Internet inter-domain interconnection view inference method and system based on multi-source data

Abstract

The invention relates to the technical field of computers, in particular to an internet inter-domain interconnection view inference method and system based on multi-source data. The method comprises the steps that multi-data-source information provided by multiple different types of data sources is acquired, a first interconnection information set is determined according to the multi-data-source information, and first interconnection information in the first interconnection information set comprises autonomous system identification pairs and interconnection positions; performing interconnection position inference according to the first interconnection information set to obtain a second interconnection information set; and according to the multi-data source information, the first interconnection information set and the second interconnection information set, carrying out interconnection facility inference to obtain an internet inter-domain interconnection view inference result. By adopting the scheme, the Internet logic and physical association view which is high in coverage and easy to update can be constructed.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a method and system for inferring an Internet inter-domain interconnection view based on multi-source data. Background Art

[0002] With the rapid development of computer network technology, the Internet has become an indispensable infrastructure for modern society. Within the vast Internet system, measuring inter-domain interconnection views is crucial for understanding network structure, optimizing network performance, and ensuring network security. Regarding the construction of inter-domain interconnection views on the Internet, mature work and technologies have primarily focused on constructing logical views. However, logical-level views have certain limitations. They struggle to accurately represent the physical interdependencies between networks and reflect the actual network operation and physical interdependencies. Summary of the Invention

[0003] The present disclosure aims to solve one of the technical problems in the related art at least to a certain extent.

[0004] To this end, the first purpose of the present disclosure is to propose an Internet inter-domain interconnection view inference method based on multi-source data to construct a high-coverage, easy-to-update Internet logical and physical association view.

[0005] The second objective of the present disclosure is to propose an Internet inter-domain interconnection view inference system based on multi-source data.

[0006] A third objective of the present disclosure is to provide an electronic device.

[0007] A fourth object of the present disclosure is to provide a computer-readable storage medium.

[0008] A fifth object of the present disclosure is to provide a computer program product.

[0009] To achieve the above objectives, the first embodiment of the present disclosure proposes a method for inferring an Internet inter-domain interconnection view based on multi-source data, comprising:

[0010] Acquire multiple data source information provided by multiple different types of data sources, and determine a first interconnection information set based on the multiple data source information, wherein the first interconnection information in the first interconnection information set includes an autonomous system identifier pair and an interconnection location;

[0011] performing interconnection location inference based on the first interconnection information set to obtain a second interconnection information set, wherein the second interconnection information in the second interconnection information set includes the autonomous system identifier pair and the interconnection location obtained after performing the interconnection location inference;

[0012] Interconnection facilities are inferred based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain an Internet inter-domain interconnection view inference result, wherein the Internet inter-domain interconnection view inference result includes a third interconnection information set, and the third interconnection information in the third interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type.

[0013] Optionally, performing interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain an Internet inter-domain interconnection view inference result includes:

[0014] performing an empirical rule-based interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain a fourth interconnection information set, wherein fourth interconnection information in the fourth interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type;

[0015] An interconnection facility inference model is used to perform interconnection facility inference based on the fourth interconnection information set to obtain an Internet inter-domain interconnection view inference result.

[0016] Optionally, the method further includes:

[0017] extracting autonomous system data features from the multiple data source information;

[0018] constructing an interconnection location inference model based on the first interconnection information set and the autonomous system data characteristics, wherein the interconnection location inference model is used to perform interconnection location inference based on the first interconnection information set to obtain a second interconnection information set;

[0019] The interconnection facility inference model is constructed according to the fourth interconnection information set and the autonomous system data characteristics.

[0020] Optionally, performing interconnection location inference based on the first interconnection information set to obtain a second interconnection information set includes:

[0021] Determining a distribution location set of each autonomous system based on the multiple data source information;

[0022] Combining the distribution location sets between different autonomous systems to obtain a pending interconnection information set;

[0023] determining, based on the first interconnection information set, an interconnection probability of each pending interconnection information in the pending interconnection information set;

[0024] The pending interconnection information whose interconnection probability meets the interconnection probability requirement is determined from the pending interconnection information set to obtain a second interconnection information set.

[0025] Optionally, determining the interconnection probability of each pending interconnection information in the pending interconnection information set according to the first interconnection information set includes:

[0026] Performing a first number of samplings with replacement on the first interconnected information set to obtain a first number of positive sub-sample sets;

[0027] Performing the first number of samplings with replacement on the pending interconnection information set to obtain a first number of temporary negative sample sets, and determining a set to be inferred corresponding to each of the temporary negative sample sets, wherein the pending interconnection information in the set to be inferred is the pending interconnection information in the pending interconnection information set excluding the temporary negative sample set;

[0028] Selecting a sub-positive sample set and a temporary negative sample set in turn, combining them, and training the initial binary classifier to obtain a first number of trained binary classifiers;

[0029] The trained binary classifier is used to perform probability inference on each pending interconnection information in the pending inference set corresponding to the trained binary classifier, so as to obtain the interconnection probability of each pending interconnection information in the pending interconnection information set.

[0030] Optionally, performing a first number of sampling with replacement on the first interconnected information set includes:

[0031] grouping the first interconnected information set according to the first confidence level of each piece of first interconnected information in the first interconnected information set to obtain a plurality of first interconnected information subsets;

[0032] determining, according to the first confidence level, a sampling ratio of each first interconnected information subset when performing sampling with replacement on the first interconnected information set;

[0033] A first number of samplings with replacement are performed on the first interconnected information set according to the sampling ratio.

