the second difference, and the sum of the matching degree are basically consistent. In actual implementation, the first difference, the second difference, and the sum of the matching degree can be normalized and then weighted summed according to their respective weights.

[0082] Optionally, based on the above embodiment, the step of calculating the difference in matching degrees between the defense review experts and the respondents assigned in all venues and all sessions to obtain the first difference includes: calculating the difference between the maximum and minimum values ​​of the matching degrees based on the matching degrees between the defense review experts and the respondents assigned in each venue and each session to obtain the first difference.

[0083] According to the defense review expert allocation tensor and the respondent allocation tensor, we can know the defense review experts and respondent in each venue and each session, obtain the matching degree of each defense review expert and respondent in the allocation result, select the maximum and minimum values ​​from them, and the first difference is the difference between the maximum and minimum values.

[0084] Optionally, on the basis of the first embodiment, the step of calculating the difference in matching between the defense review experts and the respondents assigned in all places and all sessions to obtain the first difference includes: calculating the average matching between the defense review experts and the respondents in the same place and the same session to obtain the first average matching corresponding to the place and the session. Calculating the mean of the first average matching to obtain the second average matching. According to the difference between the first average matching and the second average matching, the difference in matching between the defense review experts and the respondents assigned in all places and all sessions to obtain the first difference.

[0085] Optionally, the difference between the first average matching degree and the second average matching degree is represented by a mean square error.

[0086] On the basis of the first embodiment, optionally, the step of calculating the difference in matching between the defense review experts and the respondents assigned in the same place and the same session includes: obtaining the matching between each defense review expert and each respondent assigned in the same place and the same session. Calculating the difference between the maximum and minimum matching values ​​in the same place and the same session to obtain a third difference corresponding to the place and the session. According to the third differences corresponding to all places and all sessions, the second difference is obtained.

[0087] Specifically, the maximum value of the third differences may be selected as the second difference, or the sum of the third differences may be selected as the second difference.

[0088] Based on the first embodiment, optionally, the respondent allocates the tensor y stp The following conditions should be met:

[0089] The upper and lower limits of the number of review experts in each group must meet the requirements and are subject to constraints:

[0090]

[0091] Among them, α1 and α2 are the upper and lower limits of the number of review experts for each defense team.

[0092] when When it is an integer:

[0093]

[0094] when When not an integer:

[0095]

[0096] Where S represents the number of respondents.

[0097] Based on the first embodiment, optionally, the defense review expert assigns the tensor x mtp The following conditions should be met:

[0098] when hour:

[0099]

[0100] Otherwise when hour:

[0101]

[0102] Among them, α1 represents the upper limit of the number of defense review experts in each defense session. In order to ensure that the number of review experts in each defense group is as large as possible within the required range, and the number of review experts in each defense group in the same session is as equal as possible. Among them, p1 and p2 represent the session, and the value ranges from 1 to P.

[0103] In order to ensure the workload and quality of each defense group, it is necessary to balance the defense time of each group. Generally speaking, the school stipulates the time for students to make statements and experts to ask questions, so the defense time of the defense group is determined by the number of students defending. It is necessary to ensure that the number of students in each defense group is as equal as possible, with constraints:

[0104] when When it is an integer:

[0105]

[0106] Otherwise when When not an integer:

[0107]

[0108] Each review expert can only participate in one defense session, with the following restrictions:

[0109]

[0110] Each defense student must participate in the allocation and can only be assigned to one defense group, with the following restrictions:

[0111]

[0112] The decision variables of the review experts and the defense students are 0-1 variables with constraints:

[0113]

[0114] In undergraduate and graduate defenses, the restrictions on whether the supervisor can serve as the student's review expert are different. It is necessary to decide whether to have the following restrictions based on the actual situation:

[0115] If the evaluation experts and the defense students cannot have a teacher-student relationship, there are restrictions:

[0116]

[0117] Among them, a ms It is a Boolean variable. If it is 1, it means there is a teacher-student relationship between the review expert m and the defense student s, otherwise it is 0.

[0118] In order to solve the model more efficiently, a genetic algorithm is introduced.

[0119] Optionally, based on the first embodiment, in step S11, the defense review expert allocation tensor x is obtained. mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp Including: randomly generate the initial population, obtain the defense review expert allocation tensor x based on the initial population mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp .

