CN113824707A - Website performance dial testing measurement method and device based on knowledge graph - Google Patents
Website performance dial testing measurement method and device based on knowledge graph Download PDFInfo
- Publication number
- CN113824707A CN113824707A CN202111067717.4A CN202111067717A CN113824707A CN 113824707 A CN113824707 A CN 113824707A CN 202111067717 A CN202111067717 A CN 202111067717A CN 113824707 A CN113824707 A CN 113824707A
- Authority
- CN
- China
- Prior art keywords
- knowledge
- website performance
- graph
- layer
- extraction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012360 testing method Methods 0.000 title claims abstract description 34
- 238000000691 measurement method Methods 0.000 title claims abstract description 8
- 238000005259 measurement Methods 0.000 claims abstract description 28
- 238000000605 extraction Methods 0.000 claims description 60
- 239000013598 vector Substances 0.000 claims description 35
- 238000004364 calculation method Methods 0.000 claims description 25
- 230000009466 transformation Effects 0.000 claims description 23
- 238000000034 method Methods 0.000 claims description 19
- 238000013527 convolutional neural network Methods 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 16
- 238000013528 artificial neural network Methods 0.000 claims description 13
- 238000011176 pooling Methods 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 10
- 238000010276 construction Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000005457 optimization Methods 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1408—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic by monitoring network traffic
- H04L63/1416—Event detection, e.g. attack signature detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/142—Network analysis or design using statistical or mathematical methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1441—Countermeasures against malicious traffic
- H04L63/1458—Denial of Service
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1441—Countermeasures against malicious traffic
- H04L63/1466—Active attacks involving interception, injection, modification, spoofing of data unit addresses, e.g. hijacking, packet injection or TCP sequence number attacks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Computer Hardware Design (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Databases & Information Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Pure & Applied Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Algebra (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Environmental & Geological Engineering (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a website performance dial testing measurement method and a device based on a knowledge graph, which comprises the following steps: step 1, constructing a knowledge graph based on network dial testing; and 2, intelligently reasoning the website performance based on the graph convolution network according to the knowledge graph to realize the website performance dial testing measurement. By adopting the technical scheme of the invention, each performance index of the current website is inferred through the knowledge graph, and an evaluation report with strong interpretability is obtained.
Description
Technical Field
The invention belongs to the technical field of internet, and particularly relates to a website performance dial testing measurement method and device based on a knowledge graph.
Background
With the development of information technology and the rise of related demands, the number of websites on the internet is rapidly increasing. However, for various reasons such as operation cost and management awareness, many websites do not have security protection work and cannot guarantee the performance of the websites. If the serious performance problem of the website is not solved, a series of problems that the user cannot access, data is lost and the like can be caused, which is worthy of attention.
At present, there are many ways that can affect the performance of a website, such as Denial of Service (DoS), DNS hijacking, DNS pollution, etc. The traditional method is usually based on a fixed set of programs and modules for measuring and detecting the performance of the website, and manual intervention is needed. The traditional mode is extremely dependent on experience, has strong subjectivity and poor flexibility and robustness, and is easily influenced by a single factor or a small number of factors. Therefore, how to measure and from which aspects to measure the performance of the website is an urgent problem to be solved at present.
Disclosure of Invention
The invention aims to provide a website performance dial testing measurement method and device based on a knowledge graph.
In order to achieve the purpose, the invention adopts the following technical scheme:
a website performance dial testing measurement method based on a knowledge graph comprises the following steps:
And 2, intelligently reasoning the website performance based on the graph convolution network according to the knowledge graph to realize the website performance dial testing measurement.
Preferably, step 1 specifically comprises:
step 11, obtaining structured data, semi-structured data and unstructured data about website performance by adopting a network dial-up test mode;
step 12, extracting knowledge from the website performance semi-structured data and the unstructured data;
and step 13, carrying out entity alignment, entity disambiguation and attribute alignment on the result of the knowledge extraction so as to correct the knowledge in the website performance knowledge base.
