CN115687946A - Homologous transaction identification method based on Ethernet workshop node time information - Google Patents
Homologous transaction identification method based on Ethernet workshop node time information Download PDFInfo
- Publication number
- CN115687946A CN115687946A CN202211196222.6A CN202211196222A CN115687946A CN 115687946 A CN115687946 A CN 115687946A CN 202211196222 A CN202211196222 A CN 202211196222A CN 115687946 A CN115687946 A CN 115687946A
- Authority
- CN
- China
- Prior art keywords
- transaction
- homologous
- node
- time
- transactions
- 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
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to a homologous transaction identification method based on Ethernet room node time information, and belongs to the technical field of block chains. The method aims to utilize time information leaked by a fixed node in an Ethernet workshop network to perform characteristic expression on network attributes of transactions, and realize correlation analysis of mass transactions from a network layer. The method uses the network attribute characteristics of the transaction and takes the node as an entity to carry out transaction association. By detecting the time sequence of the broadcast of transaction flooding from a plurality of fixed nodes, the broadcast process of each transaction is uniformly characterized, and the transactions with similar broadcast processes, namely the transactions entering the network from the same source node, are clustered into an entity. The method can cover all transactions in the ether house and can resist the influence of the mixed currency contract.
Description
Technical Field
The invention relates to a homologous transaction identification method based on Ethernet room node time information, and belongs to the technical field of block chains.
Background
Since the introduction of the block chain technology, various encryption currencies relying on the block chain technology have been increasingly developed. Among them, etherhouses have received much attention from people as the first block chain platform supporting intelligent contracts. In a cryptocurrency system represented by an ether house, a user can freely apply for a pseudonym unrelated to an actual identity and perform a transaction through the pseudonym. These pseudonyms are controlled by a private key held by the user. Although all transaction records related to the pseudonym, including transaction amount, transaction time, account address, etc., are disclosed in the block for achieving decentralized consensus, and are viewed by all, no one knows the true identity represented by the pseudonym, which protects the privacy of the user.
Due to the characteristics of huge market value and anonymous identity, various malicious behaviors frequently occur in the ether house. These malicious activities directly compromise the economic interests of the user, undermining the security of the blockchain ecology. Accordingly, some research structures and companies are beginning to analyze data published on blockchains, analyze association analysis between different pseudonyms, determine different pseudonyms for the same user, to evaluate cryptocurrency assets, plan for money movement, and support law enforcement.
The existing transaction association method for the ether house can be divided into two types: based on user habits and based on specific scenarios. The user habit is based on the characteristics that a researcher reflects the user behavior habit by analyzing the characteristics of transaction time, transaction rate, transaction amount and the like of an account by using the uniqueness of a user behind a pseudonym, a heuristic rule is provided, different accounts of the same user are clustered into an entity, and then the whole network transaction is subjected to statistical analysis. Based on a specific scene, the method refers to that a researcher uses special rules under certain special scenes (such as deposit address reuse, multiple participation vacancy, self-authorization and the like) in an Etherhouse, and provides a heuristic clustering method according to the rules to associate initiators of different transactions.
However, these correlation methods all have to know the account addresses of the initiator and the acceptor of each transaction, which is affected by the mixed currency protocol. These methods fail once the user has masked the connection between the initiator and the acceptor using a coin-mix contract. Meanwhile, because the user behavior habits and the rules under special scenes are utilized, the association methods can only associate a small number of transactions meeting the conditions, but cannot play a role in most of the rest transactions.
In summary, the existing methods for performing transaction association in an ethernet workshop have the disadvantages of small coverage, poor robustness and the like.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and creatively provides a homologous transaction identification method based on the time information of Ethernet workshop nodes.
The method aims to utilize time information leaked by a fixed node in the Ethernet network to perform characteristic expression on network attributes of transactions, and realize correlation analysis on mass transactions from a network layer. The method can cover all transactions in the ether house and can resist the influence of the mixed currency contract.