[0120] Given the expert review grouping scheme, that is, x mtp becomes a constant, the objective function can be transformed into an integer programming model.

[0121] First, design the chromosome form. The chromosome uses each chromosome group to represent a review expert grouping scheme. Each chromosome in the chromosome group represents the grouping of a defense. The gene on the chromosome is the ID number of the review expert corresponding to the corresponding expert. For example, there are two defenses with two sessions each, including 6 defense experts, and two review experts in each session. One form of the chromosome representing the defense review expert allocation tensor is: [[[4,5],[2,3]],[[2,3],[1,4]]]. It means that the first session of the first venue includes experts 4 and 5, the second session includes experts 2 and 3, the first session of the second venue includes experts 2 and 3, and the second session of the second venue includes experts 1 and 4. Chromosome form 2 is similar to form 1. Each chromosome group is used to represent a review expert grouping scheme. Each chromosome in the chromosome group represents the grouping of a defense. However, the gene is represented in the form of 0-1 coding. The gene value of the corresponding position is 1, which means that the corresponding tutor is assigned to the defense group. For example, if there are two defenses with two sessions each and two reviewers in each session, the chromosomes are: [[[0,0,0,1,1,0],[0,1,1,0,0,0]],[[0,1,1,0,0,0],[1,0,0,1,0,0]]] (two forms of the same grouping scheme as the example chromosomes in form one). The respondent allocation tensor also uses the same chromosome form as above. Using the chromosome group of chromosome form one to represent the matching scheme of a reviewer, randomly generate chromosome groups of a specified population size to form an initial population. According to the matching grouping model, ensure that each chromosome group in the population meets the constraints of the model. The defense schedule tensor is reasonably assigned according to the defense review expert allocation tensor and the respondent allocation tensor, and the value assigned should be 1 only when there are both experts and respondent.

[0122] In the calculation of steps S12 and S13, the chromosome group coding in the population is converted to chromosome form two, so that the gene position is in a 0-1 coding form. The value of the model objective function is solved when the chromosome group is determined, that is, the sum of the matching degree, the weighted sum of the first difference and the second difference. Then, this objective function value and the solved defense student allocation plan are recorded in the chromosome fitness library. All chromosome groups in the population are solved using the same method and recorded in the chromosome fitness library.

[0123] After the introduction of the genetic algorithm, after step S13 and before step S14, the following steps are also included: performing selection, crossover, and mutation operator operations on the initial population to obtain more mutant individuals, and calculating the matching degree, first difference, and second difference corresponding to each chromosome.

[0124] Selection operator operation. In order to ensure the accuracy of the solution to a greater extent and prevent the population iteration from entering premature convergence, the present invention selects the tournament selection method as the selection operator. By extracting a certain number of individuals from the population each time, making them compete with each other, the best individual is selected to enter the offspring population. Repeat this operation until an offspring population of the same population size is obtained. Crossover operator operation. The chromosome group encoding in the population is converted into chromosome morphology one, and the chromosomes of the individual pairs in the population are subjected to two-point crossover according to the pre-set crossover probability to generate offspring individuals with different chromosome groups. In order to ensure the health of the generated new chromosomes, the crossover is only performed on the corresponding chromosomes in the chromosome groups of the two individuals to prevent the crossover of different pairs of chromosomes. Because the chromosomes of the new individual may not meet the constraints after the crossover, gene correction is also required. Mutation operator operation. The mutated individuals are selected from all the individual chromosome groups in the population according to the mutation probability, and single-point gene mutation is performed. Two gene points are randomly selected from the chromosome group of the mutated individual, and the genes of the two points are exchanged to form a new mutated individual. Because mutation may cause chromosomes to be inconsistent with constraints, it is also necessary to check the chromosomes at the mutation sites. If there is a conflict, gene correction is required. Population iteration. The selection, crossover, and mutation operator operations are cycled for a preset number of iterations to obtain the individual with the best fitness value (i.e., the best objective function value) in the population. The grouping scheme of the review experts and students corresponding to this individual is the grouping scheme finally solved by the genetic algorithm.