Preferably, the knowledge extraction includes entity extraction, relationship extraction and attribute extraction.
Preferably, in step 12, a text-based convolutional neural network is adopted to perform relationship extraction, where the text-based convolutional neural network is composed of a convolutional layer, a pooling layer, and a full-link layer, and specifically includes:
step 121, preprocessing the semi-structured data and the unstructured data of the website performance, and converting each word in the data into a k-dimensional word vector; for a website performance semi-structured data or unstructured data containing n words, adopting an n x k matrix for representation, and taking the n x k matrix as the input of a text convolution neural network;
step 122, extracting high-dimensional semantic features from the input n × k website performance data matrix through the convolution layer, wherein the formula of the convolution calculation is as follows:
where N is the layer number of each layer in the neural network, L(N)Is the feature vector input by Nth layer, L(N+1)Is the feature vector, W, output by the Nth layer(N)Is the weight vector of the Nth layer neural network, B(N)Is the bias vector for the nth layer;
step 123, performing maximum pooling reduction on the dimensionality of the feature map on the high-dimensional semantic features, and fully connecting the pooled features with the m classified neurons in the last layer to obtain an m-dimensional vector; calculating the m-dimensional vector by using a Softmax function to obtain probability values corresponding to the m categories respectively, and taking the category with the maximum probability value, namely the result of extracting the website performance relationship, wherein the calculation formula of Softmax is as follows:
wherein z isiConverting the output value of the multi-classification into the range of [0,1 ] by a Softmax function for the output value of the ith vector and m for the number of output vectors, namely the number of classified classes]To be distributed in the space between them.
Preferably, step 2 specifically comprises:
step 21, dividing website performance knowledge into sub-items as input of a graph volume network;
step 22, carrying out graph convolution on each node of the graph by adopting transmitting, receiving and transformation, and then completing a layer of computational transformation through the nonlinear transformation of an activation function ReLU;
step 23, repeating step 22, inputting the output of the ReLU layer into the graph convolution layer, repeatedly using graph convolution calculation and nonlinear transformation, and continuously extracting the characteristics of the website performance;
and 24, after the features are extracted for multiple times, finishing the final comprehensive knowledge reasoning of the website performance and giving a measurement result of the website performance after the knowledge reasoning.
The invention also includes a website performance dial testing measurement device based on the knowledge graph, which comprises:
a construction module for constructing a knowledge graph based on network dial testing
And the reasoning module is used for intelligently reasoning the website performance based on the graph convolution network according to the knowledge graph to realize the website performance dial testing measurement.
Preferably, the building block comprises:
the processing unit is used for obtaining structured, semi-structured and unstructured data related to the website performance by adopting a network dial-up test mode;
the extraction unit is used for carrying out knowledge extraction on the website performance semi-structured data and the unstructured data;
and the fusion unit is used for carrying out entity alignment, entity disambiguation and attribute alignment on the result of the knowledge extraction so as to correct the knowledge in the website performance knowledge base.
Preferably, the knowledge extraction includes entity extraction, relationship extraction and attribute extraction.
Preferably, the extracting unit performs the relationship extraction by using a text-based convolutional neural network, where the text-based convolutional neural network is composed of a convolutional layer, a pooling layer, and a fully-connected layer, and specifically includes:
the preprocessing component is used for preprocessing the semi-structured data and the unstructured data of the website performance and converting each word in the data into a k-dimensional word vector;
the extraction component is used for extracting high-dimensional semantic features from the input n x k website performance data matrix through the convolution layer, wherein n x k represents website performance semi-structured data or unstructured data containing n words;
and the extraction component is used for fully connecting the pooled features with the m classified neurons of the last layer through the dimensionality of the maximum pooling reduction feature graph of the high-dimensional semantic features to obtain an m-dimensional vector, and calculating the m-dimensional vector by using a Softmax function to obtain probability values respectively corresponding to the m categories, namely a website performance relation extraction result.