The method has the innovation points that: with the features of all transactions: and the network attribute takes the node as an entity to carry out transaction association. By detecting the time sequence of the transaction flooding broadcast from a plurality of fixed nodes, the broadcast process of each transaction is uniformly characterized, and the transactions with similar broadcast processes, namely the transactions entering the network from the same source node, are clustered into one entity.
The technical scheme adopted by the invention is as follows.
A homologous transaction identification method based on Ethernet room node time information comprises the following steps:
step 1: the network is monitored and transaction network information is collected.
And (4) carrying out transaction association, namely deploying a client in the Ethernet shop network, collecting data of a network layer of the transaction, and storing the data for subsequent analysis.
Specifically, step 1 may include the steps of:
step 1.1: and modifying the source code of the Etherhouse client (such as a geth client, which is called go-ethereum) and recording the transaction information.
Because the client provided by the Ethengfang is oriented to the user, the method hides the process of interaction between the node and the neighbor in the network and directly presents the final interaction result to the user. Therefore, when the source code of the Etherhouse client is to be modified, when the node receives the transaction relayed by the neighbor node, the log information of the transaction is recorded, including the timestamp of the transaction, the transaction hash, the IP of the relay node and the account of the initiator of the transaction.
Specifically, each transaction relayed by the neighbor node is described by a quadruplet (h, a, p, t), wherein t represents a timestamp of the transaction, h represents a transaction hash, p represents a relay node IP, and a represents an initiator account of the transaction.
Step 1.2: and deploying an Etherhouse client and monitoring the network.
And deploying a client operation fast synchronization mode in the Ethernet main network as a probe node.
In the fast synchronization mode, the probe node receives the block and the transaction relayed by the neighbor node, but does not verify and forward the new transaction and the block.
Step 2: and (6) data processing.
First, a time threshold T and a minimum number of copies M are set.
And dividing the recorded log information into intervals by taking T as a time interval. For data in each interval, log information is extracted (which may be extracted using python scripts), and transaction hashes, timestamps of transactions, initiator accounts of transactions, relay node IPs are saved (which may be saved in JSON format).
For each transaction tx, record tx = { h, a, (p) i ,t i ) I =0,1 \ 8230 }, counting the time for different nodes to relay a transaction to a probe node, where p i For different neighbour nodes IP, t i Is the relative time (i.e., minus the minimum timestamp for each transaction) at which the probe received the neighbor node forwarded copy of the transaction. Meanwhile, the transaction and abnormal data transaction (caused by network interruption and the like) with the number of received copies less than M are eliminated. Finally, record the set of neighbor nodes for all transactions P, P = { P = { (P) i |i＝0,1…}。
And 3, step 3: and constructing homologous trading pairs and non-homologous trading pairs.
Wherein, the homologous transaction pair refers to two transactions entering the network from the same source node. A non-homologous transaction pair refers to two transactions entering the network from different source nodes.
Based on the fact that a user cannot frequently switch access nodes in a short time, the invention provides a heuristic clustering rule for identifying homologous transaction pairs and non-homologous transaction pairs, which specifically comprises the following steps:
rule 1: any two transactions tx 1 、tx 2 If tx 1 .a＝＝tx 2 A and | tx 1 .t-tx 2 .t|<T, then tx 1 And tx 2 Are a pair of homologous transactions.
Rule 2: homologous transaction pairs tx 1 -tx 2 Respectively occur at time t 1 、t 2 (ii) a Homologous transaction pairs tx 3 -tx 4 Respectively occur at time t 3 、t 4 . If t is 1 、t 2 And t 3 、t 4 There is a time intersection, then tx 3 And tx 1 Non-homologous transaction pairs, tx 2 And tx 4 Are non-homologous transaction pairs.
And (3) aiming at the transaction obtained in the step (2), constructing a homologous transaction pair according to a rule 1, and constructing a non-homologous transaction pair according to a rule 2.