[0034] Optionally, the interconnection location includes a country location and a city location, and determining, based on the first interconnection information set, an interconnection probability of each pending interconnection information in the pending interconnection information set includes:

[0035] Determining a first subset of pending interconnection information at a country location level in the pending interconnection information set;

[0036] determining, based on the first interconnection information set, a first interconnection probability for each piece of pending interconnection information in the first pending interconnection information subset;

[0037] Determining, based on the first interconnection probability, a second subset of pending interconnection information at a city location level in the pending interconnection information set;

[0038] A second interconnection probability of each piece of pending interconnection information in the second pending interconnection information subset is determined according to the first interconnection information set.

[0039] Optionally, the interconnection facility inference model is a random forest model, and constructing the interconnection facility inference model according to the fourth interconnection information set and the autonomous system data characteristics includes:

[0040] Performing a second number of sampling with replacement on the fourth interconnected information set to obtain a second number of sub-training sets;

[0041] For each sub-training set, recursive splitting is started from the root node, and at each node, a third number of candidate features are randomly selected from the autonomous system data features corresponding to the sub-training set, and the optimal candidate features and splitting threshold are selected according to the evaluation index until the stopping condition is met, thereby obtaining a random forest model consisting of a second number of sub-decision trees.

[0042] Optionally, the autonomous system data feature includes at least one of the following:

[0043] Autonomous system attribute characteristics, wherein the autonomous system attribute characteristics include at least one of autonomous system customer cone, number of IP prefixes, and number of IP addresses;

[0044] Autonomous system interconnection characteristics, wherein the autonomous system interconnection characteristics include at least one of autonomous system business relationships and the number of autonomous system neighbors;

[0045] Geographical and facility characteristics, wherein the geographical and facility characteristics include at least one of geographical distance and the number of submarine cable landing points.

[0046] To achieve the above objectives, a second embodiment of the present disclosure proposes an Internet inter-domain interconnection view inference system based on multi-source data, comprising:

[0047] an information determining unit, configured to obtain multi-data source information provided by multiple different types of data sources, and determine a first interconnection information set based on the multi-data source information, wherein the first interconnection information in the first interconnection information set includes an autonomous system identifier pair and an interconnection location;

[0048] a location inference unit, configured to perform interconnection location inference based on the first interconnection information set to obtain a second interconnection information set, wherein the second interconnection information in the second interconnection information set includes the autonomous system identifier pair and the interconnection location obtained after performing the interconnection location inference;

[0049] A facility inference unit is configured to perform interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain an Internet inter-domain interconnection view inference result, wherein the Internet inter-domain interconnection view inference result includes a third interconnection information set, and the third interconnection information in the third interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type.

[0050] To achieve the above-mentioned objectives, a third embodiment of the present disclosure provides an electronic device, including:

[0051] a memory for storing executable program code;

[0052] The processor is used to call and run the executable program code from the memory, so that the electronic device executes the method shown in any one of the first aspects above.

[0053] To achieve the above-mentioned purpose, the fourth aspect of the present disclosure provides a computer-readable storage medium, which stores a computer program. When the computer program is executed, it implements the method shown in any one of the above-mentioned first aspects.

[0054] To achieve the above-mentioned objectives, an embodiment of the fifth aspect of the present disclosure proposes a computer program product, including a computer program, which implements the method shown in any one of the above-mentioned first aspects when executed by a processor.

[0055] In summary, the method and system provided by the present disclosure can build a high-coverage, easily updated view of the logical and physical associations of the Internet by making full use of multi-source data to infer and associate the interconnected locations and facility types between autonomous systems.

[0056] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The above and / or additional aspects and advantages of the present disclosure will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0058] Figure 1 A flowchart of a method for inferring an Internet inter-domain interconnection view based on multi-source data provided by an embodiment of the present disclosure;

[0059] Figure 2 A flowchart of a method for inferring an Internet inter-domain interconnection view based on multi-source data provided by another embodiment of the present disclosure;

[0060] Figure 3A schematic diagram of the structure of an Internet inter-domain interconnection view inference system based on multi-source data provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0061] The following describes in detail embodiments of the present disclosure, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0062] It's important to note that constructing a view of inter-domain Internet interconnection using single-dimensional data only focuses on certain characteristics. For example, while using only traceroute or Border Gateway Protocol (BGP) routing table (Routing Information Base, RIB) data to construct a logical view, it fails to reveal the business relationships between autonomous systems (ASs) or to construct the corresponding physical topology. Data from PeeringDB and Euro-IX, on the other hand, primarily focuses on large Internet Exchange Points (IXPs) and related facilities in Europe. In other words, using only single-dimensional data can easily lead to biased results and fail to fully reflect the complexity of the Internet. Therefore, inferring an inter-domain Internet interconnection view requires relying on multiple different types of data sources, each providing information at varying granularity and in different aspects. The heterogeneity, diversity, and complexity of these data sources in multi-dimensional data pose significant challenges to constructing an inter-domain Internet interconnection view.

[0063] First, the confidence and relevance of multidimensional data vary. For example, the accuracy of geolocation mapping tools can vary by region, and data in some areas may be inaccurate. There's also no direct mapping between data from different sources and dimensions. Therefore, finding an integration algorithm that balances these differences is difficult.

[0064] At the same time, some existing data sources have significant limitations in terms of accuracy and completeness. For example, most IP geolocation databases focus primarily on user IP addresses, which can lead to issues with router IP location accuracy and may even only be able to locate at the national or regional level. This can lead to misjudgments of AS interconnection locations, thereby affecting the accuracy of the interconnection view. For another example, while data sources such as PeeringDB and Euro-IX provide rich interconnection infrastructure information, this data relies primarily on proactive updates and maintenance by network operators. Therefore, some data may not have been updated in years and may not reflect dynamic changes in the network in a timely manner. Due to the accuracy limitations of different data sources, the process of building an interconnection view may also encounter information conflicts between different data sources, further increasing the complexity of data fusion.