Preferably, the metric module comprises:
the dividing unit is used for dividing the website performance knowledge into sub-items as the input of the graph convolution network;
the first calculation unit is used for carrying out graph convolution on each node of the graph by adopting transmission, reception and transformation, and then completing one layer of calculation transformation through the nonlinear transformation of an activation function ReLU;
the second calculation unit is used for inputting the output of the ReLU layer into the graph convolution layer, repeatedly using graph convolution calculation and nonlinear transformation and continuously extracting the characteristics of the website performance;
and the measurement unit is used for finishing the final website performance comprehensive knowledge inference after the characteristics are extracted for multiple times and giving a measurement result of the website performance after the knowledge inference.
The method for measuring the website performance by dialing and deducing the performance indexes of the current website through the knowledge graph to obtain the evaluation report with strong interpretability.
Drawings
FIG. 1 is a basic relationship diagram of a web site performance metric knowledge graph;
FIG. 2 is an exemplary diagram of website ping information mapped as a triplet of knowledge graph;
FIG. 3 is a flow chart of website performance knowledge graph construction;
FIG. 4 is a schematic diagram of relationship extraction based on a text convolutional neural network;
FIG. 5 is a diagram illustrating a website performance knowledge inference process;
FIG. 6 is a schematic diagram of a partial subgraph construction;
FIG. 7 is a schematic diagram of graph convolution calculation;
FIG. 8 is a schematic diagram of a Laplace matrix;
FIG. 9 is a relearning-based intelligent optimization;
FIG. 10 is a flow chart of a method for knowledge-graph based website performance dial-up measurement;
fig. 11 is a schematic structural diagram of a website performance dial-up measurement device based on a knowledge graph.
Detailed Description
The present invention will be described in further detail with reference to the following detailed description and accompanying drawings.
As shown in fig. 10, the present invention provides a method for measuring website performance dialing test based on knowledge graph, which comprises:
step S1, constructing knowledge graph based on network dial testing
The invention completes the measurement of the website performance through the knowledge graph technology and provides a reliable evaluation report. The website performance knowledge graph is constructed by the following steps: the method comprises the steps of performing all-round measurement on a website based on a network dial-up measurement technology to obtain measured data, and gradually constructing a website performance measurement knowledge graph through processes of information extraction and the like.
Taking the example of hopping on a DNS server, the construction of a knowledge graph starts with the access of a domain name and maps it to a basic knowledge graph triplet, as shown in fig. 1. Based on the website performance relationship graph shown in fig. 1, the nodes in the knowledge graph can be gradually expanded by continuously inputting data about the website performance.
The mapping of the knowledge graph is realized by extracting the network dial-up test items as entities in the website performance knowledge graph and using the measurement indexes, test places, test results and other related information of the network dial-up test as attributes of the entities, as shown in fig. 2.
As shown in fig. 3, constructing a web site performance knowledge graph includes: the method comprises the following steps of website performance knowledge expression, website performance knowledge extraction and website performance knowledge fusion.
The website performance knowledge expression specifically comprises the following steps:
structured, semi-structured, unstructured data about the performance of a website is obtained in a network dial-up test mode.
1) Structured data
Through a network dial testing technology, various performance indexes related to a specific website at the current moment can be measured, including PING, DNS, routing, IPv6 and the like, and corresponding structured data are generated. The structured data of the website performance indexes can be directly fused into the knowledge graph.
2) Semi-structured and unstructured data
Under an ideal state, the website performance knowledge graph can be built step by step based on the support of the structured website performance data. However, in the construction of the actual knowledge graph, not only the structured website performance data but also the semi-structured and unstructured website performance data need to be considered, which is an indispensable component and can be used as a powerful supplement for the structured website performance data. In addition, a small amount of unstructured website performance data can be extracted and utilized to enhance the integrity of the knowledge graph. Therefore, constructing a knowledge graph requires discriminating and processing a variety of data.