And 4, step 4: and unifying the reference nodes of different transaction time stamps to obtain a unified expression.
Specifically, P in IP set P is selected 0 And 3, serving as a reference node, and relatively converting the homologous transaction pair and the non-homologous transaction pair obtained in the step 3. Time stamp t of each transaction i Subtracting p from 0 Time stamp t corresponding to node 0 ，t i ＝t i -t 0 。
The time series in the transaction are then reordered according to the IP order in P, building a relative time series with the same meaning (relative to the same node) and in the same order for each comparison.
And 5: analyzing the broadcasting process of the homologous transaction pairs, performing feature representation on the similarity of the homologous transaction pairs, and unifying the feature representation of the homologous transaction pairs and the non-homologous transaction pairs.
Specifically, step 5 includes the steps of:
step 5.1: the similarity of the transaction pairs is represented.
Since the network topology is constant over a short period of time. Two flood broadcasts from the same source node will have a certain probability of reaching the designated node with the same path. If two nodes arrive with the same path, the time difference between the two nodes in the two broadcasts is consistent.
Therefore, the absolute time difference and the time difference trend change of the two transactions in the transaction pair are respectively counted, and the similarity of the two broadcasting processes of the homologous transaction pair is reflected.
Calculating absolute time difference box of transaction pair by a classification method in feature engineering t The calculation is as follows:
wherein the content of the first and second substances,
representing two transactions tx in a transaction pair separately 1 、tx 2 By node p i Relative timestamp of relay time; i is a set time interval and represents tolerance to time fluctuation; n represents the number of relative time stamps in the received time series.
The time-varying trend between different nodes is another key attribute. Calculating the variation trend of any two nodes by adopting a quotient obtaining method t The method comprises the following steps:
wherein the content of the first and second substances,
respectively representing two transactions tx in a transaction pair 1 、tx 2 By node p i 、p j Relative timestamp of relay time. And item 2 is the reciprocal of item 1 in order to avoid the effect of denominator selection on the results.
Step 5.2: and (5) feature extraction.
And respectively extracting features of the absolute time difference sequence and the change trend sequence of the transaction pair, and extracting features for describing the similarity of the homologous transaction pair for training and learning the model.
Step 6: and (5) carrying out model training and cross validation on the original data set obtained in the step (4) and the feature data set obtained in the step (5), thereby realizing the identification of homologous transaction pairs.
Advantageous effects
Compared with the prior art, the method of the invention has the following advantages:
(1) In the method, the probe nodes are deployed to be connected to the selected fixed nodes, and the transaction pairs entering the network from the same source node are clustered. The invention covers most of the transactions of the whole network and can resist the influence of the mixed currency contract. In addition, the probe nodes deployed by the invention only need to show complete and normal behaviors and are hidden and difficult to monitor.
(2) By analyzing the behavior pattern of the user, the invention provides a heuristic construction method of homologous transaction pairs and non-homologous transaction pairs, and homologous transaction pairs and non-homologous transaction pairs are constructed by utilizing network layer information of transactions and serve as data sets for subsequent analysis.
(3) The invention designs a new characteristic representation mode to describe the similarity of the homologous transaction in the network to the network attribute.
Drawings
FIG. 1 is a flow chart of transaction association of the method of the present invention.
Detailed Description
The method of the present invention will be described in further detail with reference to the accompanying drawings and examples. It should be noted that the practice of the present invention is not limited to the following examples, and any modification or variation of the present invention may be made without departing from the scope of the present invention.
Examples
A homologous transaction identification method based on Ethernet room node time information is disclosed.
FIG. 1 is a flow chart of an Etherhouse transaction correlation method with fixed node time leakage. Depending on the flow chart in fig. 1, the method specifically comprises the following steps:
step 1: downloading the geth client source code from the Etherhouse official website, modifying the client code and recording the received transaction copy. The maximum number of links of the geth client is set to its default value (50), while relaxing the limitation on the number of actively accessing nodes, allowing actively accessing nodes to exceed 1/3 of the maximum number of connections. The final number of access nodes is between 15-35.