[0125] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for allocating defense review experts, characterized in that: The method comprises: Get the number of defense venues and defense sessions; Get the defense review expert allocation tensor x mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp , the defense review experts assign tensor x mtp represents whether the defense review expert m is arranged in the defense team of the tth defense venue and the pth defense session; the respondent allocation tensor represents y stp Indicates whether the respondent s is assigned to the defense team of the tth defense venue and the pth defense session; the defense review schedule tensor k tp Indicates whether the defense is scheduled at the tth defense venue and the pth defense session; Obtain the similarity between the research direction of the defense review experts and the defense materials of the respondent; According to the defense review experts, the tensor x is allocated mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp The matching degree between the defense review experts and the respondents assigned to each venue and each session is calculated with the obtained similarity; Calculate the difference in matching between the defense review experts and the respondents in all venues and all sessions to obtain the first difference; Calculate the difference in matching between the defense review experts and the respondents assigned in the same venue and the same session to obtain the second difference; Obtain the average matching degree between each respondent and the defense review expert, and calculate the dispersion degree of the average matching degree; determine the weight of the first difference according to the dispersion degree of the average matching degree; For each respondent, the matching degree between the respondent and each defense review expert is obtained to obtain the first matching degree sequence corresponding to the respondent; the respondent is clustered according to the similarity of the first matching degree sequence between the respondent to obtain the respondent cluster; the weight of the second difference is determined according to the clustering degree of each respondent cluster; According to the weight of the first difference and the weight of the second difference, the matching degree of the defense review experts and the respondent assigned to all sessions in all venues, the first difference and the second difference are summed up, and the best defense review expert allocation tensor, respondent allocation tensor, and defense schedule tensor are selected based on the summation results.

2. A method for allocating defense experts as claimed in claim 1, characterized in that: According to the defense review experts, the tensor x is allocated mtp , the respondent allocates tensor y stp 、Defense schedule tensor k tp The matching degree between the defense review experts and the respondents assigned to each venue and each session is calculated based on the obtained similarity, including: The matching degree q between the defense review experts and the respondents assigned to each venue and each session is calculated according to the following formula: mstp : q mstp =x mtp y stp c ms k tp Among them, q mstp That is, the matching degree between the defense expert m and the respondent s at the tth defense venue and the pth defense session, c ms It indicates the similarity between the research direction of defense review expert m and the defense materials of respondent s; The matching degree Q between the defense review experts and the respondents assigned to all sessions in all venues is calculated according to the following formula: Among them, T represents the number of venues, P represents the number of sessions, M represents the number of defense review experts, and S represents the number of respondents.

3. A method for allocating defense review experts as claimed in claim 1, characterized in that: The calculation of the matching degree difference between the defense review experts and the respondents assigned in all venues and all sessions to obtain the first difference includes: According to the matching degree between the defense review experts and the respondents assigned to each venue and each session, the difference between the maximum and minimum matching degrees is calculated to obtain the first difference.

4. A method for allocating defense review experts as claimed in claim 1, characterized in that: The calculation of the matching degree difference between the defense review experts and the respondents assigned in all venues and all sessions to obtain the first difference includes: Calculate the average matching degree between the defense review experts and the respondents in the same venue and session to obtain the first average matching degree corresponding to the venue and session; Calculate the mean of the first average matching degree to obtain a second average matching degree; A first difference is obtained according to the difference between the first average matching degree and the second average matching degree.

5. The method for allocating defense review experts according to claim 1, characterized in that: The second difference obtained by calculating the matching difference between the defense review experts and the respondents assigned in the same venue and the same session includes: Obtain the matching degree between each defense review expert and each respondent assigned in the same venue and the same session; Calculate the difference between the maximum and minimum values ​​of the matching degree in the same venue and the same session to obtain the third difference corresponding to the venue and the session; The second difference is obtained based on the third differences corresponding to all venues and all sessions.

6. A method for allocating defense review experts as claimed in claim 3, characterized in that: The respondent allocates the tensor y stp The following conditions should be met: when When it is an integer: when When not an integer: in, Indicates rounding up and rounding down respectively.

7. A method for allocating defense review experts as claimed in claim 2, characterized in that: The defense review experts assign tensor x mtp The following conditions should be met: when hour: Otherwise when hour: Among them, α1 represents the upper limit of the defense review experts in each defense session.

8. A computer device comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the processor executes the computer program, the steps of a method for allocating defense review experts according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium, wherein a computer program is stored in the storage medium, characterized in that: When the computer program is executed, the steps of a method for allocating defense review experts according to any one of claims 1 to 7 are implemented.

Images