The website performance knowledge extraction specifically comprises the following steps:
ambiguous relationship names and relationship correspondences may exist between items in the semi-structured and unstructured data, so knowledge extraction including entity extraction, relationship extraction, and attribute extraction is performed on the items. The invention performs the relation extraction based on the Text Convolutional Neural network (Text Convolutional Neural Networks), as shown in fig. 4.
1) Data pre-processing
Before extracting the relation by using a text convolution neural network, preprocessing semi-structured data and unstructured data of website performance, and converting each word in the data into a k-dimensional word vector. For a website performance semi-structured data or unstructured data containing n words, an n x k matrix can be used for representation, and the n x k matrix is used as an input of the text convolution neural network.
2) Text convolutional neural network
The text convolution neural network provided by the invention is composed of a convolution layer, a pooling layer and a full-connection layer. Firstly, extracting high-dimensional semantic features from an input n x k website performance data matrix through a convolution layer, wherein the formula of convolution calculation is as follows:
where N is the layer number of each layer in the neural network, L(N)Is the feature vector input by Nth layer, L(N+1)Is the feature vector, W, output by the Nth layer(N)Is the weight vector of the Nth layer neural network, B(N)Is the bias vector for the nth layer.
To preserve the most discriminative features, the dimensionality of the feature map is reduced using maximal pooling. And fully connecting the pooled features with the m classified neurons of the last layer to obtain an m-dimensional vector. And finally, calculating the m-dimensional vector obtained in the last step by using a Softmax function to obtain probability values corresponding to the m categories respectively, and taking the category with the maximum probability value, namely the result of extracting the website performance relation, wherein the calculation formula of Softmax is as follows:
wherein z isiOf the ith vectorThe output value m is the number of output vectors, namely the number of classified categories, and the output value of the multi-classification is converted into the range of [0,1 ] by a Softmax function]To be distributed in the space between them.
The invention regards the relation extraction problem as a classification task, and trains a convolutional neural network model for extracting the website performance relation by using a random gradient descent method, so that the relation extraction is more intelligent and can be continuously learned.
The website performance knowledge fusion specifically comprises the following steps:
in the knowledge fusion process of the website performance knowledge base, because the triples of data acquired by knowledge expression and triples acquired by knowledge extraction cannot be guaranteed to be completely correct, the method performs entity alignment, entity disambiguation and attribute alignment, and corrects the knowledge in the website performance knowledge base.
1) Entity alignment
The method comprises the steps of judging whether two or more entities with different information sources belong to the same entity in the real world or not by using entity alignment in a knowledge graph, gathering entities representing the same object, constructing an alignment relation among the entities, and fusing information contained in the entities.
2) Entity disambiguation
Entity disambiguation disambiguates the word ambiguity based on contextual information in the data.
3) Attribute alignment
The attribute alignment judges whether two or more attributes can represent the same attribute, and information fusion is carried out on the attributes with different sources or different names but the same representation, so that richer and more accurate information is obtained.
Step S2, intelligent inference of website performance based on knowledge graph
2.1 Intelligent inference of website performance based on graph convolution network
On the basis of the website performance knowledge Graph constructed in the previous step, the method combines a Graph convolution network (Graph relational Networks) method to reason the website performance knowledge about various performance indexes extracted intelligently, and is shown in fig. 5.
2.1.1 partitioning sub-items
For knowledge inference of the graph, compared with inference based on a global structure, inference introducing a local structure enables feature granularity in a graph neural network to be finer and computation cost to be low. The network dial testing technology is used as a main body, sub-projects are divided, a local sub-graph of the website performance knowledge graph is obtained and used as the input of the graph convolution network, and the local sub-graph is constructed as shown in fig. 6.
2.1.2 graph convolution network
And (3) transmitting (send), receiving (receive), transforming (transform) each node of the graph, performing graph convolution, and performing nonlinear transformation of an activation function (ReLU) to complete a layer of computational transformation.