Step 2: considering the routing state and the packet propagation speed of the ether house P2P network, a time threshold T =10min and a minimum copy number M =15 are set.
And (4) counting the relay nodes and the relay time stamps of each transaction and calculating a union set P of the relay nodes.
And step 3: and carrying out heuristic clustering on the transactions in different intervals respectively according to the provided heuristic clustering rules 1 and 2, and constructing homologous transaction pairs and non-homologous transaction pairs.
And 4, step 4: selecting node P in P 0 . With p 0 The relay timestamps of all other nodes are relativised for the reference node. Note p 0 There should be a node in the list of all transaction relay nodes. The specific relatizing method is to each transaction, the node p i Relay time stamp t of i ＝t i -t 0 ，t 0 Is node p 0 The relay timestamp of (1). Then according to the node sequence in P, the time of all transactions is countedThe stamps are reordered to ensure that the time stamp sequences for all transactions have the same order. And finally, obtaining an original transaction pair data set, wherein sample data in the data set is two time stamp sequences of the transaction pair, and the tag is determined according to the homology of the sample data and the time stamp sequences. If the transaction pair is a homologous transaction pair, the label is 1; if it is a non-homologous transaction pair, its label is 0. There were 5704 samples in the dataset, with the homologous and non-homologous transaction pairs in half each.
And 5: similarity representation is carried out on the data sets of the transaction according to the formulas 1 and 2, wherein I =50. The two time stamp sequences of each transaction pair will be merged into a vector representing the degree of similarity, comprising a sequence of absolute time differences and a sequence of trends in change. And performing feature extraction on the absolute time difference sequence and the variation trend sequence in each vector representing the similarity degree. The common features of the two include the mean, standard deviation, minimum, first quartile, median, third quartile, maximum, range, kurtosis, skewness, absolute mean deviation (AAD), median Absolute Deviation (MAD), quartile range (IQR), 13-D total. Further, the extraction of the proportional features of the key values for the sequence of absolute time differences comprises greater than 0, less than 0, equal to 0, greater than b, less than-b, greater than 2 xb, less than-2 xb, greater than 3 xb, less than-3 xb, greater than 4 xb, less than-4 xb, where b is related to the value of I, where b =5. For the variation trend sequence, the extracted proportional features of the key numerical values comprise more than 0, less than 0, positive number sum, negative number sum, more than 3, less than-3, more than 4, less than-4, more than and less than-5. Finally, the two time stamp sequences of each pair of transaction pairs are converted into a 48-dimensional feature vector, and feature data sets are generated from the original transaction pair data sets.
Step 6: and selecting a classifier to perform feature learning and identifying homologous transaction pairs. The traditional machine learning methods (SVM, decisionTree, randomForest and XGboost) and the multilayer perceptron are verified, and the most appropriate classifier model is selected.
And performing Grid _ Search on the traditional machine learning model, selecting the result of the optimal parameter, and comparing the results. The optimal parameter of the SVM is max _ depth =6, max _feature =0.3, min _samples _split =4. The optimal parameters of DecisionTree are max _ depth =6, max _lives =0.3, min _samples _ split =4. The optimum parameters for RandomForest are n _ estimators =15, max _lives =0.5, max _depth =8, min _samples _split =12. The optimal parameters of XGBoost are learning _ rate =0.3222, n \\ estimators =500, max_depth =10, subsample =0.8, and colsample_byte =0.9. The multilayer perception achieves the optimal effect when a 3-layer neural network is used, the RELU is used as an activation function, and the FocalLoss is used as a loss function.
