Website data processing method, device and equipment

Abstract

The embodiment of the invention provides a website data processing method and device, equipment and a storage medium, and the method comprises the steps: obtaining website instance data which comprises a plurality of website instances; the initialized particle swarm is evolved through a competitive particle swarm algorithm and website instance data, and particles in the particle swarm represent an instance selection scheme of the website instance data; and according to the evolved particle swarm, instance selection is carried out in the website instance data to obtain a target instance subset, and the target instance subset is used for classification model training of phishing websites. Therefore, the instance selection scheme in the website instance data is optimized through the competitive particle swarm algorithm, and the instance subset used for classification model training of the phishing website is selected from the website instance data according to the optimized instance selection scheme, so that the quality of training data of the classification model is improved; the accuracy of phishing website detection based on the classification model is improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technologies, and in particular, to a method, apparatus, device, and storage medium for processing website data. Background Art

[0002] The existence of phishing websites poses certain security risks to users when surfing the Internet. With the development of artificial intelligence technologies, artificial intelligence technologies can be used to detect phishing websites.

[0003] In related technologies, a large amount of website data is obtained, and an artificial intelligence model is trained based on the website data to obtain a classification model for phishing websites, and then the classification model is used to detect phishing websites.

[0004] However, website data has disadvantages such as a large amount of noisy data and data redundancy, resulting in poor performance of the trained classification model, thereby affecting the detection accuracy of phishing websites. Summary of the Invention

[0005] Embodiments of the present disclosure provide a method, apparatus, and device for processing website data to solve the problem that the website data has disadvantages such as a large amount of noisy data and data redundancy, resulting in poor performance of the trained classification model, thereby affecting the detection accuracy of phishing websites.

[0006] In a first aspect, embodiments of the present disclosure provide a method for processing website data, including: obtaining website instance data, where the website instance data includes multiple website instances; evolving an initialized particle swarm through a competitive particle swarm algorithm and the website instance data, where particles in the particle swarm represent instance selection schemes for the website instance data; and performing instance selection in the website instance data according to the evolved particle swarm to obtain a target instance subset, where the target instance subset is used for training a classification model for phishing websites.

[0007] In some embodiments, one evolution process of the particle swarm includes: selecting instances from the website instance data according to the instance selection schemes represented by the particles in the particle swarm to obtain the subsets of instances corresponding to the particles in the particle swarm; evaluating the subsets of instances corresponding to the particles in the particle swarm in the target dimension related to instance selection to obtain the target values corresponding to the particles in the particle swarm in the target dimension; performing a clustering operation on the particles in the particle swarm according to the target values corresponding to the particles in the particle swarm in the target dimension to obtain a plurality of particle clusters; comparing the particles in the plurality of particle clusters respectively to determine the winner particles and the loser particles in the plurality of particle clusters; evolving the winner particles in the plurality of particle clusters according to the evolution strategy corresponding to the winner particles to obtain the evolved particles corresponding to the winner particles in the plurality of particle clusters; evolving the loser particles in the plurality of particle clusters according to the evolution strategy corresponding to the loser particles to obtain the evolved particles corresponding to the loser particles in the plurality of particle clusters; and performing an environmental selection operation on the particles in the particle swarm, the evolved particles corresponding to the winner particles in the plurality of particle clusters, and the evolved particles corresponding to the loser particles in the plurality of particle clusters to determine the particles in the particle swarm after the current evolution.

[0008] In some embodiments, the target dimension includes a first dimension representing the fishing network detection error and / or a second dimension representing the subset complexity. The evaluating the subsets of instances corresponding to the particles in the particle swarm in the target dimension related to instance selection to obtain the target values corresponding to the particles in the particle swarm in the target dimension includes: using the subsets of instances corresponding to the particles in the particle swarm as a plurality of independent training data sets to train the classification model respectively to obtain the model training errors of the subsets of instances corresponding to the particles in the particle swarm; determining the target values corresponding to the particles in the particle swarm in the first dimension according to the model training errors of the subsets of instances corresponding to the particles in the particle swarm; and / or determining the target values corresponding to the particles in the particle swarm in the second dimension according to the complexity-related features of the subsets of instances corresponding to the particles in the particle swarm.

[0009] In some embodiments, clustering the particles in the particle swarm according to the target values corresponding to the particles in the particle swarm in the target dimension to obtain a plurality of particle clusters includes: determining the angles between the particles in the particle swarm and a plurality of reference vectors according to the target values corresponding to the particles in the particle swarm in the target dimension, where the plurality of reference vectors are uniformly distributed in the target space formed by the target dimension; clustering the plurality of particles into a plurality of initial clusters according to the angles between the particles in the particle swarm and the plurality of reference vectors, where the plurality of initial clusters correspond one-to-one to the plurality of reference vectors; and adjusting the particles in the initial clusters with an actual particle number greater than the cluster capacity and the particles in the initial clusters with an actual particle number less than the cluster capacity according to the plurality of initial clusters and the plurality of reference vectors to obtain the plurality of particle clusters.

[0010] In some embodiments, comparing the particles in the plurality of particle clusters respectively to determine the winner particles and the loser particles in the plurality of particle clusters includes: for any one of the plurality of particle clusters, determining the target value corresponding to the ideal particle corresponding to the any one of the particle clusters in the target dimension according to the target value corresponding to the particles in the any one of the particle clusters in the target dimension; determining the Chebyshev value of the particles in the any one of the particle clusters according to the target value corresponding to the particles in the any one of the particle clusters in the target dimension, the target value corresponding to the ideal particle corresponding to the any one of the particle clusters in the target dimension, and the plurality of reference vectors; and determining the winner particles and the loser particles in the any one of the particle clusters by comparing the Chebyshev values of the particles in the any one of the particle clusters.

[0011] In some embodiments, the evolutionary strategy corresponding to the winner particles is an evolutionary strategy based on a Gaussian model. Evolving the winner particles in the plurality of particle clusters according to the evolutionary strategy corresponding to the winner particles to obtain the evolutionary particles corresponding to the winner particles in the plurality of particle clusters includes: constructing the Gaussian model, where the mean matrix in the Gaussian model and the covariance matrix in the Gaussian model are determined according to the position matrix of the winner particles in the plurality of particle clusters; performing random sampling according to the Gaussian model to obtain an updated particle position matrix; and guiding the evolution of the winner particles in the plurality of particle clusters according to the updated particle position matrix to obtain the evolutionary particles corresponding to the winner particles in the plurality of particle clusters.

[0012] In some embodiments, the evolutionary strategy corresponding to the loser particles is an evolutionary strategy based on constraint conditions, where the constraint conditions include in-domain learning conditions and inter-domain learning conditions. Evolving the loser particles in the multiple particle clusters according to the evolutionary strategy corresponding to the loser particles to obtain the evolved particles corresponding to the loser particles in the multiple particle clusters includes: when the current evolution satisfies the in-domain learning conditions, for the same particle cluster among the multiple particle clusters, guiding the evolution of the loser particles in the same particle cluster according to the position matrix of the winner particles in the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster; when the current evolution satisfies the inter-domain learning conditions, for the same particle cluster among the multiple particle clusters, guiding the evolution of the loser particles in the same particle cluster according to the position matrix of the winner particles in the same particle cluster and the position matrix of the winner particles in the adjacent clusters of the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster; where the in-domain learning conditions include that the conditional probability value in the current evolution is less than the probability threshold, and the inter-domain learning conditions include that the conditional probability value is greater than or equal to the probability threshold, and the conditional probability value is randomly generated in the current evolution.

[0013] In some embodiments, performing an environmental selection operation on the particles in the particle swarm, the evolved particles corresponding to the winner particles in the multiple particle clusters, and the evolved particles corresponding to the loser particles in the multiple particle clusters to determine the particles in the particle swarm after the current evolution includes: combining the particles in the particle swarm, the evolved particles corresponding to the winner particles in the multiple particle clusters, and the evolved particles corresponding to the loser particles in the multiple particle clusters into a temporary swarm; determining the non-dominance relationship of the particles in the temporary swarm according to the objective values of the particles in the temporary swarm in the target dimension; stratifying the particles in the temporary swarm according to the non-dominance relationship of the particles in the temporary swarm to determine the non-dominance layer where the particles in the temporary swarm are located; and selecting the particles after the current evolution in the temporary swarm according to the non-dominance layer where the particles in the temporary swarm are located and the objective values of the particles in the temporary swarm in the target dimension.

[0014] In a second aspect, an embodiment of the present disclosure provides a website data processing device, including: an acquisition unit, configured to acquire website instance data, where the website instance data includes multiple website instances; an evolution unit, configured to evolve an initialized particle swarm through a competitive particle swarm algorithm and the website instance data, where the particles in the particle swarm represent instance selection schemes of the website instance data; and a selection unit, configured to perform instance selection in the website instance data according to the evolved particle swarm to obtain a target instance subset, where the target instance subset is used for training a classification model of phishing websites.