1) Graph convolution calculation
The graph convolution calculation is to perform graph fourier transform on the graph signal in the space domain, perform point-by-point product on the transformed data and the convolution kernel, and then recover the data to the space domain where the original graph signal is located by using inverse fourier transform, thereby completing the feature extraction of the graph signal, and the specific process is shown in fig. 7.
The graph fourier transform obtains a spectrogram of a graph signal by projecting the graph signal on an eigenvector of a laplacian matrix, which is obtained by subtracting a degree matrix and an adjacent matrix of the graph, as shown in fig. 8.
The data after graph convolution changes the value less than zero into zero and the positive value into the same value by using a ReLU function, so as to realize single-side suppression, wherein the function of the ReLU is as follows:
ReLU=max(0,x)
2.1.3 Multi-layer feature extraction
And repeating 2.1.2, inputting the output of the ReLU layer into the graph convolution layer, repeating graph convolution calculation and nonlinear transformation on the data, and continuously extracting the characteristics of the website performance until convergence so as to enable the characteristics to be higher and abstract.
And finally, after the characteristics are extracted for multiple times, the output result of the graph convolution neural network is a graph, the missing relationship in the graph is predicted in the graph, a comprehensive measurement result of the website performance after the knowledge reasoning is given, and the final website performance comprehensive knowledge reasoning is completed.
2.2 Intelligent Re-learning based optimization
As shown in FIG. 9, the present invention designs a web site performance knowledge graph with openness in view of sharing data with a third party web site performance knowledge base through an open API interface. The website performance knowledge map can be continuously supplemented and updated by fusing website performance knowledge from different sources, and the capability of intelligently reasoning the website performance is enhanced through the calculation of a convolutional neural network, so that the intelligent optimization based on relearning is realized. Since the website performance data of the third party knowledge base is usually structured, the invention makes it go directly to the knowledge fusion phase.
In the process of continuously updating the website performance knowledge graph, not only the website performance knowledge is updated and newly added, but also the conflict between the newly added knowledge and the old knowledge can be caused. Because the website performance knowledge has very strong timeliness, the invention preferentially tends to newly add the website performance knowledge to update the knowledge in the knowledge map.
As shown in fig. 11, the present invention further provides a website performance dial-up measurement device based on a knowledge graph, which includes the following steps:
a construction module for constructing a knowledge graph based on network dial testing
And the reasoning module is used for intelligently reasoning the website performance based on the graph convolution network according to the knowledge graph to realize the website performance dial testing measurement.
Further, the building module comprises:
the processing unit is used for obtaining structured, semi-structured and unstructured data related to the website performance by adopting a network dial-up test mode;
the extraction unit is used for performing knowledge extraction on the website performance semi-structured data and the unstructured data, and the knowledge extraction comprises entity extraction, relation extraction and attribute extraction;
and the fusion unit is used for carrying out entity alignment, entity disambiguation and attribute alignment on the result of the knowledge extraction so as to correct the knowledge in the website performance knowledge base.
Further, the extraction unit adopts a text-based convolutional neural network for relational extraction, wherein the text-based convolutional neural network is composed of a convolutional layer, a pooling layer and a full-link layer, and specifically comprises:
the preprocessing component is used for preprocessing the semi-structured data and the unstructured data of the website performance and converting each word in the data into a k-dimensional word vector;
the extraction component is used for extracting high-dimensional semantic features from the input n x k website performance data matrix through the convolution layer, wherein n x k represents website performance semi-structured data or unstructured data containing n words;
and the extraction component is used for fully connecting the pooled features with the m classified neurons of the last layer through the dimensionality of the maximum pooling reduction feature graph of the high-dimensional semantic features to obtain an m-dimensional vector, and calculating the m-dimensional vector by using a Softmax function to obtain probability values respectively corresponding to the m categories, namely a website performance relation extraction result.