[0065] Second, existing observational data has significant limitations in coverage. While some measurement platforms have deployed multiple measurement points globally, these points are unevenly distributed, with insufficient data coverage in certain regions. The observable routing data cannot effectively cover global AS connectivity, and the number of available samples is limited. Furthermore, while some open communities provide AS interconnection information, this data is released voluntarily by network operators and is therefore primarily focused on ASes within large IXPs and large companies, with insufficient coverage of smaller or private interconnection points. Furthermore, coverage varies significantly between different data sources. For example, the overlap between traceroute and BGP data is low, making it difficult to adopt simple methods such as intersection analysis during data fusion, as this would otherwise make it difficult to form a complete interconnection view.

[0066] Furthermore, regarding the construction of inter-domain interconnection views on the Internet, mature related work and technologies primarily focus on constructing logical views, including inferring AS business relationships and AS path inference, resulting in a certain dataset. However, logical-level views have certain limitations. They struggle to accurately represent the physical interdependencies between networks and reflect the actual network operation and physical interdependencies. Research on the correlation between logical and physical views has primarily fallen into two categories: one based on the extraction and simple matching of public data, and the other relying on web crawlers and LG active measurement. Research methods based on the extraction and simple matching of public data often face coverage issues. Due to the limited sources of public data and the lack of timely updates, they fail to fully capture network interconnection information. Furthermore, with the continuous advancement of network protection measures, methods that rely on web crawlers and LG active measurement are becoming increasingly difficult to automatically obtain data, and may even be restricted or prohibited.

[0067] The present disclosure is described in detail below with reference to specific embodiments.

[0068] In the first embodiment, if Figure 1 As shown, Figure 1 This is a flow chart of a method for inferring an internet inter-domain interconnection view based on multi-source data, provided in an embodiment of the present disclosure. This method can be implemented using a computer program and run on a system for inferring an internet inter-domain interconnection view based on multi-source data. The computer program can be integrated into an application or run as a standalone tool application.

[0069] Among them, the Internet inter-domain interconnection view inference method based on multi-source data can be executed by an electronic device.

[0070] For example, the method for inferring an Internet inter-domain interconnection view based on multi-source data includes the following steps:

[0071] S101, obtaining multi-data source information provided by multiple different types of data sources, and determining a first interconnected information set based on the multi-data source information;

[0072] According to some embodiments, the first interconnection information in the first interconnection information set includes an autonomous system identifier pair and an interconnection location.

[0073] In some embodiments, at the logical view level, a pair of AS interconnections serves as a basic unit and can be represented as an autonomous system identifier pair, specifically in a two-tuple format (AS1, AS2).

[0074] In some embodiments, at the physical view level, the interconnection location indicates that (AS1, AS2) are interconnected at location C. This interconnection location may include multiple levels of locations. For example, the interconnection location may be accurate to the country level or the city level, including country location C1 and city location C2. In this case, the first interconnection information may be represented as a four-tuple (AS1, AS2, C1, C2), indicating that AS1 and AS2 are interconnected at country location C1 and city location C2.

[0075] S102, performing interconnection location inference based on the first interconnection information set to obtain a second interconnection information set;

[0076] According to some embodiments, the second interconnection information in the second interconnection information set includes an autonomous system identifier pair and an interconnection location obtained after performing interconnection location inference.

[0077] It should be noted that the autonomous system identifier pair and the interconnection location obtained after the interconnection location inference are the autonomous system identifier pair and the interconnection location further inferred in addition to the first interconnection information set.

[0078] S103: Inferring interconnection facilities based on the multi-data source information, the first interconnection information set, and the second interconnection information set to obtain an Internet inter-domain interconnection view inference result.

[0079] According to some embodiments, the inferred result of the inter-domain internet interconnection view includes a third interconnection information set, wherein the third interconnection information in the third interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnected device type. The third interconnection information can serve as basic data for the inferred result of the inter-domain internet interconnection view to construct the inter-domain internet interconnection view.

[0080] For example, the third interconnection information can be represented as a quintuple (AS1, AS2, C1, C2, F), indicating that AS1 and AS2 are interconnected in country C1 and city C2 via infrastructure type F. At the logical level of the Internet inter-domain interconnection view, (AS1, AS2) can be used as a key to associate with the physical view, and thus associated with (C1, C2, F) in the physical view.

[0081] In summary, the method provided in this embodiment can build a high-coverage, easily updated view of the logical and physical associations of the Internet by making full use of multi-source data to infer and associate the interconnected locations and facility types between autonomous systems.

[0082] Another embodiment of the present disclosure provides a method for inferring an Internet inter-domain interconnection view based on multi-source data. The method can be executed by an electronic device.

[0083] For example, the method for inferring an Internet inter-domain interconnection view based on multi-source data may include the following steps:

[0084] S201, obtaining multi-data source information provided by multiple different types of data sources, and determining a first interconnected information set based on the multi-data source information;

[0085] According to some embodiments, Figure 2 This is a flow chart of a method for inferring an Internet domain interconnection view based on multi-source data provided by another embodiment of the present disclosure. Figure 2 As shown, the first interconnected information set can be determined based on information from multiple data sources based on empirical rules; wherein, the empirical rules include but are not limited to empirical rules based on physical facility information positioning, empirical rules based on single area edge AS positioning, empirical rules based on path detection results positioning, etc.

[0086] For example, data sources like PeeringDB and Euro-IX provide rich interconnection facility information, including the city where the IXP or facility is located, member ASes, and the interconnection policies of these member ASes. Examining all ASes connected to the same IXP, if the business relationship between the interconnected ASes is inferred to be peer-to-peer (p2p), and both ASes use a relatively loose interconnection policy (Open or Selective), then it can be determined that the two ASes have established a connection at the IXP (facility type: IXP) and are interconnected in the city where the IXP is located. Similarly, examining all ASes connected to the same facility, if at least one of the interconnected ASes uses a relatively strict interconnection policy (Restrictive or No), then it can be determined that both ASes have established a connection at the facility (facility type: Direct Connection) and are interconnected in the city where the facility is located.