[0015] In a third aspect, an embodiment of the present disclosure provides an electronic device, including: at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the website data processing method described in the first aspect above.

[0016] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium, in which computer-executable instructions are stored, and when a processor executes the computer-executable instructions, the website data processing method described in the first aspect above is implemented.

[0017] In a fifth aspect, an embodiment of the present disclosure provides a computer program product, which includes computer-executable instructions, and when a processor executes the computer-executable instructions, the website data processing method described in the first aspect above is implemented.

[0018] In the website data processing method, apparatus and device provided by the embodiments of the present disclosure, the particles in the particle swarm represent the instance selection scheme of website instance data, and the particle swarm is evolved through the competitive particle swarm algorithm to optimize the instance selection scheme of website instance data. According to the evolved particle swarm, instance selection is performed in the website instance data, which improves the accuracy of instance selection and can select representative instances from the website instance data. The instance subset composed of these instances has greatly reduced data in terms of noise, redundant data, etc. compared with the website instance data. Using these instance subsets for the training of the phishing website classification model can effectively improve the detection accuracy of the classification model for phishing websites. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present disclosure. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 It is a schematic flowchart of the website data processing method provided by the embodiment of the present disclosure; Figure 2 It is a schematic flowchart of an optimization process of the particle swarm in the website data processing method provided by the embodiment of the present disclosure; Figure 3 It is an example diagram of the evolution process of the particle swarm; Figure 4 It is a structural block diagram of the website data processing device provided by the embodiment of the present disclosure; Figure 5 Schematic diagram of the hardware structure of the electronic device provided by the embodiments of the present disclosure. Detailed implementation manners

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are some but not all of the embodiments of the present disclosure. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present disclosure without creative efforts shall fall within the protection scope of the present disclosure.

[0022] First, the related technologies involved in the embodiments of the present disclosure will be introduced.

[0023] Phishing website detection refers to identifying whether a website is a phishing website based on the characteristic information of the website. A classification model can be trained through a large amount of website data and used for phishing website detection.

[0024] Instance selection refers to selecting a representative or discriminative subset from a large-scale dataset to reduce data redundancy, reduce noise interference, and improve the computational efficiency of subsequent data processing.

[0025] The competitive particle swarm optimization (CPSO) is a stochastic optimization algorithm based on swarm intelligence. This algorithm combines the advantages of the particle swarm optimization algorithm and the competitive learning algorithm. Specifically, this algorithm maps the solution space of the problem to the search space of particles, and solves the problem by evolving particles in the search space by combining the competitive learning strategy.

[0026] In the related technologies of phishing website detection, the website data used for training the classification model has disadvantages such as a large amount of noise data and data redundancy, resulting in a decrease in the training efficiency of the classification model and poor training effects of the classification model, and the detection accuracy of the classification model for phishing websites is relatively low.

[0027] To solve the above problems, the embodiments of the present disclosure combine instance selection and the competitive particle algorithm, optimize the instance selection scheme through the competitive particle algorithm, perform instance selection on the website instance data according to the optimized instance selection scheme, and obtain a representative and less redundant instance subset. Such an instance subset is used for training the classification model of phishing websites, which can effectively improve the training efficiency and training effects of the classification model and improve the detection accuracy of the classification model for phishing websites.

[0028] Figure 1 Schematic diagram of the process of the website data processing method provided by the embodiments of the present disclosure. AsFigure 1 As shown, the website data processing method includes: S101, obtaining website instance data, where the website instance data includes multiple website instances.

[0029] Among them, one website instance corresponds to one website. A website instance refers to website sample data, and the website instance includes the feature information of the website, such as the internet protocol (IP) address, the length of the uniform resource locator (URL), website traffic, etc.

[0030] Optionally, the website instance further includes the website label of the website, and the website label indicates whether the website is a phishing website. In this way, the subsequent subset of instances can be used for the supervised training of the classification model.

[0031] In this embodiment, the website instance data input by the user can be obtained; alternatively, the website instances can be read from the database; alternatively, the website instance data sent by other devices can be received; alternatively, the website instance data collected by the data collection service can be obtained.

[0032] S102, evolving the initialized particle swarm through the competitive particle swarm algorithm and the website instance data, where the particles in the particle swarm represent the instance selection schemes of the website instance data.

[0033] Among them, the competitive particle swarm algorithm can refer to the foregoing explanation and will not be elaborated here.

[0034] Among them, there are multiple particles in the particle swarm. One particle represents an instance selection scheme of the website instance data, and the instance selection scheme corresponding to the particle can be represented by the position matrix of the particle and the velocity matrix of the particle.

[0035] In this embodiment, first, the random initialization of the particle swarm is performed, including the random initialization of the position matrix of the particles and the velocity matrix of the particles. Then, the particle swarm can be evolved multiple times through the competitive particle swarm algorithm and the website instance data to obtain the finally evolved particle swarm. In each evolution process of the particle swarm, the instance selection can be performed in the website instance data according to the instance selection schemes represented by the particles in the particle swarm, and the subset of instances corresponding to the particles in the particle swarm (that is, the subset of instances corresponding to the instance selection scheme represented by the particle) can be obtained. According to the quality of the scheme reflected by the subset of instances corresponding to the particle (that is, the quality of the instance selection scheme represented by the particle), the competition and evolution between the particles are performed to optimize the solution of the instance selection scheme.

[0036] S103, performing instance selection in the website instance data according to the evolved particle swarm to obtain the target subset of instances, and the target subset of instances is used for the training of the phishing website classification model.

[0037] In this embodiment, according to the instance selection scheme represented by the particles in the evolved particle swarm, instance selection can be performed on the website instance data to obtain a target instance subset. The number of particles is multiple, and the obtained target instance subsets are also multiple. These target instance subsets can be used for training the classification model of phishing websites, or a part of the target instance subsets can be selected from these target instance subsets for training the classification model of phishing websites to improve the training effect of the classification model.

[0038] In the embodiment of the present application, the instance selection scheme of website instance data is represented by the particles in the particle swarm, and the particle swarm is evolved by the competitive particle swarm algorithm to optimize and solve the instance selection scheme of website instance data, improving the accuracy of instance selection for website instance data, obtaining an instance subset with strong representativeness and less redundancy, so as to improve the training efficiency and training effect of the classification model through the instance subset with strong representativeness and less redundancy, and improve the detection accuracy of the classification model for phishing websites.

[0039] In some embodiments, the instance selection scheme of website instance data includes the selection schemes respectively corresponding to multiple website instances in the website instance data. Among them, the selection schemes respectively corresponding to the multiple website instances may include the selected probabilities respectively corresponding to the multiple website instances and the selection labels respectively corresponding to the multiple website instances. The selection labels include participating in the selection or not participating in the selection.

[0040] In this embodiment, the selection labels respectively corresponding to the multiple website instances can be represented by the position matrix of the particles in the particle swarm, and the selected probabilities respectively corresponding to the multiple website instances can be represented by the velocity matrix of the particles in the particle swarm. Thus, the instance selection scheme of website instance data is accurately represented by combining the position matrix and the velocity matrix of the particles in the particle swarm.

[0041] Optionally, the number of elements in the position matrix of the particle is equal to the number of website instances in the website instance data, and the number of elements in the velocity matrix of the particle is equal to the number of website instances in the website instance data.

[0042] Furthermore, the element values in the position matrix of the particle are binary values. The number of elements in the position matrix of the particle may be equal to the number of website instances in the website instance data, and the selection labels respectively corresponding to the multiple website instances in the website instance data are represented by multiple binary values in the position matrix of the particle. The number of elements in the velocity matrix of the particle may be equal to the number of website instances in the website instance data.

[0043] For example, the website instance data is represented as {I1, I2, I3, I4, I5, I6}, where I1, I2, I3, I4, I5, and I6 respectively represent different website instances; the position matrix of the first particle in the particle swarm is [1, 0, 1, 1, 0, 1], and this position matrix indicates that I1, I3, I4, and I6 are involved in the selection, while I2 and I5 are not; the position matrix of the second particle in the particle swarm is [0, 1, 1, 0, 1, 1], and this position matrix indicates that I2, I3, I5, and I6 are involved in the selection, while I1 and I4 are not.

[0044] The competitive particle swarm algorithm improves the search efficiency (i.e., the evolution efficiency) of particles during the evolution process through a competition mechanism. However, it may increase additional computational overhead. Especially in the instance selection optimization of large-scale instance datasets, it may face the problem of insufficient computing resources, resulting in limited optimization performance. To solve this problem, the embodiments of the present disclosure improve the evolution strategies of loser particles and winner particles during the particle evolution process, and specific details can be referred to the descriptions of the subsequent embodiments.