Further, the metric module includes:
the dividing unit is used for dividing the website performance knowledge into sub-items as the input of the graph convolution network;
the first calculation unit is used for carrying out graph convolution on each node of the graph by adopting transmission, reception and transformation, and then completing one layer of calculation transformation through the nonlinear transformation of an activation function ReLU;
the second calculation unit is used for inputting the output of the ReLU layer into the graph convolution layer, repeatedly using graph convolution calculation and nonlinear transformation and continuously extracting the characteristics of the website performance;
and the measurement unit is used for finishing the final website performance comprehensive knowledge inference after the characteristics are extracted for multiple times and giving a measurement result of the website performance after the knowledge inference.
The website performance dial-up measurement method and device based on the knowledge graph integrates the website IP, speed measurement and DNS information, establishes the website security knowledge graph, and improves the website hijacking, wall-sticking and pollution detection capabilities. In addition, the method supports the input of various website performance data, can improve the updating efficiency of the knowledge base in the knowledge map, and ensures the timeliness and the accuracy of the website performance knowledge so as to provide a more accurate evaluation report.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.
Claims (10)
1. A website performance dial testing measurement method based on a knowledge graph is characterized by comprising the following steps:
step 1, establishing a knowledge graph based on network dial testing
And 2, intelligently reasoning the website performance based on the graph convolution network according to the knowledge graph to realize the website performance dial testing measurement.
2. The method for website performance dial-up measurement based on knowledge-graph as claimed in claim 1, wherein step 1 specifically comprises:
step 11, obtaining structured data, semi-structured data and unstructured data about website performance by adopting a network dial-up test mode;
step 12, extracting knowledge from the website performance semi-structured data and the unstructured data;
and step 13, carrying out entity alignment, entity disambiguation and attribute alignment on the result of the knowledge extraction so as to correct the knowledge in the website performance knowledge base.
3. The method of claim 2, wherein the knowledge extraction comprises an entity extraction, a relationship extraction, and an attribute extraction.
4. The method according to claim 2, wherein in step 12, the relation extraction is performed by using a text-based convolutional neural network, the text-based convolutional neural network is composed of a convolutional layer, a pooling layer and a full link layer, and specifically comprises:
step 121, preprocessing the semi-structured data and the unstructured data of the website performance, and converting each word in the data into a k-dimensional word vector; for a website performance semi-structured data or unstructured data containing n words, adopting an n x k matrix for representation, and taking the n x k matrix as the input of a text convolution neural network;
step 122, extracting high-dimensional semantic features from the input n × k website performance data matrix through the convolution layer, wherein the formula of the convolution calculation is as follows:
where N is the layer number of each layer in the neural network, L(N)Is the feature vector input by Nth layer, L(N+1)Is the feature vector, W, output by the Nth layer(N)Is the weight vector of the Nth layer neural network, B(N)Is the bias vector for the nth layer;
step 123, performing maximum pooling reduction on the dimensionality of the feature map on the high-dimensional semantic features, and fully connecting the pooled features with the m classified neurons in the last layer to obtain an m-dimensional vector; calculating the m-dimensional vector by using a Softmax function to obtain probability values corresponding to the m categories respectively, and taking the category with the maximum probability value, namely the result of extracting the website performance relationship, wherein the calculation formula of Softmax is as follows:
wherein z isiConverting the output value of the multi-classification into the range of [0,1 ] by a Softmax function for the output value of the ith vector and m for the number of output vectors, namely the number of classified classes]To be distributed in the space between them.
5. The method for website performance dial-up measurement based on knowledge-graph as claimed in claim 2, wherein step 2 specifically comprises:
step 21, dividing website performance knowledge into sub-items as input of a graph volume network;
step 22, carrying out graph convolution on each node of the graph by adopting transmitting, receiving and transformation, and then completing a layer of computational transformation through the nonlinear transformation of an activation function ReLU;
step 23, repeating step 22, inputting the output of the ReLU layer into the graph convolution layer, repeatedly using graph convolution calculation and nonlinear transformation, and continuously extracting the characteristics of the website performance;
and 24, after the features are extracted for multiple times, finishing the final comprehensive knowledge reasoning of the website performance and giving a measurement result of the website performance after the knowledge reasoning.