[0087] It should be noted that the interconnected AS pair must be a neighboring AS that appears in the BGP routing table collected by the observation point (Vantage Point, VP).

[0088] For example, edge ASes are typically located at the bottom of the network hierarchy and do not provide paid traffic forwarding services to other ASes. Their primary function is to provide network access services to local users, maintaining a relatively fixed position within the network. Therefore, we can first identify edge ASes from inferred AS business relationship data. These ASes act as customer ASes or peer ASes in business relationships. We can then construct a set of edge ASes by filtering the customer cone to include only ASes within the AS itself. Next, based on the AS prefix data provided by RouteViews and the longest prefix match principle, we calculate the IP address ranges owned by all ASes. Subsequently, we randomly sample an IP address from every / 24 range within the AS's IP address range. Using an IP geolocation mapping tool (such as MaxMind GeoLite2), we map these sampled IP addresses to specific geographic locations, thereby determining the AS's distribution area. We can then examine the distribution areas of edge ASes and their interconnected ASes. If an edge AS is located in a single region, and its interconnected ASes are also distributed within that region, we can assume that the connection between the two ASes occurs within that region, and the facility type is direct.

[0089] For example, the actual traffic paths within an AS network can be determined by collecting traceroute data regularly measured by Ark and RIPE Atlas, i.e., IP-level path information. The bdrmapIT tool developed by CAIDA is then used to map each IP on the traceroute path to an AS. For sequences of IPs with the same consecutive AS identification number after mapping, the first and last IP addresses can be used as the border router IP addresses. Inter-domain interconnection occurs between border routers in two adjacent ASes. Next, simply geo-map the border router IP addresses to determine the location of the AS interconnection. Finally, the ki3 platform is used to obtain the location distribution information of existing submarine cables and their landing points worldwide, and to examine AS interconnections that have been located. If the landing point of the same submarine cable exists near both ends (within 100 kilometers) and the two ends are separated by ocean, the AS pair can be considered interconnected via the submarine cable.

[0090] S202, extracting autonomous system data features from multiple data source information, and constructing an interconnected location inference model based on the first interconnected information set and the autonomous system data features;

[0091] According to some embodiments, the interconnection location inference model is used to perform interconnection location inference based on the first interconnection information set to obtain the second interconnection information set.

[0092] In some embodiments, the interconnected location inference model includes, but is not limited to, a Bagging-PU classifier. Bagging-PU (Bootstrap Aggregating for Positive and Unlabeled Learning) is an ensemble learning method designed for positive and unlabeled learning (PU Learning) scenarios. Its core concept is to alleviate the class imbalance and noise interference problems in PU data through bootstrap sampling and classifier integration.

[0093] According to some embodiments, the autonomous system data features include but are not limited to at least one of autonomous system attribute features, autonomous system interconnection features, and geographical and facility features.

[0094] In some embodiments, the autonomous system attribute feature includes but is not limited to at least one of an autonomous system customer cone (CustomerCone), the number of IP prefixes, and the number of IPs.

[0095] The autonomous system customer cone refers to an AS and the collection of all its downstream customers.

[0096] The number of IP prefixes refers to the number of prefixes owned by the AS, that is, the total number of IP prefixes announced by the AS.

[0097] The number of IP addresses refers to the number of IP addresses owned by the AS, that is, the total number of IP addresses managed by the AS. It can be estimated by the number of prefixes owned by the AS and the size of each prefix.

[0098] In some embodiments, the autonomous system interconnection feature includes but is not limited to at least one of an autonomous system business relationship and a number of autonomous system neighbors.

[0099] The autonomous system business relationship refers to the business relationship type between ASes, namely, peer-to-peer interconnection, customer-provider, and provider-customer relationship, which can be represented by 0, 1, and -1 respectively.

[0100] The number of autonomous system neighbors refers to the number of three types of neighboring ASs (providers, customers, and peers) that the AS is directly connected to.

[0101] In some embodiments, the geographical and facility characteristics include, but are not limited to, at least one of geographical distance and the number of submarine cable landing points.

[0102] The geographic distance refers to the top five closest straight-line distances between two ASes (if the number is less than five, infinity is used to fill the gap), which can be calculated based on the latitude and longitude information obtained by sampling.

[0103] The number of submarine cable landing points refers to the number of landing points belonging to the same submarine cable in the areas where the two ASs are located.

[0104] S203: Using an interconnection location inference model to perform interconnection location inference based on the first interconnection information set to obtain a second interconnection information set.

[0105] According to some embodiments, the interconnection location inference model can determine the distribution location set of each autonomous system based on information from multiple data sources; combine the distribution location sets between different autonomous systems to obtain a pending interconnection information set; determine the interconnection probability of each pending interconnection information in the pending interconnection information set based on the first interconnection information set; determine the pending interconnection information whose interconnection probability meets the interconnection probability requirements from the pending interconnection information set to obtain a second interconnection information set.

[0106] In some embodiments, the distribution of each AS can be obtained by sampling the IP address segments of the AS and combining them with IP geolocation mapping. The sampled locations can be clustered using the DBSCAN algorithm with a radius no greater than 60 kilometers to determine the distribution area of ​​the edge AS.

[0107] In some embodiments, when combining the distribution location sets between different autonomous systems, two autonomous systems can be selected in sequence and the distribution location sets of the two autonomous systems can be combined. For example, the different locations distributed by AS1 and AS2 can be combined to obtain all possible interconnected locations between them as unlabeled samples.

[0108] In some embodiments, the interconnection probability requirement may be, for example, that the interconnection probability is greater than an interconnection probability threshold. The interconnection probability threshold is not specifically a fixed threshold, and the interconnection probability threshold may be determined according to actual application scenarios.