The result shows that the multi-layer perceptron is the most suitable classifier model, and can identify the homologous transaction pair with an F1 score of 97%, namely whether two transactions enter the network from the same source node. The cognate transaction pair recognition effect of different classifiers is shown in table 1:
TABLE 1 homologous transaction pair recognition Effect of different classifiers
| Classifier | Precision | Recall | F1 |
| SVM | 0.9149 | 0.8597 | 0.886441485 |
| DT | 0.8787 | 0.8588 | 0.86863604 |
| RF | 0.9639 | 0.9474 | 0.955578779 |
| XGBoost | 0.9712 | 0.9658 | 0.968492473 |
| Neural Network | 0.9747 | 0.97969 | 0.97718863 |
Contrast verification
In the embodiment, the homologous transaction identification method based on the time information of the EtherFang nodes is compared with other transaction association methods, and the method has better homologous transaction pair identification capability.
The method of comparison (https:// ambient. Lu/bitstream/10993/3972/1/bioryukov-tikhomirov-evaluation-and-linkabiiity. Pdf) describes the network attribute of one transaction by using the relay time stamps of the first N nodes, then assigns weights to the time stamps, and judges whether the two transactions are homologous or not by using the similarity of the weight vectors.
The comparison method uses the concept of adjusted anonymity degree (adjusted anonymity degree) to illustrate the effect of the correlation, and a lower value indicates that the correlation method is more effective.
For better comparison, the same way will be used in this example to measure the effect of the correlation. Adjusted anonymity d adj The calculation method is as follows:
d adJ ＝1-(1-e)(1-d)
where e represents the median squared error between the probability distribution and the true distribution. p is a radical of i Representing the probability that the transaction is homologous to the target transaction. d is the ratio of the calculated information entropy to the maximum entropy. N represents the number of transactions.
To ensure generality, this example will calculate the probability p that two transactions are homologous in the following way:
p＝p 0 pre+(1-p 0 )(1-pre)
wherein p is 0 Represents the mean probability of the output probability distribution of the classifier of the present invention, and pre represents the recognition accuracy of the model.
The result shows that the method adjusts the anonymity degree d adj Adjustment anonymity d for which =0.563 is much higher than the contrast method adj =0.879, the timestamp of the fixed N nodes used in the present invention can better represent the network attribute of the transaction, resulting in better transaction association effect. The effect of the previous N node timestamp is compared with the fixed N node timestamp, as shown in table 2:
TABLE 2 comparison of the previous N-node timestamp and the fixed N-node timestamp effect
| Method | Time stamp | Adaptive network | Number of nodes | d adj |
| Comparison method | First | Bitcoin | 1000 | 0.879 |
| Method for producing a composite material | Fixed | Ethereum | 35 | 0.563 |
While the embodiments of the present invention have been described in connection with the drawings and examples, it will be apparent to those skilled in the art that various modifications can be made without departing from the principles of this patent, and it is intended to cover all modifications that are within the scope of this patent.