[0045] Figure 2 It is a schematic flowchart of an optimization process of the particle swarm in the website data processing method provided by the embodiments of the present disclosure. As Figure 2 shown, an optimization process of the particle swarm includes: S201, according to the instance selection scheme represented by the particles in the particle swarm, perform instance selection in the website instance data to obtain the subset of instances corresponding to the particles in the particle swarm.

[0046] In this embodiment, the particle swarm includes multiple particles, and one particle represents an instance selection scheme. For each particle in the particle swarm, instances can be selected in the website instance data according to the instance selection scheme represented by the particle, and the selected instances form the subset of instances corresponding to the particle. In this way, the subset of instances corresponding to each particle in the particle swarm is obtained respectively.

[0047] In a possible implementation manner, the position matrix of the particles in the particle swarm represents the selection tags corresponding to multiple website instances, and the velocity matrix of the particles in the particle swarm represents the selected probabilities corresponding to multiple website instances. Instance selection can be performed in the website instance data according to the position matrix of the particles in the particle swarm and the velocity matrix of the particles in the particle swarm to obtain the subset of instances corresponding to the particles in the particle swarm.

[0048] In this implementation method, for any particle in the particle swarm, instance selection is performed in the website instance data according to the selection label represented by the element value of the i-th element in the position matrix of the any particle and the selection probability represented by the element value of the i-th element in the velocity matrix of the any particle, to obtain the i-th instance in the instance subset corresponding to the any particle, where i is greater than or equal to 1. According to this process, instance subsets corresponding to each particle in the particle swarm can be obtained. Thus, the instance selection scheme is accurately represented by the velocity matrix of the particles in the particle swarm and the position matrix of the particles in the particle swarm, improving the accuracy of instance selection from website instances to obtain an instance subset.

[0049] S202. On the target dimension related to instance selection, evaluate the instance subsets corresponding to the particles in the particle swarm to obtain the target values corresponding to the particles in the particle swarm on the target dimension.

[0050] Among them, the target dimension related to instance selection can be one or more dimensions. The target dimension is the evaluation dimension of instance selection, and the instance subsets of instance selection are evaluated on the target dimension.

[0051] Among them, the target value corresponding to the particle on the target dimension reflects the quality of the instance subset corresponding to the particle on the target dimension, and further reflects the pros and cons of the instance selection scheme represented by the particle.

[0052] In this embodiment, the instance subsets corresponding to the particles in the particle swarm can be evaluated according to the subset evaluation method corresponding to the target dimension to obtain the target values corresponding to the particles in the particle swarm on the target dimension.

[0053] In one example, the subset evaluation method corresponding to the target dimension may include one or more of the following: the subset evaluation formula corresponding to the target dimension, the subset evaluation condition corresponding to the target dimension, and the subset evaluation model corresponding to the target dimension, so as to improve the accuracy and diversity of evaluating the instance subsets corresponding to the particles through these subset evaluation methods. The following describes these evaluation methods separately.

[0054] When the subset evaluation method corresponding to the target dimension includes the subset evaluation formula corresponding to the target dimension, the subset evaluation formula corresponding to the target dimension can be calculated based on the instance subsets corresponding to the particles in the particle swarm to obtain the target values corresponding to the particles in the particle swarm on the target dimension.

[0055] When the subset evaluation method corresponding to the target dimension includes the subset evaluation conditions corresponding to the target dimension, it is possible to judge the subset evaluation conditions satisfied by the instance subset corresponding to the particle in the particle swarm in the subset evaluation conditions corresponding to the target dimension, and determine the target value corresponding to the particle in the target dimension according to the subset evaluation conditions satisfied by the instance subset corresponding to the particle. For example, if the target dimension corresponds to multiple subset evaluation conditions, and the multiple subset evaluation conditions respectively correspond to different target values, the target value corresponding to the particle in the target dimension is the target value corresponding to the subset evaluation conditions satisfied by the instance subset corresponding to the particle.

[0056] When the subset evaluation method corresponding to the target dimension includes the subset evaluation model corresponding to the target dimension, the subset evaluation model corresponding to the target dimension can be a classification model for phishing websites. The classification model is trained by the instance subset corresponding to the particle in the particle swarm, and the target value corresponding to the particle in the target dimension is determined according to the training index of the classification model.

[0057] In a possible implementation manner, the target dimension includes a first dimension representing the phishing network detection error and / or a second dimension representing the subset complexity. S202 includes: using the instance subsets corresponding to the particles in the particle swarm as multiple independent training data sets to train the classification model respectively, obtaining the model training error of the instance subsets corresponding to the particles in the particle swarm, and determining the target value corresponding to the particles in the particle swarm in the first dimension according to the model training error of the instance subsets corresponding to the particles in the particle swarm; and / or determining the target value corresponding to the particles in the particle swarm in the second dimension according to the complexity-related features of the instance subsets corresponding to the particles in the particle swarm. Among them, the phishing network detection error can reflect the effect of the instance subset used for training the classification model of the phishing network, and the subset complexity can reflect the data volume of the instance subset. Evaluating the instance subset from these two target dimensions can accurately reflect the advantages and disadvantages of the instance selection scheme corresponding to the particle, and improve the accuracy of subsequent particle competition and evolution.

[0058] In this implementation manner, when the target dimension includes a first dimension representing the fishing network detection error: for any particle in the particle swarm, the real example subset corresponding to the particle is used as a training data set, and the classification model is trained based on the training data set to obtain the training error during the training of the classification model. This training error is the model training error of the real example subset corresponding to the particle, and it is determined that the target value corresponding to the particle in the first dimension is this model training error. When the target dimension includes a second dimension representing the subset complexity: for any particle in the particle swarm, based on the complexity-related features of the real example subset corresponding to the particle, the target value corresponding to the particle in the second dimension is determined. The complexity-related features of the real example subset may include one or more of the following: the number of website instances in the real example subset, the number of website instances with the website label of fishing website in the real example subset, and the number of website instances with the website label of normal website in the real example subset.

[0059] When the target dimension includes the above first dimension and / or the above second dimension, in the competitive particle swarm algorithm, the objective function for the particle swarm evolution can be: minimizing the target value corresponding to the particles in the particle swarm in the first dimension and / or minimizing the target value corresponding to the particles in the particle swarm in the second dimension, so as to extract a real example subset with good model training effect, less noise, low redundancy and representativeness from the website instance data.

[0060] For example, the formula of the objective function can be expressed as: where, represents the real example subset corresponding to the particle, represents the target value corresponding to the particle in the first dimension, represents the target value corresponding to the particle in the second dimension.

[0061] S203. According to the target values corresponding to the particles in the particle swarm in the target dimension, perform a clustering operation on the particles in the particle swarm to obtain multiple particle clusters.

[0062] In this embodiment, the target values corresponding to the particles in the particle swarm in the target dimension can be used as one of the reference data for clustering. The particles in the particle swarm can be clustered according to the set clustering rules or clustering algorithms to obtain multiple particle clusters.

[0063] In a possible implementation, S203 includes: determining the angles between the particles in the particle swarm and multiple reference vectors respectively according to the target values corresponding to the particles in the particle swarm in the target dimension; clustering the multiple particles into multiple initial clusters according to the angles between the particles in the particle swarm and the multiple reference vectors respectively, where the multiple initial clusters correspond to the multiple reference vectors one by one; adjusting the particles in the initial clusters with the actual number of particles greater than the cluster capacity and the particles in the initial clusters with the actual number of particles less than the cluster capacity according to the multiple initial clusters and the multiple reference vectors to obtain multiple particle clusters. Thus, by introducing reference vectors uniformly distributed in the target space and combining the angles between the particles in the particle swarm and the multiple reference vectors, the particles in the particle swarm are clustered, improving the accuracy of clustering the particles. Among them, the multiple reference vectors are uniformly distributed in the target space composed of the target dimension.

[0064] Among them, the target space composed of the target dimension may include the coordinate space composed of the target dimension. The coordinate space composed of the target dimension refers to the coordinate system space with the target dimension as the coordinate axis, and this coordinate space includes at least one quadrant space in the coordinate system with the target dimension as the coordinate axis.

[0065] Optionally, when the target dimension includes a first dimension and a second dimension, the target space composed of the target dimension includes at least one quadrant space in the two-dimensional coordinate system with the first dimension and the second dimension as the coordinate axes. If the number of target dimensions is three, the target space may include at least one quadrant space in the three-dimensional coordinate system with the three target dimensions as the coordinate axes, and so on.

[0066] Among them, the reference vector is a vector with the starting point at the origin and uniformly distributed in the target space.