[0109] According to some embodiments, a first number of samplings with replacement can be performed on the first interconnected information set to obtain a first number of sub-positive sample sets; a first number of samplings with replacement can be performed on the pending interconnected information set to obtain a first number of temporary negative sample sets, and a set to be inferred corresponding to each temporary negative sample set is determined, wherein the pending interconnected information in the set to be inferred is the pending interconnected information in the set to be inferred excluding the temporary negative sample set; a sub-positive sample set and a temporary negative sample set are selected in turn and combined to train the initial binary classifier to obtain a first number of trained binary classifiers; the trained binary classifier is used to perform probability inference on each piece of pending interconnected information in the set to be inferred corresponding to the trained binary classifier to obtain the interconnection probability of each piece of pending interconnected information in the set to be inferred.

[0110] In some embodiments, the initial binary classifier may be, for example, a random forest model, a decision tree, a support vector machine (SVM), or the like.

[0111] Taking a scenario as an example, the first Internet information collection Pending Internet Information Collection The initial binary classifier f and the number of sampling times (i.e., the first number) S are used as inputs, and the interconnection probability of each pending interconnection information in the pending interconnection information set is determined according to the following steps:

[0112] S2031, sampling: Perform S sampling with replacement to obtain S sub-positive sample sets Similarly, from Randomly select samples from the set to form S temporary negative samples The remaining part of the set to be inferred

[0113] S2032, binary classifier training: for each subset Train the binary classifier f i ;

[0114] S2033, Probabilistic Inference: For each set to be inferred Use the trained binary classifier f i Make probability inferences for each of these samples;

[0115] S2034, result integration: Assume that any unlabeled sample has not been drawn in the first Sk times, and is drawn from the S-k+1th to the Sth times (i.e., the last k times), and is assigned probability pi in the i-th inference probability. Then the final probability inference result is integrated as follows:

[0116]

[0117] It should be noted that the first interconnection information set includes both high-confidence positive samples and low-confidence positive samples. For example, ITDK provides mappings between nodes and ASs, nodes and locations, and provides information on the interconnections between nodes. If interconnected nodes belong to a pair of interconnected ASs, it can be inferred that these ASes are interconnected at the node's location. However, ITDK's topology data may contain significant errors, so the positive samples generated by ITDK data are low-confidence positive samples. Subsequently, based on the AS connections provided by AS Rank, the ASs at both ends of the connection can be examined separately.

[0118] In this case, when performing a first number of samplings with replacement on the first interconnected information set, the first interconnected information set can be grouped according to the first confidence level of each piece of first interconnected information in the first interconnected information set to obtain a plurality of first interconnected information subsets; a sampling ratio for each first interconnected information subset when performing sampling with replacement on the first interconnected information set can be determined according to the first confidence level; and a first number of samplings with replacement can be performed on the first interconnected information set according to the sampling ratio. Therefore, by grouping positive samples according to confidence level and setting different sampling ratios during the sampling process, the challenge of limited data source accuracy can be overcome, information from different data sources can be compared and cross-validated, and obviously erroneous or contradictory data can be identified and eliminated. At the same time, a confidence assessment mechanism can be introduced to weight the data by distinguishing confidence levels, thereby controlling the impact of errors in the data source and achieving a balance between information utilization and accuracy.

[0119] According to some embodiments, to improve inference accuracy and efficiency, a funnel-like design can be employed: high-level interconnected location inference is first performed, and based on this inference result, low-level unlabeled samples are generated, and their probabilities are then inferred. In each step of interconnected location inference, the algorithm principles and model structure are consistent; only the granularity of the interconnected locations represented by the samples varies.

[0120] For example, when the interconnection locations include a country location C1 and a city location C2, a first subset of pending interconnection information at the country location level in the pending interconnection information set can be determined; based on the first interconnection information set, a first interconnection probability of each piece of pending interconnection information in the first subset of pending interconnection information can be determined; based on the first interconnection probability, a second subset of pending interconnection information at the city location level in the pending interconnection information set can be determined; based on the first interconnection information set, a second interconnection probability of each piece of pending interconnection information in the second subset of pending interconnection information can be determined.

[0121] S204, performing empirical rule-based interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain a fourth interconnection information set;

[0122] According to some embodiments, the fourth interconnection information in the fourth interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type.

[0123] In some embodiments, as Figure 2 As shown in FIG, the empirical rules used in inferring interconnection facilities include but are not limited to the empirical rules related to sea connection, the empirical rules related to direct connection, and the empirical rules related to IXP.

[0124] It should be noted that in addition to the AS interconnections located by the path detection results in the multi-data source information, the connections under investigation also include some inferred AS interconnection location results. That is to say, it is necessary to infer the interconnection facilities based on empirical rules for the multi-data source information, the first interconnection information set, and the second interconnection information set to improve the accuracy of obtaining the fourth interconnection information set.

[0125] S205, constructing an interconnection facility inference model based on the fourth interconnection information set and autonomous system data characteristics;

[0126] According to some embodiments, the interconnected facility inference model may be, for example, a random forest model. Random forest is an ensemble learning method that achieves high classification accuracy and robustness by constructing multiple decision trees and integrating their results. For example, the random forest model may be a three-class random forest model.

[0127] In some embodiments, a second number of samplings with replacement can be performed on the fourth interconnected information set to obtain a second number of sub-training sets; for each sub-training set, recursive splitting is performed starting from the root node, and at each node, a third number of candidate features are randomly selected from the autonomous system data features corresponding to the sub-training set, and the optimal candidate feature and splitting threshold are selected according to the evaluation index until the stopping condition is met, thereby obtaining a random forest model composed of a second number of sub-decision trees.