6. A website performance dial-up measurement device based on a knowledge graph is characterized by comprising:
the construction module is used for constructing a knowledge graph based on network dial testing;
and the reasoning module is used for intelligently reasoning the website performance based on the graph convolution network according to the knowledge graph to realize the website performance dial testing measurement.
7. The apparatus according to claim 6, wherein the building module comprises:
the processing unit is used for obtaining structured, semi-structured and unstructured data related to the website performance by adopting a network dial-up test mode;
the extraction unit is used for carrying out knowledge extraction on the website performance semi-structured data and the unstructured data;
and the fusion unit is used for carrying out entity alignment, entity disambiguation and attribute alignment on the result of the knowledge extraction so as to correct the knowledge in the website performance knowledge base.
8. The apparatus according to claim 7, wherein the knowledge extraction comprises an entity extraction, a relationship extraction, and an attribute extraction.
9. The apparatus according to claim 7, wherein the extracting unit performs the relationship extraction by using a convolutional neural network based on text, the convolutional neural network is composed of a convolutional layer, a pooling layer and a full link layer, and specifically comprises:
the preprocessing component is used for preprocessing the semi-structured data and the unstructured data of the website performance and converting each word in the data into a k-dimensional word vector;
the extraction component is used for extracting high-dimensional semantic features from the input n x k website performance data matrix through the convolution layer, wherein n x k represents website performance semi-structured data or unstructured data containing n words;
and the extraction component is used for fully connecting the pooled features with the m classified neurons of the last layer through the dimensionality of the maximum pooling reduction feature graph of the high-dimensional semantic features to obtain an m-dimensional vector, and calculating the m-dimensional vector by using a Softmax function to obtain probability values respectively corresponding to the m categories, namely a website performance relation extraction result.
10. The apparatus of claim 6, wherein the metrics module comprises:
the dividing unit is used for dividing the website performance knowledge into sub-items as the input of the graph convolution network;
the first calculation unit is used for carrying out graph convolution on each node of the graph by adopting transmission, reception and transformation, and then completing one layer of calculation transformation through the nonlinear transformation of an activation function ReLU;
the second calculation unit is used for inputting the output of the ReLU layer into the graph convolution layer, repeatedly using graph convolution calculation and nonlinear transformation and continuously extracting the characteristics of the website performance;
and the measurement unit is used for finishing the final website performance comprehensive knowledge inference after the characteristics are extracted for multiple times and giving a measurement result of the website performance after the knowledge inference.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111067717.4A CN113824707A (en) | 2021-09-13 | 2021-09-13 | Website performance dial testing measurement method and device based on knowledge graph |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111067717.4A CN113824707A (en) | 2021-09-13 | 2021-09-13 | Website performance dial testing measurement method and device based on knowledge graph |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113824707A true CN113824707A (en) | 2021-12-21 |
Family
ID=78914403
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111067717.4A Pending CN113824707A (en) | 2021-09-13 | 2021-09-13 | Website performance dial testing measurement method and device based on knowledge graph |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113824707A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114861072A (en) * | 2022-07-05 | 2022-08-05 | 浙商银行股份有限公司 | Graph convolution network recommendation method and device based on interlayer combination mechanism |
CN114978586A (en) * | 2022-04-12 | 2022-08-30 | 东北电力大学 | Power grid attack detection method and system based on attack genes and electronic equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190019088A1 (en) * | 2017-07-14 | 2019-01-17 | Guangdong Shenma Search Technology Co., Ltd. | Knowledge graph construction method and device |