[0067] In the process of determining the angles between the particles in the particle swarm and the multiple reference vectors respectively according to the target values corresponding to the particles in the particle swarm in the target dimension, the target values corresponding to the multiple reference vectors in the target dimension can be determined respectively. For any particle in the particle swarm, the angles between the any particle and the multiple reference vectors can be calculated according to the target value corresponding to the any particle in the target dimension and the target values corresponding to the multiple reference vectors in the target dimension respectively, so as to obtain the angles between the any particle and the multiple reference vectors respectively.

[0068] Optionally, the calculation formula for the angles between the particles in the particle swarm and the multiple reference vectors is expressed as: Among them, represents the j-th reference vector, represents the particle, represents the target value corresponding to the particle in the k-th target dimension, represents the target value corresponding to the reference vector in the k-th target dimension, and M represents the number of target dimensions. represents the angles between the particle and multiple reference vectors.

[0069] In the process of clustering multiple particles into multiple initial clusters according to the angles between the particles in the particle swarm and multiple reference vectors, multiple reference vectors respectively correspond to different clusters. For any particle in the particle swarm, the particle can be divided into the cluster corresponding to the reference vector with the smallest angle with the particle according to the angles between the particle and multiple reference vectors. After the particles in the particle swarm are divided, multiple initial clusters can be obtained, and the number of particles in different initial clusters may be different.

[0070] Optionally, the clustering formula for multiple initial clusters is expressed as: where represents the j-th initial cluster, represents the particle swarm, represents the set of reference vectors, represents the angle between particle s and the j-th reference vector.

[0071] In the process of adjusting the particles in the initial clusters with the actual number of particles greater than the cluster capacity and the particles in the initial clusters with the actual number of particles less than the cluster capacity according to multiple initial clusters and multiple reference vectors to obtain multiple particle clusters, the cluster capacity can be preset in advance. For the initial clusters with the actual number of particles greater than the cluster capacity, the particles that need to be adjusted out in the initial cluster can be determined according to the angles between the particles in the initial cluster and the reference vector corresponding to the initial cluster. According to the angle between the particle that needs to be adjusted out and the reference vector corresponding to the initial cluster with the actual number of particles less than the cluster capacity, the target cluster is determined in the initial cluster with the actual number of particles less than the cluster capacity, and the particle that needs to be adjusted out is divided into the target cluster. After the particles in multiple initial clusters are adjusted, multiple particle clusters are obtained.

[0072] Optionally, the cluster capacity is 2, that is, each of the finally obtained multiple particle clusters includes two particles, so as to determine the winning particles and losing particles in each particle cluster in the subsequent competition process.

[0073] In one example, in the adjustment process, the initial clusters with the actual number of particles greater than the cluster capacity only retain the two particles closest to the reference vector corresponding to the initial cluster, and move the remaining particles to other unfilled clusters.

[0074] S204. Compare the particles in multiple particle clusters respectively to determine the winning particles and losing particles in multiple particle clusters.

[0075] In this embodiment, among multiple particle clusters, the particles in the same particle cluster can be respectively compared in terms of performance. The particle that wins in the performance comparison is determined as the winner particle in the particle cluster, and the particle that loses in the performance comparison is determined as the loser particle in the particle cluster.

[0076] Optionally, the performance index of the particle adopts the Chebyshev value of the particle. The winner particle and the loser particle in the same particle cluster are determined by comparing the Chebyshev values of the particles in the same particle cluster.

[0077] In a possible implementation manner, S204 includes: for any one of the multiple particle clusters, determining the target value corresponding to the ideal particle of any one particle cluster in the target dimension according to the target values corresponding to the particles in any one particle cluster in the target dimension; determining the Chebyshev value of the particles in any one particle cluster according to the target values corresponding to the particles in any one particle cluster in the target dimension, the target value corresponding to the ideal particle of any one particle cluster in the target dimension, and multiple reference vectors; determining the winner particle and the loser particle in any one particle cluster by comparing the Chebyshev values of the particles in any one particle cluster. Thereby, the accuracy of representing the particle performance by the Chebyshev value of the particle is improved, and the accuracy of comparing and obtaining the winner particle and the loser particle is improved.

[0078] In the process of determining the target value corresponding to the ideal particle of any one particle cluster in the target dimension according to the target values corresponding to the particles in any one particle cluster in the target dimension for any one of the multiple particle clusters, for any one particle cluster, the minimum value can be determined among the target values corresponding to the particles in the any one particle cluster in the target dimension, and the target value corresponding to the ideal particle of the any one particle cluster in the target dimension is determined as this minimum value.

[0079] In the process of determining the Chebyshev value of the particles in any one particle cluster, for any one particle cluster, the target value difference between the particle and the ideal particle in the target dimension can be calculated according to the target value corresponding to the particle in the target dimension of the any one particle cluster and the target value corresponding to the ideal particle of the any one particle cluster in the target dimension, and this target value difference reflects the gap between the particle and the ideal particle in the target dimension; the target value of the reference vector corresponding to the particle cluster to which the particle belongs in the target dimension can be determined according to the determination method of the target value of the target dimension; the Chebyshev value of the particle is determined according to the target value difference between the particle and the ideal particle in the target dimension and the target value of the reference vector corresponding to the particle cluster to which the particle belongs in the target dimension.

[0080] Optionally, if there are multiple target dimensions, the Chebyshev value of the particle can be determined as the maximum value among the target value differences between the particle and the ideal particle in the target dimensions.

[0081] Optionally, the calculation formula of the Chebyshev value of a particle is expressed as: where represents an ideal particle, represents the target value corresponding to the ideal particle in the i-th target dimension, represents the target value corresponding to particle s in the i-th target dimension, represents the reference vector corresponding to the j-th particle cluster, represents the target value of the reference vector corresponding to the j-th particle cluster in the target dimension.

[0082] In the process of determining the winner particle and the loser particle in any particle cluster by comparing the Chebyshev values of the particles in any particle cluster, the larger the Chebyshev value of a particle, the worse the performance of the particle. The Chebyshev values of the particles in the same particle cluster will be compared, and the particle with the larger Chebyshev value in the same particle cluster will be determined as the loser particle, and the particle with the smaller Chebyshev value in the same particle cluster will be determined as the winner particle.

[0083] S205. According to the evolutionary strategy corresponding to the winner particle, perform an evolutionary operation on the winner particles in multiple particle clusters to obtain the evolutionary particles corresponding to the winner particles in multiple particle clusters.

[0084] In this embodiment, different evolutionary strategies can be adopted for the winner particle and the loser particle. By designing different evolutionary strategies for the winner particle and the loser particle, personalized evolution of the winner particle and the loser particle can be realized, and the evolutionary effect can be improved. The evolutionary operation can be performed on the winner particles in multiple particle clusters according to the evolutionary strategy corresponding to the winner particle, that is, update the position matrix and velocity matrix of the winner particle to obtain the updated particle, and the updated particle is the evolutionary particle corresponding to the winner particle.

[0085] Optionally, the evolutionary strategy corresponding to the winner particle is an evolutionary strategy based on the Gaussian model. In the evolutionary strategy based on the Gaussian model, the evolution of the winner particle is assisted by the Gaussian model, thereby improving the diversity of the evolution of the winner particle, and further improving the evolutionary effect of the particle swarm.

[0086] In a possible implementation, S205 includes: constructing a Gaussian model, where the mean matrix and covariance matrix in the Gaussian model are determined based on the position matrices of the winner particles in multiple particle clusters; performing random sampling according to the Gaussian model to obtain an updated particle position matrix; and guiding the evolution of the winner particles in multiple particle clusters based on the updated particle position matrix to obtain the evolved particles corresponding to the winner particles in multiple particle clusters. Thus, by using the Gaussian model to assist the evolution of the winner particles, the diversity of the evolution of the winner particles is improved, compensating for the defect that the competitive particle swarm optimization algorithm ignores the quality of the winner particles and leads to low optimization performance. Moreover, the computational complexity of the evolution strategy based on the Gaussian model is small, which is beneficial to saving computing resources and improving the evolution efficiency.

[0087] In the process of constructing the Gaussian model, calculate the mean matrix according to the position matrices of the winner particles in multiple particle clusters; calculate the covariance matrix according to the position matrices and the mean matrix of the winner particles in multiple particle clusters; and the Gaussian model can be constructed by combining the mean matrix and the covariance matrix.

[0088] Optionally, calculate the mean of the position matrices of the winner particles in multiple particle clusters to obtain the mean matrix, and the calculation formula is expressed as: where N represents the number of particles in the particle swarm, N / 2 represents the number of winner particles, represents the position matrix of the i-th winner particle, represents the mean matrix.

[0089] Optionally, the calculation formula of the covariance matrix is expressed as: where cov represents the covariance matrix.

[0090] Optionally, the Gaussian model is expressed as: where x represents the input data of the Gaussian model.