[0128] Taking a scenario as an example, the training set The number of decision trees (i.e., the second quantity) T, the feature subset size m, and the node splitting evaluation index (such as the Gini index and information gain) are used as input to construct a random forest model according to the following steps:

[0129] S2051, sampling: Perform T times of sampling with replacement to generate a sub-training set

[0130] Among them, the training set It can be the first interconnected information set or a sample set reconstructed based on information from multiple data sources. Each sample is represented in the form of (AS1, AS2, C1, C2), and its corresponding facility type F can be obtained through the random forest model.

[0131] S2052, Decision Tree Training: For each Recursively split from the root node until the stopping condition (such as node purity threshold, maximum depth) is met; at each node, randomly select m candidate features from all data features; select the optimal candidate feature and split threshold based on the evaluation index;

[0132] S2053, voting mechanism: for new sample x new , T decision trees in the random forest {h1,h2,...,h T The sample is classified and the classification result is obtained by ensembling using the majority voting method:

[0133]

[0134] in, Represents the classification result; k represents the predicted category; the above formula indicates that it is necessary to find the k value that maximizes the sum expression on the right side of the equal sign.

[0135] S206: Using an interconnection facility inference model to perform interconnection facility inference based on the fourth interconnection information set, to obtain an Internet inter-domain interconnection view inference result.

[0136] According to some embodiments, the fourth interconnection information set may be input into a random forest model for classification to obtain an Internet inter-domain interconnection view inference result.

[0137] According to some embodiments, this embodiment uses data from 2023 as an example. In the basic data constructed by combining information from multiple data sources, AS interconnection location data covers 50,478 ASes and 266,165 pairs of AS interconnections, while AS interconnection facilities cover 26,018 ASes and 151,496 pairs of AS interconnections. In the inference results obtained during the model inference phase, AS interconnection location data covers 21,011 ASes and 109,545 pairs of AS interconnections, while AS interconnection facilities cover 38,925 ASes and 229,392 pairs of AS interconnections. Overall, the final results of this embodiment cover 51,990 ASes and 340,502 pairs of AS interconnections. Compared to similar work, the 2016 interconnection location dataset provided by CAIDA covers 13,174 ASes and 310,738 pairs of AS interconnections. The method provided in this embodiment improves AS and AS interconnection coverage by 295% and 9.58%, respectively.

[0138] In addition, a shortest distance-first AS interconnection location inference algorithm was implemented as a baseline. Experimental results show that the baseline algorithm achieves an accuracy rate of 37.98% and a recall rate of 55.85% for AS interconnection location inference. This method, however, achieves an accuracy rate of 73.23% and a recall rate of 71.54%, exceeding the baseline algorithm by 92.8% and 28.1%, respectively. Furthermore, this method achieves an accuracy rate of 98% and a recall rate of 98% for AS interconnection facility inference, demonstrating the superiority of the method provided by this embodiment.

[0139] In summary, the method provided in this embodiment overcomes the challenges of single-dimensional data easily leading to biased results and multi-dimensional data being difficult to integrate and utilize. Based on multi-source data, through the comprehensive application of various measurement techniques and analytical methods, combined with empirical rules and machine learning methods, it can comprehensively utilize multi-dimensional data in terms of sample construction and feature design. Secondly, to overcome the challenge of limited existing observation coverage, by dividing inference into two stages, namely interconnected location inference and interconnected facility inference, the observable data is utilized with different emphasis at different stages, thus maximizing data utilization.

[0140] In order to implement the above embodiments, the present disclosure also proposes an Internet inter-domain interconnection view inference system based on multi-source data.

[0141] For example, Figure 3 This is a schematic diagram of the structure of an Internet domain interconnection view inference system based on multi-source data provided by an embodiment of the present disclosure. Figure 3 As shown, the Internet inter-domain interconnection view inference system 300 based on multi-source data includes:

[0142] An information determining unit 310 is configured to obtain multiple data source information provided by multiple different types of data sources, and determine a first interconnection information set based on the multiple data source information, wherein the first interconnection information in the first interconnection information set includes an autonomous system identifier pair and an interconnection location;

[0143] a location inference unit 320 configured to perform interconnection location inference based on the first interconnection information set to obtain a second interconnection information set, wherein the second interconnection information in the second interconnection information set includes the autonomous system identifier pair and the interconnection location obtained after performing the interconnection location inference;

[0144] The facility inference unit 330 is configured to perform interconnection facility inference based on the multi-data source information, the first interconnection information set, and the second interconnection information set to obtain an inference result of an inter-domain Internet interconnection view, wherein the inference result of the inter-domain Internet interconnection view includes a third interconnection information set, and the third interconnection information in the third interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type.

[0145] Optionally, the facility inference unit 330 is configured to perform interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set, and when obtaining an Internet inter-domain interconnection view inference result, specifically to:

[0146] performing an empirical rule-based interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain a fourth interconnection information set, wherein the fourth interconnection information in the fourth interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type;

[0147] The interconnection facility inference model is adopted to perform interconnection facility inference according to the fourth interconnection information set, and an inference result of the Internet inter-domain interconnection view is obtained.

[0148] Optionally, the multi-source data-based Internet inter-domain interconnection view inference system 300 further includes a model building unit for:

[0149] Extracting autonomous system data features from multiple data source information;

[0150] constructing an interconnection location inference model based on the first interconnection information set and the autonomous system data characteristics, wherein the interconnection location inference model is used to perform interconnection location inference based on the first interconnection information set to obtain a second interconnection information set;

[0151] An interconnection facility inference model is constructed based on the fourth interconnection information set and autonomous system data characteristics.

[0152] Optionally, the location inference unit 320 is configured to perform interconnection location inference based on the first interconnection information set, and when obtaining the second interconnection information set, specifically to:

[0153] Determine the distribution location set of each autonomous system based on information from multiple data sources;

[0154] Combining the distribution location sets between different autonomous systems to obtain a pending interconnection information set;

[0155] Determining, based on the first interconnection information set, an interconnection probability of each pending interconnection information in the pending interconnection information set;

[0156] The pending interconnection information whose interconnection probability meets the interconnection probability requirement is determined from the pending interconnection information set to obtain a second interconnection information set.