CN111428054A (en) * | 2020-04-14 | 2020-07-17 | 中国电子科技网络信息安全有限公司 | Construction and storage method of knowledge graph in network space security field |
CN112148888A (en) * | 2020-09-18 | 2020-12-29 | 南京邮电大学 | Knowledge graph construction method based on graph neural network |
CN112699280A (en) * | 2020-12-31 | 2021-04-23 | 北京天融信网络安全技术有限公司 | Website monitoring method, website map establishing method and device and electronic equipment |
CN113010691A (en) * | 2021-03-30 | 2021-06-22 | 电子科技大学 | Knowledge graph inference relation prediction method based on graph neural network |
-
2021
- 2021-09-13 CN CN202111067717.4A patent/CN113824707A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190019088A1 (en) * | 2017-07-14 | 2019-01-17 | Guangdong Shenma Search Technology Co., Ltd. | Knowledge graph construction method and device |
CN111428054A (en) * | 2020-04-14 | 2020-07-17 | 中国电子科技网络信息安全有限公司 | Construction and storage method of knowledge graph in network space security field |
CN112148888A (en) * | 2020-09-18 | 2020-12-29 | 南京邮电大学 | Knowledge graph construction method based on graph neural network |
CN112699280A (en) * | 2020-12-31 | 2021-04-23 | 北京天融信网络安全技术有限公司 | Website monitoring method, website map establishing method and device and electronic equipment |
CN113010691A (en) * | 2021-03-30 | 2021-06-22 | 电子科技大学 | Knowledge graph inference relation prediction method based on graph neural network |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114978586A (en) * | 2022-04-12 | 2022-08-30 | 东北电力大学 | Power grid attack detection method and system based on attack genes and electronic equipment |
CN114978586B (en) * | 2022-04-12 | 2023-07-04 | 东北电力大学 | Power grid attack detection method and system based on attack genes and electronic equipment |
CN114861072A (en) * | 2022-07-05 | 2022-08-05 | 浙商银行股份有限公司 | Graph convolution network recommendation method and device based on interlayer combination mechanism |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111159425A (en) | Temporal knowledge graph representation method based on historical relationship and double-graph convolution network | |
CN114816909B (en) | Real-time log detection early warning method and system based on machine learning | |
CN110837602A (en) | User recommendation method based on representation learning and multi-mode convolutional neural network | |
CN110138595A (en) | Time link prediction technique, device, equipment and the medium of dynamic weighting network | |
CN113824707A (en) | Website performance dial testing measurement method and device based on knowledge graph | |
CN113783874B (en) | Network security situation assessment method and system based on security knowledge graph | |
CN113326377A (en) | Name disambiguation method and system based on enterprise incidence relation | |
CN112231305A (en) | Digital power grid system and method based on digital twinning | |
CN112182306B (en) | Uncertain graph-based community discovery method | |
CN112700056A (en) | Complex network link prediction method, complex network link prediction device, electronic equipment and medium | |
CN113032238A (en) | Real-time root cause analysis method based on application knowledge graph | |
CN114462520A (en) | Network intrusion detection method based on traffic classification | |
CN111367908A (en) | Incremental intrusion detection method and system based on security assessment mechanism | |
Li et al. | Street-Level Landmarks Acquisition Based on SVM Classifiers. | |
CN111562541A (en) | Software platform for realizing electric energy meter detection data management by applying CART algorithm | |
CN117312578B (en) | Construction method and system of non-genetic carrier spectrum | |
CN118095341A (en) | SimRank similarity calculation method based on deep neural network | |
CN114254738A (en) | Double-layer evolvable dynamic graph convolution neural network model construction method and application | |
CN116629362A (en) | Interpreteable time graph reasoning method based on path search | |
CN114757391B (en) | Network data space design and application method oriented to service quality prediction | |
CN106296420A (en) | A kind of community discovery method | |
CN114793170A (en) | DNS tunnel detection method, system, equipment and terminal based on open set identification | |
CN114385827A (en) | Conference knowledge graph-oriented retrieval method | |
CN113962748A (en) | Method for aligning users of heterogeneous e-commerce platform by using holomorphic information representation based on meta-path | |
CN113255977A (en) | Intelligent factory production equipment fault prediction method and system based on industrial internet |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211221 |
|
RJ01 | Rejection of invention patent application after publication |