Claims (3)
1. A homologous transaction identification method based on Ethernet room node time information is characterized by comprising the following steps:
step 1: monitoring a network and collecting transaction network information;
firstly, deploying a client in an Ethernet shop network, collecting data of a network layer of a transaction, and then storing the data for subsequent analysis;
step 2: processing data;
firstly, setting a time threshold T and a minimum copy number M;
dividing the recorded log information into intervals by taking T as a time interval; for data in each interval, extracting log information, and storing transaction hash, a time stamp of the transaction, an account of an initiator of the transaction and a relay node IP;
for each transaction tx, record tx = { h, a, (p) i ,t i ) I =0,1 \8230; }, h denotes transaction hash, a denotes time when the initiator account of the transaction statistically different nodes relay the transaction to the probe node, p i As different neighbor nodesPoint IP, t i The relative time when the probe receives the transaction copy forwarded by the neighbor node; meanwhile, rejecting the transactions and abnormal data transactions with the number of received copies less than M;
finally, record the set of neighbor nodes for all transactions P, P = { P = { (P) i |i＝0,1…}；
And step 3: constructing homologous transaction pairs and non-homologous transaction pairs;
wherein, the homologous transaction pair refers to two transactions entering the network from the same source node; a non-homologous transaction pair refers to two transactions entering the network from different source nodes;
the heuristic clustering rule for identifying the homologous transaction pairs and the non-homologous transaction pairs is adopted, and specifically comprises the following steps:
rule 1: any two transactions tx 1 、tx 2 If tx 1 .a＝＝tx 2 A and | tx 1 .t-tx 2 .t|<T, then tx 1 And tx 2 Are a homologous transaction pair;
rule 2: homologous transaction pairs tx 1 -tx 2 Respectively occurring at time t 1 、t 2 (ii) a Homologous transaction pairs tx 3 -tx 4 Respectively occur at time t 3 、t 4 (ii) a If t is 1 、t 2 And t 3 、t 4 There is a time intersection, then tx 3 And tx 1 Non-homologous transaction pairs, tx 2 And tx 4 A non-homologous transaction pair;
aiming at the transaction obtained in the step 2, constructing a homologous transaction pair according to a rule 1, and constructing a non-homologous transaction pair according to a rule 2;
and 4, step 4: unifying reference nodes of different transaction timestamps to obtain unified expression;
selecting P in IP set P 0 As a reference node, the homologous transaction pair and the non-homologous transaction pair obtained in the step 3 are subjected to relativity; time stamp t of each transaction i Subtracting p from 0 Time stamp t corresponding to node 0 ，t i ＝t i -t 0 ；
Then, reordering the time sequences in the transaction according to the IP sequence in the P, and constructing a relative time sequence with the same meaning and the same sequence for each comparison;
and 5: analyzing the broadcasting process of the homologous transaction pairs, performing characteristic representation on the similarity of the homologous transaction pairs, and unifying the characteristic representation of the homologous transaction pairs and the non-homologous transaction pairs;
and 6: and (5) performing model training and cross validation on the original data set obtained in the step (4) and the feature data set obtained in the step (5), so as to realize identification of the homologous transaction pair.
2. The method for identifying the same source transaction based on the time information of the EtherFang nodes as claimed in claim 1, wherein step 1 comprises the steps of:
step 1.1: modifying the source code of the Ethernet house client and recording transaction information;
modifying the source code of the Ethernet house client, and recording the log information of the transaction when the node receives the transaction relayed by the neighbor node, wherein the log information comprises a timestamp of the transaction, transaction hash, a relay node IP and an initiator account of the transaction;
describing each transaction relayed by the neighbor node by adopting a quadruple (h, a, p, t), wherein t represents a timestamp of the transaction, h represents a transaction hash, p represents a relay node IP, and a represents an initiator account of the transaction;
step 1.2: deploying an Etheng client and monitoring a network;
deploying a client operation fast synchronization mode in an Ethernet main network as a probe node;
in the fast synchronization mode, the probe node receives the block and the transaction relayed by the neighbor node, but does not verify and forward the new transaction and the block.