[0157] Optionally, when the location inference unit 320 is configured to determine the interconnection probability of each pending interconnection information in the pending interconnection information set based on the first interconnection information set, it is specifically configured to:

[0158] Performing a first number of samplings with replacement on the first interconnected information set to obtain a first number of positive sub-sample sets;

[0159] Performing a first number of samplings with replacement on the pending interconnection information set to obtain a first number of temporary negative sample sets, and determining a pending inference set corresponding to each temporary negative sample set, wherein the pending interconnection information in the pending inference set is the pending interconnection information in the pending interconnection information set excluding the temporary negative sample set;

[0160] Selecting a sub-positive sample set and a temporary negative sample set in turn, combining them, and training the initial binary classifier to obtain a first number of trained binary classifiers;

[0161] The trained binary classifier is used to perform probability inference on each pending interconnection information in the pending inference set corresponding to the trained binary classifier, and the interconnection probability of each pending interconnection information in the pending interconnection information set is obtained.

[0162] Optionally, when the location inference unit 320 is configured to perform a first number of samplings with replacement on the first interconnected information set, it is specifically configured to:

[0163] grouping the first interconnected information set according to the first confidence level of each piece of first interconnected information in the first interconnected information set to obtain a plurality of first interconnected information subsets;

[0164] Determining, according to the first confidence level, a sampling ratio of each first interconnected information subset when performing sampling with replacement on the first interconnected information set;

[0165] A first number of samplings with replacement are performed on the first interconnected information set according to a sampling ratio.

[0166] Optionally, the interconnection location includes a country location and a city location. When the location inference unit 320 is configured to determine the interconnection probability of each pending interconnection information in the pending interconnection information set based on the first interconnection information set, it is specifically configured to:

[0167] Determining a first subset of pending interconnection information at a national location level in the pending interconnection information set;

[0168] Determining, based on the first interconnection information set, a first interconnection probability for each piece of pending interconnection information in the first subset of pending interconnection information;

[0169] Determining, based on the first interconnection probability, a second subset of pending interconnected information at a city location level in the pending interconnected information set;

[0170] A second interconnection probability of each piece of pending interconnection information in a second subset of pending interconnection information is determined according to the first interconnection information set.

[0171] Optionally, the interconnected facility inference model is a random forest model, and the model building unit is configured to build the interconnected facility inference model according to the fourth interconnected information set and the autonomous system data characteristics, specifically for:

[0172] Performing a second number of sampling with replacement on the fourth interconnected information set to obtain a second number of sub-training sets;

[0173] For each sub-training set, recursive splitting is started from the root node, and at each node, a third number of candidate features are randomly selected from the autonomous system data features corresponding to the sub-training set. The optimal candidate features and splitting threshold are selected according to the evaluation index until the stopping condition is met, and a random forest model consisting of the second number of sub-decision trees is obtained.

[0174] Optionally, the autonomous system data feature includes at least one of the following:

[0175] Autonomous system attribute characteristics, wherein the autonomous system attribute characteristics include at least one of autonomous system customer cone, number of IP prefixes, and number of IP addresses;

[0176] Autonomous system interconnection characteristics, wherein the autonomous system interconnection characteristics include at least one of autonomous system business relationships and the number of autonomous system neighbors;

[0177] Geographical and facility characteristics, wherein the geographical and facility characteristics include at least one of geographical distance and the number of submarine cable landing points.

[0178] It should be noted that the aforementioned explanation of the embodiment of the Internet inter-domain interconnection view inference method based on multi-source data is also applicable to the Internet inter-domain interconnection view inference system based on multi-source data in this embodiment, and will not be repeated here.

[0179] In summary, the system provided by the embodiments of the present disclosure can build a high-coverage, easily updated view of the logical and physical associations of the Internet by making full use of multi-source data to infer and associate the interconnected locations and facility types between autonomous systems.

[0180] In order to implement the above embodiments, the present disclosure also proposes an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

[0181] In order to implement the above embodiments, the present disclosure further proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0182] In order to implement the above embodiments, the present disclosure further provides a computer program product, including a computer program, which implements the methods provided in the above embodiments when executed by a processor.

[0183] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this disclosure are in compliance with relevant laws and regulations and do not violate public order and good morals.

[0184] It is important to note that personal information collected from users should be used for legitimate and reasonable purposes and should not be shared or sold beyond these legitimate uses. Furthermore, such collection / sharing should be conducted only after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign an agreement / authorization that includes the relevant user information before using the feature. Furthermore, any necessary steps must be taken to safeguard and secure access to such personal information and ensure that others with access to personal information comply with its privacy policy and procedures.

[0185] This disclosure contemplates providing implementations that allow users to selectively block the use or access of personal information data. Specifically, this disclosure contemplates providing hardware and / or software to prevent or block access to such personal information data. Risks can be minimized by limiting data collection and deleting data once it is no longer needed. Furthermore, where applicable, such personal information can be de-identified to protect user privacy.

[0186] In the descriptions of the aforementioned embodiments, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are mutually inconsistent.

[0187] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. Throughout the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0188] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present disclosure belong.

[0189] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM) or flash memory, a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0190] It should be understood that various parts of the present disclosure can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement the present invention: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0191] Those skilled in the art will appreciate that all or part of the steps in the method for implementing the above-mentioned embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0192] In addition, the functional units in the various embodiments of the present disclosure may be integrated into a single processing module, each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0193] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present disclosure have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. A person of ordinary skill in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present disclosure.