[0091] In the subsequent random sampling and the evolution process of the winner particles, random sampling can be performed based on the Gaussian model to obtain an updated particle position matrix. With the updated particle position matrix, the evolution of the winner particles is guided to obtain the evolved particles of the winner particles. In the guiding process, the velocity matrix of the evolved particles can be determined according to the updated particle position matrix, the position matrix of the winner particles, and the velocity matrix of the winner particles; the position matrix of the evolved particles can be determined according to the velocity matrix of the evolved particles and the velocity matrix of the winner particles. Thus, based on the updated particle position matrix, the guidance of the position and velocity evolution of the winner particles is realized, improving the diversity of the position evolution and velocity evolution of the winner particles.

[0092] Optionally, determine the position matrix difference between the updated particle position matrix and the position matrix of the winner particle, and obtain the velocity matrix of the evolved particle by weighting the position matrix difference and the velocity matrix of the winner particle. The formula for the evolution process of the winner particle can be expressed as: Wherein, represents the position matrix of the winner particle, represents the updated particle position matrix, represents the velocity matrix of the winner particle, and respectively represent the weight parameter corresponding to the velocity matrix of the winner particle and the weight parameter corresponding to the position matrix difference. These two weight parameters can be random values between 0 and 1. represents the velocity matrix of the evolved particle corresponding to the successful particle, represents the position matrix of the evolved particle corresponding to the successful particle.

[0093] S206. According to the evolution strategy corresponding to the loser particle, perform an evolution operation on the loser particles in multiple particle clusters to obtain the evolved particles corresponding to the loser particles in the multiple particle clusters.

[0094] In this embodiment, an evolution operation can be performed on the loser particles in multiple particle clusters according to the evolution strategy corresponding to the loser particle, that is, update the position matrix and velocity matrix of the loser particle to obtain an updated particle, and the updated particle is the evolved particle corresponding to the winner particle.

[0095] Optionally, the evolution strategy corresponding to the loser particle is an evolution strategy based on constraint conditions, so as to improve the convergence speed of the loser particle during the evolution process and improve the ability of the competitive particle swarm to solve the optimization problem of the instance selection scheme for large-scale website instance data.

[0096] In a possible implementation, the constraint conditions include the intra-domain learning condition and the inter-domain learning condition. S206 includes: when the current evolution meets the intra-domain learning condition, for the same particle cluster among multiple particle clusters, according to the position matrix of the winner particle in the same particle cluster, guiding the evolution of the loser particles in the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster; when the current evolution meets the inter-domain learning condition, for the same particle cluster among multiple particle clusters, according to the position matrix of the winner particle in the same particle cluster and the position matrix of the winner particle in the adjacent cluster of the same particle cluster, guiding the evolution of the loser particles in the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster. Through the evolution process based on the intra-domain learning condition and the inter-domain learning condition, the convergence speed of the loser particles is improved, and the diversity of the evolution of the loser particles is also ensured, solving the deficiency of poor performance of the competitive particle algorithm in optimizing the instance selection scheme for large-scale website instance data (especially the multi-objective optimization problem when there are multiple target dimensions).

[0097] Among them, the intra-domain learning condition includes that the conditional probability value in the current evolution is less than the probability threshold, and the inter-domain learning condition includes that the conditional probability value is greater than or equal to the probability threshold. The conditional probability value is randomly generated in the current evolution to improve the diversity of the evolution of the loser particles.

[0098] When the current evolution meets the intra-domain learning condition, according to the position matrix of the winner particle in the same particle cluster, guiding the evolution of the loser particles in the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster, thereby realizing the evolution of the loser particles through the intra-domain learning method and improving the convergence speed of the evolution of the loser particles.

[0099] The evolution process of this intra-domain learning may include: in the same particle cluster, the velocity matrix of the evolved particle corresponding to the loser particle can be determined according to the position matrix of the winner particle, the position matrix of the loser particle, and the velocity matrix of the loser particle; the position matrix of the evolved particle can be determined according to the velocity matrix of the evolved particle and the position matrix of the loser particle.

[0100] Optionally, in the same particle cluster, the position matrix difference between the position matrix of the winner particle and the position matrix of the loser particle is determined, and by weighting the position matrix difference and the velocity matrix of the loser particle, the velocity matrix of the evolved particle corresponding to the loser particle is obtained.

[0101] Optionally, the formula representation of the intra-domain evolution process of the loser particle is as follows: Among them, represents the position matrix of the loser particle, Represents the velocity matrix of the loser particles, and respectively represent the weight parameter corresponding to the velocity matrix of the loser particles and the weight parameter corresponding to the position matrix difference. These two weight parameters can be random values between 0 and 1. Represents the velocity matrix of the evolved particles corresponding to the loser particles, Represents the position matrix of the evolved particles corresponding to the loser particles.

[0102] When the current evolution meets the inter-domain learning condition, based on the position matrix of the winner particles in the same particle cluster and the position matrix of the winner particles in the adjacent cluster of the same particle cluster, guide the evolution of the loser particles in the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster, so as to realize the evolution of the loser particles through the inter-domain learning method.

[0103] The evolution process of this inter-domain learning may include: determining the velocity matrix of the evolved particles corresponding to the loser particles according to the position matrix of the winner particles in the same particle cluster, the position matrix of the winner particles in the adjacent cluster of the same particle cluster, the position matrix of the loser particles in the same particle cluster, and the velocity matrix of the loser particles; determining the position matrix of the evolved particles according to the velocity matrix corresponding to the loser particles and the position matrix of the loser particles.

[0104] Optionally, the formula representation of the inter-domain evolution process of the loser particles is: where, Represents the position matrix of the winner in the adjacent cluster, and respectively represent the weight parameters. These two weight parameters can be random values between 0 and 1.

[0105] S207. Perform an environmental selection operation on the particles in the particle swarm, the evolved particles corresponding to the winner particles in multiple particle clusters, and the evolved particles corresponding to the loser particles in multiple particle clusters to determine the particles in the particle swarm after the current evolution.

[0106] In this embodiment, the particles in the particle swarm refer to the particles before the current evolution, including the winner particles and the loser particles. During the environmental selection process, the instance selection of website instance data can be performed according to the instance selection scheme represented by the evolved particles corresponding to the winner particles, and the subset of instances corresponding to the evolved particles can be obtained. The target value corresponding to the evolved particles in the target dimension can be obtained by evaluating the subset of instances in the target dimension. The instance selection of website instance data can be performed according to the instance selection scheme represented by the evolved particles corresponding to the loser particles, and the subset of instances corresponding to the evolved particles can be obtained. The target value corresponding to the evolved particles in the target dimension can be obtained by evaluating the subset of instances in the target dimension. According to the target values corresponding to the particles in the particle swarm in the target dimension, the target values corresponding to the evolved particles of the winner particles in the target dimension, and the target values corresponding to the evolved particles of the loser particles in the target dimension, particle selection is performed among the particles in the particle swarm, the evolved particles corresponding to the winner particles, and the evolved particles corresponding to the loser particles to obtain the particles after the current evolution. The particles after the current evolution form the particle swarm after the current evolution.

[0107] In a possible implementation manner, S207 may include: combining the particles in the particle swarm, the evolved particles corresponding to the winner particles in multiple particle clusters, and the evolved particles corresponding to the loser particles in multiple particle clusters into a temporary group; determining the non-dominated relationship of the particles in the temporary group according to the target values of the particles in the temporary group in the target dimension; stratifying the particles in the temporary group according to the non-dominated relationship of the particles in the temporary group to determine the non-dominated layer where the particles in the temporary group are located; and selecting the particles after the current evolution in the temporary group according to the non-dominated layer where the particles in the temporary group are located and the target values of the particles in the temporary group in the target dimension. Thus, by stratifying the particles according to the non-dominated relationship of the particles, the particles with better performance in the target dimension are selected as the particles after the current evolution, improving the accuracy and efficiency of environmental selection.

[0108] In this implementation manner, first, the particles in the particle swarm, the evolutionary particles corresponding to the winner particles in the multiple particle clusters, and the evolutionary particles corresponding to the loser particles in the multiple particle clusters are merged to obtain a temporary swarm. Assuming the number of particles in the particle swarm is N, then the number of particles in the merged temporary swarm is 2N. The instance selection of the website instance data can be performed according to the instance selection scheme represented by the particles in the temporary swarm to obtain the subset of instances corresponding to the particles in the temporary swarm; in the target dimension, the subset of instances corresponding to the particles in the temporary swarm is evaluated to obtain the target values of the particles in the temporary swarm in the target dimension; according to the target values of the particles in the temporary swarm in the target dimension, the non-dominated relationship of the particles in the temporary swarm is determined. The non-dominated relationship of the particles includes whether the particle is dominated by other particles. In the case where the particle is dominated by other particles, the non-dominated relationship of the particle further includes the identification information of the other particles that dominate the particle. The other particles refer to the remaining particles in the temporary swarm except this particle.