3. The method for identifying a same source transaction based on the time information of the Etherhouse nodes as claimed in claim 1, wherein the step 5 comprises the steps of:
step 5.1: a similarity representation of the transaction pairs;
respectively counting absolute time difference and time difference trend changes of two transactions in the transaction pair, and reflecting the similarity of two broadcasting processes of the homologous transaction pair;
calculating the absolute time difference box of the transaction pair by a box dividing method in the characteristic engineering t The calculation is as follows:
wherein the content of the first and second substances,
respectively representing two transactions tx in a transaction pair 1 、tx 2 By node p i Relative timestamp of relay time; i is a set time interval and represents tolerance to time fluctuation; n represents the number of relative timestamps in the received time sequence;
the time variation trend between different nodes is another key attribute; calculating the variation trend of any two nodes by adopting a quotient obtaining method t The method comprises the following steps:
wherein the content of the first and second substances,
respectively representing two transactions tx in a transaction pair 1 、tx 2 By node p i 、p j Relative timestamp during relaying, and item 2 is the reciprocal of item 1, so as to avoid the influence of denominator selection on the result;
step 5.2: extracting characteristics;
and respectively extracting features of the absolute time difference sequence and the variation trend sequence of the transaction pair, and extracting features for describing the similarity of the homologous transaction pair for training and learning the model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211196222.6A CN115687946A (en) | 2022-09-29 | 2022-09-29 | Homologous transaction identification method based on Ethernet workshop node time information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211196222.6A CN115687946A (en) | 2022-09-29 | 2022-09-29 | Homologous transaction identification method based on Ethernet workshop node time information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115687946A true CN115687946A (en) | 2023-02-03 |
Family
ID=85064673
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211196222.6A Pending CN115687946A (en) | 2022-09-29 | 2022-09-29 | Homologous transaction identification method based on Ethernet workshop node time information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115687946A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116389163A (en) * | 2023-05-26 | 2023-07-04 | 中科链安(北京)科技有限公司 | Block chain transaction originating node IP tracking method, risk monitoring method and risk monitoring device |
-
2022
- 2022-09-29 CN CN202211196222.6A patent/CN115687946A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116389163A (en) * | 2023-05-26 | 2023-07-04 | 中科链安(北京)科技有限公司 | Block chain transaction originating node IP tracking method, risk monitoring method and risk monitoring device |
| CN116389163B (en) * | 2023-05-26 | 2023-08-01 | 中科链安(北京)科技有限公司 | Block chain transaction originating node IP tracking method, risk monitoring method and risk monitoring device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110334548B (en) | Data anomaly detection method based on differential privacy | |
| Boshmaf et al. | Integro: Leveraging victim prediction for robust fake account detection in OSNs. | |
| Alvarez et al. | Strategic voting in British elections | |
| CN111614690B (en) | Abnormal behavior detection method and device | |
| Zhao et al. | A machine learning based trust evaluation framework for online social networks | |
| Giatsoglou et al. | Retweeting activity on twitter: Signs of deception | |
| Short et al. | Using blockchain technologies to improve security in federated learning systems | |
| Rintyarna et al. | Mapping acceptance of Indonesian organic food consumption under Covid-19 pandemic using Sentiment Analysis of Twitter dataset | |
| CN109831545B (en) | Domain name abuse processing method and system based on block chain | |
| CN110598129B (en) | Cross-social network user identity recognition method based on two-stage information entropy | |
| Yamak et al. | Detection of multiple identity manipulation in collaborative projects | |
| CN115687946A (en) | Homologous transaction identification method based on Ethernet workshop node time information | |
| CN110445772B (en) | Internet host scanning method and system based on host relationship | |
| Lu et al. | Sharing models or coresets: A study based on membership inference attack | |
| Wang et al. | Trustworthy and efficient crowdsensed data trading on sharding blockchain | |
| Han et al. | Data valuation for vertical federated learning: An information-theoretic approach | |
| Jones et al. | Behind the mask: A computational study of Anonymous' presence on Twitter | |
| CN115022326A (en) | Block chain Byzantine fault-tolerant consensus method based on collaborative filtering recommendation | |
| Yu et al. | Robust clustering of ethereum transactions using time leakage from fixed nodes | |
| Feng et al. | An improved X-means and isolation forest based methodology for network traffic anomaly detection | |
| Asadian et al. | Identification of Sybil attacks on social networks using a framework based on user interactions | |
| CN111556148B (en) | Data sharing method, data sharing platform and device | |
| Cai et al. | An unsupervised approach for detecting group shilling attacks in recommender systems based on topological potential and group behaviour features | |
| Buccafurri et al. | Experimenting with certified reputation in a competitive multi-agent scenario | |
| CN116186629A (en) | Financial customer classification and prediction method and device based on personalized federal learning |
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 |