Claims

1. A method for inferring an Internet inter-domain interconnection view based on multi-source data, characterized in that: include: Acquire multiple data source information provided by multiple different types of data sources, and determine a first interconnection information set based on the multiple data source information, wherein the first interconnection information in the first interconnection information set includes an autonomous system identifier pair and an interconnection location; performing interconnection location inference based on the first interconnection information set to obtain a second interconnection information set, wherein the second interconnection information in the second interconnection information set includes the autonomous system identifier pair and the interconnection location obtained after performing the interconnection location inference; Interconnection facilities are inferred based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain an Internet inter-domain interconnection view inference result, wherein the Internet inter-domain interconnection view inference result includes a third interconnection information set, and the third interconnection information in the third interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type.

2. The method according to claim 1, characterized in that The performing interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain an Internet inter-domain interconnection view inference result includes: performing an empirical rule-based interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain a fourth interconnection information set, wherein fourth interconnection information in the fourth interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type; An interconnection facility inference model is used to perform interconnection facility inference based on the fourth interconnection information set to obtain an Internet inter-domain interconnection view inference result.

3. The method according to claim 2, characterized in that The method further comprises: extracting autonomous system data features from the multiple data source information; constructing an interconnection location inference model based on the first interconnection information set and the autonomous system data characteristics, wherein the interconnection location inference model is used to perform interconnection location inference based on the first interconnection information set to obtain a second interconnection information set; The interconnection facility inference model is constructed according to the fourth interconnection information set and the autonomous system data characteristics.

4. The method according to claim 1, wherein The performing interconnection location inference based on the first interconnection information set to obtain the second interconnection information set includes: Determining a distribution location set of each autonomous system based on the multiple data source information; Combining the distribution location sets between different autonomous systems to obtain a pending interconnection information set; determining, based on the first interconnection information set, an interconnection probability of each pending interconnection information in the pending interconnection information set; The pending interconnection information whose interconnection probability meets the interconnection probability requirement is determined from the pending interconnection information set to obtain a second interconnection information set.

5. The method according to claim 4, characterized in that The step of determining, based on the first interconnection information set, the interconnection probability of each pending interconnection information in the pending interconnection information set includes: Performing a first number of samplings with replacement on the first interconnected information set to obtain a first number of positive sub-sample sets; Performing the first number of samplings with replacement on the pending interconnection information set to obtain a first number of temporary negative sample sets, and determining a set to be inferred corresponding to each of the temporary negative sample sets, wherein the pending interconnection information in the set to be inferred is the pending interconnection information in the pending interconnection information set excluding the temporary negative sample set; Selecting a sub-positive sample set and a temporary negative sample set in turn, combining them, and training the initial binary classifier to obtain a first number of trained binary classifiers; The trained binary classifier is used to perform probability inference on each pending interconnection information in the pending inference set corresponding to the trained binary classifier, so as to obtain the interconnection probability of each pending interconnection information in the pending interconnection information set.

6. The method according to claim 5, characterized in that The performing a first number of sampling with replacement on the first interconnected information set includes: grouping the first interconnected information set according to the first confidence level of each piece of first interconnected information in the first interconnected information set to obtain a plurality of first interconnected information subsets; determining, according to the first confidence level, a sampling ratio of each first interconnected information subset when performing sampling with replacement on the first interconnected information set; A first number of samplings with replacement are performed on the first interconnected information set according to the sampling ratio.

7. The method according to claim 4, characterized in that The interconnection location includes a country location and a city location. Determining, based on the first interconnection information set, the interconnection probability of each pending interconnection information in the pending interconnection information set includes: Determining a first subset of pending interconnection information at a country location level in the pending interconnection information set; determining, based on the first interconnection information set, a first interconnection probability for each piece of pending interconnection information in the first subset of pending interconnection information; Determining, based on the first interconnection probability, a second subset of pending interconnection information at a city location level in the pending interconnection information set; A second interconnection probability of each piece of pending interconnection information in the second pending interconnection information subset is determined according to the first interconnection information set.

8. The method according to claim 3, characterized in that The interconnection facility inference model is a random forest model, and constructing the interconnection facility inference model according to the fourth interconnection information set and the autonomous system data characteristics includes: Performing a second number of sampling with replacement on the fourth interconnected information set to obtain a second number of sub-training sets; For each sub-training set, recursive splitting is started from the root node, and at each node, a third number of candidate features are randomly selected from the autonomous system data features corresponding to the sub-training set, and the optimal candidate features and splitting threshold are selected according to the evaluation index until the stopping condition is met, thereby obtaining a random forest model consisting of a second number of sub-decision trees.

9. The method according to claim 3, characterized in that The autonomous system data characteristics include at least one of the following: Autonomous system attribute characteristics, wherein the autonomous system attribute characteristics include at least one of autonomous system customer cone, number of IP prefixes, and number of IP addresses; Autonomous system interconnection characteristics, wherein the autonomous system interconnection characteristics include at least one of autonomous system business relationships and the number of autonomous system neighbors; Geographical and facility characteristics, wherein the geographical and facility characteristics include at least one of geographical distance and the number of submarine cable landing points.

10. An Internet inter-domain interconnection view inference system based on multi-source data, characterized in that: include: an information determining unit, configured to obtain multi-data source information provided by multiple different types of data sources, and determine a first interconnection information set based on the multi-data source information, wherein the first interconnection information in the first interconnection information set includes an autonomous system identifier pair and an interconnection location; a location inference unit, configured to perform interconnection location inference based on the first interconnection information set to obtain a second interconnection information set, wherein the second interconnection information in the second interconnection information set includes the autonomous system identifier pair and the interconnection location obtained after performing the interconnection location inference; A facility inference unit is configured to perform interconnection facility inference based on the multiple data source information, the first interconnection information set, and the second interconnection information set to obtain an Internet inter-domain interconnection view inference result, wherein the Internet inter-domain interconnection view inference result includes a third interconnection information set, and the third interconnection information in the third interconnection information set includes an autonomous system identifier pair, an interconnection location, and an interconnection device type.

Images