[0109] Optionally, determining the non-dominated relationship of the particles in the temporary swarm according to the target values of the particles in the temporary swarm in the target dimension may include: if the target values of the first particle in all target dimensions are not inferior to the target values of the second particle in the corresponding target dimensions and the target values of the first particle in at least one target dimension are better than the target values of the second particle in the at least one target dimension, then it is determined that the first particle dominates the second particle. Wherein, the first particle and the second particle are any two particles in the temporary swarm.

[0110] If there is no particle in the temporary swarm that dominates the first particle, then it can be determined that the non-dominated relationship of the first particle is that the first particle is not dominated by other particles.

[0111] In the case where the target dimension includes the first dimension and the second dimension, if the first particle and the second particle meet the following first condition and second condition, then it is determined that the first particle dominates the second particle. The first condition is: the target value of the first particle in the first dimension is not greater than the target value of the second particle in the first dimension, and the target values of the first particle in the second dimension are not greater than the target values of the second particle in the second dimension. The second condition is: the target value of the first particle in the first dimension is less than the target value of the second particle in the first dimension, or the target value of the first particle in the second dimension is less than the target value of the second particle in the second dimension. The first dimension and the second dimension can refer to the foregoing embodiments and will not be elaborated herein.

[0112] In this implementation manner, after determining the non-dominated relationship of the particles in the temporary swarm, the particles whose non-dominated relationship is not dominated by other particles can be classified into the first non-dominated layer, and the particles whose non-dominated relationship is not dominated by the particles in the i-th non-dominated layer can be classified into the (i + 1)-th non-dominated layer, where i is greater than or equal to 1. Finally, all the particles in the temporary swarm are classified into the corresponding non-dominated layers. After that, the particles after the current evolution can be selected from the temporary swarm according to the hierarchical order of the non-dominated layers where the particles in the temporary swarm are located and the objective values of the particles in the temporary swarm in the target dimension.

[0113] Optionally, selecting the particles after the current evolution from the temporary swarm according to the hierarchical order of the non-dominated layers where the particles in the temporary swarm are located and the objective values of the particles in the temporary swarm in the target dimension includes: starting from the first non-dominated layer, selecting all the particles in the entire non-dominated layer as the particles after the current evolution until the selection stop condition is met. The selection stop condition is: if all the particles in the L-th non-dominated layer are selected as the particles after the current evolution, the number of the particles after the current evolution will be greater than the swarm size (i.e., the number of particles in the particle swarm), where L is greater than or equal to 1. According to the swarm size and the number of the particles after the current evolution, determine the number of the remaining particles to be selected; according to the objective values of the particles in the L-th non-dominated layer in the target dimension, select the number of the remaining particles to be selected from the L-th non-dominated layer. Finally, the selection of the particles after the current evolution is completed. Thus, by combining the hierarchical order and the objective value, the selection accuracy and efficiency of the particles after the current evolution are improved.

[0114] Among them, the formula for calculating the number of the remaining particles to be selected can be expressed as: In this formula, N represents the swarm size, K represents the number of the remaining particles to be selected, represents the i-th non-dominated layer. By subtracting the total number of particles in the first L - 1 non-dominated layers (i.e., the number of the particles after the current evolution that have been selected) from N, the number of the remaining particles to be selected is obtained.

[0115] Optionally, according to the objective values of the particles in the L-th non-dominated layer in the objective dimension, select the remaining number of particles to be selected in the L-th non-dominated layer, which may include: sorting the particles in the L-th non-dominated layer according to the objective values of the particles in the L-th non-dominated layer in the objective dimension to obtain the sorting order of the particles in the L-th non-dominated layer in the objective dimension; determining the objective crowding degree of the particles in the L-th non-dominated layer according to the sorting order of the particles in the L-th non-dominated layer in the objective dimension; based on the objective crowding degree of the particles in the L-th non-dominated layer, select the remaining number of particles to be selected from the L-th non-dominated layer in descending order of the objective crowding degree, and the remaining number of particles to be selected is the particles after the current evolution.

[0116] Among them, the objective crowding degree of a particle indicates the sparsity degree of the particle in the objective space. If the objective crowding degree of a particle is large, it means that the neighbors of the particle are few, that is, the sparsity degree of the particle in the objective space is large, and more "exploratory" solutions can be retained. Selecting particles in descending order of the objective crowding degree can select particles with large objective crowding degrees, not limited to particle evolution in a small area, and improves the exploration ability of particle evolution.

[0117] In this optional solution, if there are multiple objective dimensions, then the particles in the L-th non-dominated layer can be sorted in each objective dimension. Taking the objective dimensions including the first dimension and the second dimension as an example, sort the particles in the L-th non-dominated layer according to the objective values of the particles in the L-th non-dominated layer in the first dimension to obtain the sorting order of the particles in the L-th non-dominated layer in the first dimension; sort the particles in the L-th non-dominated layer according to the objective values of the particles in the L-th non-dominated layer in the second dimension to obtain the sorting order of the particles in the L-th non-dominated layer in the second dimension.

[0118] In this optional solution, in the process of determining the objective crowding degree of the particles in the L-th non-dominated layer according to the sorting order of the particles in the L-th non-dominated layer in the objective dimension, first, determine the edge particles and the middle particles according to the sorting order of the particles in the L-th non-dominated layer in the objective dimension. The edge particles refer to the particles located at both ends in the sorting order, and the middle particles refer to the particles located in the middle position in the sorting order; determine that the objective crowding degree of the edge particles is a set value, and this set value can be an infinite value; determine the objective crowding degree of the middle particles according to the objective values corresponding to the edge particles in the objective dimension and the objective values of the neighbor particles of the middle particles in the objective dimension. The neighbor particles of the middle particles refer to the particles adjacent to the middle particles in the sorting order in the objective dimension.

[0119] In the case where the target dimension includes a first dimension and a second dimension, determining the boundary particles and the intermediate particles according to the sorting order of the particles in the L-th non-dominated layer in the target dimension may include: determining the particles sorted at both ends as boundary particles according to the sorting order of the particles in the L-th non-dominated layer in the first dimension, and determining the particles sorted at both ends as boundary particles according to the sorting order of the particles in the L-th non-dominated layer in the second dimension; determining the remaining particles in the L-th non-dominated layer except the boundary particles as intermediate particles. In the case where the target dimension includes more dimensions, this process can be referred to.

[0120] Determining the target crowding degree of the intermediate particles according to the target values corresponding to the boundary particles in the target dimension and the target values of the neighbor particles of the intermediate particles in the target dimension may include: determining the difference between the target values corresponding to the boundary particles in the target dimension, determining the difference between the target values of the neighbor particles of the intermediate particles in the target dimension, and determining the target crowding degree of the intermediate particles according to the difference between the target values corresponding to the boundary particles in the target dimension and the difference between the target values of the neighbor particles in the target dimension.

[0121] Optionally, the calculation formula for the target crowding degree of the intermediate particles is expressed as: Wherein, represents the target crowding degree of the i-th intermediate particle, and represent the target values of the neighbor particles in the m-th target dimension. The neighbor particles include the particles adjacent to the i-th intermediate particle in the sorting order of the particles in the L-th non-dominated layer in the m-th target dimension, and respectively represent the maximum and minimum values of the target values of the particles in the L-th non-dominated layer in the m-th target dimension, which can be obtained from the target values of the boundary particles in the m-th target dimension.

[0122] In the embodiments of the present application, a primary evolution process of the particle swarm is provided. In this evolution process, the evolution effect and efficiency of the particle swarm are improved from multiple aspects (such as the evolution strategy for the winner particles, the evolution strategy for the loser particles, and the environmental selection operation for the particles), thereby improving the effect and efficiency of optimizing the instance selection scheme for website instance data and improving the quality of the target instance subset.

[0123] As an example, Figure 3 is an example diagram of the evolution process of the particle swarm. As shown in Figure 3As shown in the figure, first, the particle swarm P is obtained through initialization. Next, P is evolved: spatial partitioning (clustering and comparing the particles in the clusters obtained by clustering) is performed on the particle swarm P to obtain winner particles and loser particles; for the winner particles, an evolutionary strategy based on the Gaussian model is used for evolution to obtain the evolved particles corresponding to the winner particles; for the loser particles, an evolutionary strategy based on the constraint conditions is used for evolution to obtain the evolved particles corresponding to the loser particles; then, environmental selection is performed on all particles to select the particles after this evolution, and the particle swarm after this evolution is obtained. It is judged whether the evolution end condition is satisfied (such as whether the number of evolutions is greater than or equal to the number threshold). If it is satisfied, the evolution ends; otherwise, the next evolution is performed. Finally, the particle swarm P after the evolution ends is output.

[0124] Corresponding to the website data processing method in the above embodiment Figure 4 is a structural block diagram of the website data processing device provided in the embodiment of the present disclosure. For the sake of description, only the parts related to the embodiment of the present disclosure are shown. Referring to Figure 4 , the website data processing device 400 includes: an acquisition unit 401, configured to acquire website instance data, where the website instance data includes a plurality of website instances; an evolution unit 402, configured to evolve the initialized particle swarm through a competitive particle swarm algorithm and the website instance data, and the particles in the particle swarm represent the instance selection schemes of the website instance data; a selection unit 403, configured to perform instance selection in the website instance data according to the evolved particle swarm to obtain a target instance subset, and the target instance subset is used for training a classification model of phishing websites.

[0125] In some embodiments, during one evolution process of the particle swarm, the evolution unit 402 is specifically configured to: select instances from the website instance data according to the instance selection schemes represented by the particles in the particle swarm, so as to obtain a subset of instances corresponding to the particles in the particle swarm; evaluate the subset of instances corresponding to the particles in the particle swarm in the target dimension related to instance selection, so as to obtain the target values corresponding to the particles in the particle swarm in the target dimension; perform a clustering operation on the particles in the particle swarm according to the target values corresponding to the particles in the particle swarm in the target dimension, so as to obtain a plurality of particle clusters; compare the particles in the plurality of particle clusters respectively to determine the winner particles and the loser particles in the plurality of particle clusters; perform an evolution operation on the winner particles in the plurality of particle clusters according to the evolution strategies corresponding to the winner particles, so as to obtain the evolved particles corresponding to the winner particles in the plurality of particle clusters; perform an evolution operation on the loser particles in the plurality of particle clusters according to the evolution strategies corresponding to the loser particles, so as to obtain the evolved particles corresponding to the loser particles in the plurality of particle clusters; perform an environmental selection operation on the particles in the particle swarm, the evolved particles corresponding to the winner particles in the plurality of particle clusters, and the evolved particles corresponding to the loser particles in the plurality of particle clusters, so as to determine the particles in the particle swarm after the current evolution.

[0126] In some embodiments, the target dimension includes a first dimension representing the fishing network detection error and / or a second dimension representing the subset complexity. The evolution unit 402 is specifically configured to: use the subsets of instances corresponding to the particles in the particle swarm as a plurality of independent training data sets to train the classification model respectively, so as to obtain the model training errors of the subsets of instances corresponding to the particles in the particle swarm; determine the target values corresponding to the particles in the particle swarm in the first dimension according to the model training errors of the subsets of instances corresponding to the particles in the particle swarm; and / or determine the target values corresponding to the particles in the particle swarm in the second dimension according to the complexity-related features of the subsets of instances corresponding to the particles in the particle swarm.

[0127] In some embodiments, the evolution unit 402 is specifically configured to: determine the included angles between the particles in the particle swarm and a plurality of reference vectors according to the target values corresponding to the particles in the particle swarm in the target dimension, where the plurality of reference vectors are evenly distributed in the target space formed by the target dimension; cluster the plurality of particles into a plurality of initial clusters according to the included angles between the particles in the particle swarm and the plurality of reference vectors, and the plurality of initial clusters correspond to the plurality of reference vectors one by one; adjust the particles in the initial clusters with the actual number of particles greater than the cluster capacity and the particles in the initial clusters with the actual number of particles less than the cluster capacity according to the plurality of initial clusters and the plurality of reference vectors, so as to obtain a plurality of particle clusters.

[0128] In some embodiments, the evolution unit 402 is specifically configured to: for any one of the multiple particle clusters, determine the target value corresponding to the ideal particle of any one of the particle clusters in the target dimension according to the target value corresponding to the particles in any one of the particle clusters in the target dimension; determine the Chebyshev value of the particles in any one of the particle clusters according to the target value corresponding to the particles in any one of the particle clusters in the target dimension, the target value corresponding to the ideal particle of any one of the particle clusters in the target dimension, and multiple reference vectors; determine the winner particle and the loser particle in any one of the particle clusters by comparing the Chebyshev values of the particles in any one of the particle clusters.

[0129] In some embodiments, the evolution strategy corresponding to the winner particle is an evolution strategy based on a Gaussian model. The evolution unit 402 is specifically configured to: construct a Gaussian model, where the mean matrix in the Gaussian model and the covariance matrix in the Gaussian model are determined according to the position matrix of the winner particles in the multiple particle clusters; perform random sampling according to the Gaussian model to obtain an updated particle position matrix; and guide the evolution of the winner particles in the multiple particle clusters according to the updated particle position matrix to obtain the evolved particles corresponding to the winner particles in the multiple particle clusters.

[0130] In some embodiments, the evolution strategy corresponding to the loser particle is an evolution strategy based on constraint conditions. The constraint conditions include in-domain learning conditions and inter-domain learning conditions. The evolution unit 402 is specifically configured to: in the case where the current evolution satisfies the in-domain learning conditions, for the same particle cluster among the multiple particle clusters, guide the evolution of the loser particles in the same particle cluster according to the position matrix of the winner particles in the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster; in the case where the current evolution satisfies the inter-domain learning conditions, for the same particle cluster among the multiple particle clusters, guide the evolution of the loser particles in the same particle cluster according to the position matrix of the winner particles in the same particle cluster and the position matrix of the winner particles in the adjacent cluster of the same particle cluster to obtain the evolved particles corresponding to the loser particles in the same particle cluster; where the in-domain learning conditions include that the conditional probability value in the current evolution is less than the probability threshold, and the inter-domain learning conditions include that the conditional probability value is greater than or equal to the probability threshold, and the conditional probability value is randomly generated in the current evolution.

[0131] In some embodiments, the evolution unit 402 is specifically configured to: combine the particles in the particle swarm, the evolutionary particles corresponding to the winning particles in the multiple particle clusters, and the evolutionary particles corresponding to the losing particles in the multiple particle clusters into a temporary swarm; determine the non-dominated relationship of the particles in the temporary swarm according to the target values of the particles in the temporary swarm in the target dimension; stratify the particles in the temporary swarm according to the non-dominated relationship of the particles in the temporary swarm to determine the non-dominated layer where the particles in the temporary swarm are located; and select the particles after the current evolution in the temporary swarm according to the non-dominated layer where the particles in the temporary swarm are located and the target values of the particles in the temporary swarm in the target dimension.

[0132] The website data processing device provided in this embodiment can be used to execute the technical solutions of the embodiments of the above website data processing method. The implementation principles and technical effects are similar and will not be described in detail here.

[0133] Reference Figure 5 , which shows a schematic structural diagram of an electronic device 500 suitable for implementing the embodiments of the present disclosure. The electronic device 500 can be a terminal device or a server. Among them, the terminal device can include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, personal digital assistants (Personal Digital Assistant, abbreviated as PDA), tablet computers (Portable Android Device, abbreviated as PAD), portable multimedia players (Portable Media Player, abbreviated as PMP), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs and desktop computers. Figure 5 The shown electronic device is only an example and should not bring any limitation to the functions and usage scope of the embodiments of the present disclosure.

[0134] As Figure 5 shown, the electronic device 500 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to the program stored in the read-only memory (Read Only Memory, abbreviated as ROM) 502 or the program loaded from the storage device 508 into the random access memory (Random Access Memory, abbreviated as RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. The input / output (I / O) interface 505 is also connected to the bus 504.

[0135] Typically, the following devices can be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 can allow the electronic device 500 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 5 the electronic device 500 with various devices is shown, it should be understood that it is not required to implement or have all the shown devices. More or fewer devices can be alternatively implemented or had.

[0136] Specifically, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product that includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication device 509, or installed from the storage device 508, or installed from the ROM 502. When the computer program is executed by the processing device 501, the above functions defined in the method of the embodiment of the present disclosure are executed.

[0137] It should be noted that the computer-readable medium described above can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. In the present disclosure, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable signal medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0138] The above computer-readable medium can be included in the above electronic device; or it can exist separately and not be assembled into the electronic device.

[0139] The above computer-readable medium carries one or more programs, and when the one or more programs are executed by the electronic device, the electronic device is caused to execute the method shown in the above embodiments.

[0140] Computer program code for performing the operations of this disclosure may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may execute entirely on the user's computer, partially on the user's computer, execute as a stand-alone software package, execute partially on the user's computer and partially on a remote computer, or execute entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0141] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions noted in the blocks may occur in a different order than noted in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0142] The units involved in the embodiments described in this disclosure may be implemented in software or in hardware. Among them, the name of the unit does not constitute a limitation on the unit itself in some cases. For example, the acquisition unit may also be described as "the unit for acquiring the page image and page description text of the web page to be detected".

[0143] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), systems on a chip (SOC), complex programmable logic devices (CPLD), and so on.

[0144] In the context of the present disclosure, a machine-readable medium may be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0145] The above description is only a preferred embodiment of the present disclosure and an illustration of the applied technical principles. Those skilled in the art should understand that the scope of the disclosure involved in the present disclosure is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosure concept. For example, a technical solution formed by mutually replacing the above features with technical features having similar functions (but not limited to) disclosed in the present disclosure.

[0146] In addition, although the operations are depicted in a particular order, this should not be construed as requiring that the operations be performed in the particular order shown or in sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the present disclosure. Certain features described in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, the various features described in the context of a single embodiment may also be implemented separately or in any suitable sub-combination in multiple embodiments.

[0147] Although the subject matter has been described in language specific to structural features and / or methodological logical acts, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are merely example forms of implementing the claims.

Claims

1. A website data processing method, characterized in that: include: Acquire website instance data, wherein the website instance data includes multiple website instances; Evolving an initialized particle swarm through a competitive particle swarm algorithm and the website instance data, wherein particles in the particle swarm represent instance selection schemes of the website instance data; According to the evolved particle swarm, instance selection is performed in the website instance data to obtain a target instance subset, and the target instance subset is used for classification model training of phishing websites.

2. The website data processing method according to claim 1, characterized in that: The particle swarm's one-time evolution process includes: According to the instance selection scheme represented by the particles in the particle swarm, instance selection is performed in the website instance data to obtain an instance subset corresponding to the particles in the particle swarm; On a target dimension related to instance selection, evaluating a subset of instances corresponding to particles in the particle swarm to obtain target values ​​corresponding to the particles in the particle swarm on the target dimension; According to the target values ​​corresponding to the particles in the particle swarm in the target dimension, clustering operations are performed on the particles in the particle swarm to obtain a plurality of particle clusters; Comparing the particles in the plurality of particle clusters respectively, and determining winner particles in the plurality of particle clusters and loser particles in the plurality of particle clusters; According to the evolution strategy corresponding to the winner particle, the winner particles in the plurality of particle clusters are subjected to evolution operation to obtain the evolved particles corresponding to the winner particles in the plurality of particle clusters; According to the evolution strategy corresponding to the loser particles, the loser particles in the plurality of particle clusters are subjected to evolution operations to obtain evolved particles corresponding to the loser particles in the plurality of particle clusters; An environmental selection operation is performed on particles in the particle swarm, evolved particles corresponding to winner particles in the plurality of particle clusters, and evolved particles corresponding to loser particles in the plurality of particle clusters to determine particles in the particle swarm that have undergone a current evolution.

3. The website data processing method according to claim 2, characterized in that: The target dimension includes a first dimension representing a phishing network detection error and / or a second dimension representing a subset complexity. The instance subset corresponding to the particles in the particle swarm is evaluated on the target dimension related to the instance selection to obtain a target value corresponding to the particles in the particle swarm on the target dimension, including: Using instance subsets corresponding to the particles in the particle swarm as multiple independent training data sets, respectively training the classification model to obtain model training errors of the instance subsets corresponding to the particles in the particle swarm; determining target values ​​corresponding to the particles in the particle swarm in the first dimension according to the model training errors of the instance subsets corresponding to the particles in the particle swarm; And / or, determining the target value corresponding to the particles in the particle swarm in the second dimension according to the complexity-related characteristics of the instance subset corresponding to the particles in the particle swarm.

4. The website data processing method according to claim 2, characterized in that: The step of performing a clustering operation on the particles in the particle swarm according to the target values ​​corresponding to the particles in the particle swarm on the target dimension to obtain a plurality of particle clusters includes: Determine angles between particles in the particle swarm and a plurality of reference vectors according to target values ​​corresponding to particles in the particle swarm in the target dimension, wherein the plurality of reference vectors are evenly distributed in a target space formed by the target dimension; Clustering the particles into a plurality of initial clusters according to the angles between the particles in the particle group and the plurality of reference vectors, wherein the plurality of initial clusters correspond to the plurality of reference vectors one by one; According to the multiple initial clusters and the multiple reference vectors, particles in the initial clusters whose actual particle number is greater than the cluster capacity and particles in the initial clusters whose actual particle number is less than the cluster capacity are adjusted to obtain the multiple particle clusters.

5. The website data processing method according to claim 2, characterized in that: The step of comparing the particles in the plurality of particle clusters respectively to determine winner particles in the plurality of particle clusters and loser particles in the plurality of particle clusters comprises: For any particle cluster among the multiple particle clusters, determining a target value corresponding to an ideal particle corresponding to any particle cluster in the target dimension according to the target value corresponding to the particles in the any particle cluster in the target dimension; Determine the Chebyshev value of the particles in any particle cluster according to the target value corresponding to the particles in any particle cluster in the target dimension, the target value corresponding to the ideal particle corresponding to any particle cluster in the target dimension, and the multiple reference vectors; By comparing the Chebyshev values ​​of the particles in any particle cluster, a winner particle in any particle cluster and a loser particle in any particle cluster are determined.

6. The website data processing method according to any one of claims 2 to 5, characterized in that: The evolutionary strategy corresponding to the winner particle is an evolutionary strategy based on a Gaussian model. According to the evolutionary strategy corresponding to the winner particle, the winner particles in the multiple particle clusters are subjected to evolutionary operations to obtain evolved particles corresponding to the winner particles in the multiple particle clusters, including: Constructing the Gaussian model, wherein the mean matrix and the covariance matrix in the Gaussian model are determined according to the position matrix of the winner particles in the plurality of particle clusters; Perform random sampling according to the Gaussian model to obtain an updated particle position matrix; According to the updated particle position matrix, the evolution of the winner particles in the plurality of particle clusters is guided to obtain the evolved particles corresponding to the winner particles in the plurality of particle clusters.

7. The website data processing method according to any one of claims 2 to 5, characterized in that: The evolutionary strategy corresponding to the loser particles is an evolutionary strategy based on constraint conditions, wherein the constraint conditions include intra-domain learning conditions and inter-domain learning conditions. The evolutionary strategy corresponding to the loser particles is used to perform evolutionary operations on the loser particles in the multiple particle clusters to obtain evolved particles corresponding to the loser particles in the multiple particle clusters, including: When the current evolution satisfies the in-domain learning condition, for a same particle cluster of the multiple particle clusters, according to the position matrix of the winner particles in the same particle cluster, the evolution of the loser particles in the same particle cluster is guided to obtain the evolved particles corresponding to the loser particles in the same particle cluster; When the current evolution satisfies the inter-domain learning condition, for a same particle cluster among the multiple particle clusters, according to the position matrix of the winner particles in the same particle cluster and the position matrix of the winner particles in the adjacent clusters of the same particle cluster, the evolution of the loser particles in the same particle cluster is guided to obtain the evolved particles corresponding to the loser particles in the same particle cluster; Among them, the intra-domain learning condition includes that the conditional probability value in the current evolution is less than the probability threshold, and the inter-domain learning condition includes that the conditional probability value is greater than or equal to the probability threshold, and the conditional probability value is randomly generated in the current evolution.

8. The website data processing method according to any one of claims 2 to 5, characterized in that: The step of performing an environmental selection operation on the particles in the particle swarm, the evolved particles corresponding to the winner particles in the plurality of particle clusters, and the evolved particles corresponding to the loser particles in the plurality of particle clusters to determine the particles in the particle swarm that have undergone the current evolution includes: Merging particles in the particle group, evolved particles corresponding to winner particles in the plurality of particle clusters, and evolved particles corresponding to loser particles in the plurality of particle clusters into a temporary group; Determining, in the temporary group, a non-dominant relationship of particles in the temporary group according to target values ​​of particles in the temporary group under the target dimension; According to the non-dominated relationship of the particles in the temporary group, the particles in the temporary group are layered to determine the non-dominated layer where the particles in the temporary group are located; The particles that have undergone the current evolution are selected from the temporary group according to the non-dominated layer where the particles in the temporary group are located and the target values ​​of the particles in the temporary group under the target dimension.

9. A website data processing device, characterized in that: include: An acquisition unit, used to acquire website instance data, wherein the website instance data includes a plurality of website instances; An evolution unit, used for evolving an initialized particle swarm through a competitive particle swarm algorithm and the website instance data, wherein particles in the particle swarm represent instance selection schemes of the website instance data; The selection unit is used to select instances in the website instance data according to the evolved particle swarm to obtain a target instance subset, and the target instance subset is used for classification model training of phishing websites.

10. An electronic device, characterized in that: include: at least one processor and memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the website data processing method according to any one of claims 1 to 8.

Images