CN112084373B - Graph embedding-based multi-source heterogeneous network user alignment method - Google Patents
Graph embedding-based multi-source heterogeneous network user alignment method Download PDFInfo
- Publication number
- CN112084373B CN112084373B CN202010778910.8A CN202010778910A CN112084373B CN 112084373 B CN112084373 B CN 112084373B CN 202010778910 A CN202010778910 A CN 202010778910A CN 112084373 B CN112084373 B CN 112084373B
- Authority
- CN
- China
- Prior art keywords
- user
- network
- nodes
- users
- node
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000005295 random walk Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000013528 artificial neural network Methods 0.000 claims abstract description 14
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims 1
- 238000004364 calculation method Methods 0.000 abstract description 3
- 238000012360 testing method Methods 0.000 description 4
- 244000046052 Phaseolus vulgaris Species 0.000 description 3
- 235000010627 Phaseolus vulgaris Nutrition 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 240000006677 Vicia faba Species 0.000 description 1
- 235000010749 Vicia faba Nutrition 0.000 description 1
- 235000002098 Vicia faba var. major Nutrition 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008451 emotion Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/901—Indexing; Data structures therefor; Storage structures
- G06F16/9024—Graphs; Linked lists
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/12—Use of codes for handling textual entities
- G06F40/151—Transformation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Economics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a graph embedding-based multi-source heterogeneous network user alignment method, which is characterized by comprising the following steps of: 1) calculating the similarity of user attributes through a user name and a social role; 2) obtaining a node sequence of the heterogeneous network through a random walk algorithm, and analyzing the mutual relation among nodes; 3) calculating the node sequence by using an embedding algorithm to obtain an embedded representation of the network; 4) and training the multilayer neural network to align the users according to the attribute similarity and the structural features of the users. The multi-source heterogeneous network user alignment method based on graph embedding can be used for user alignment of an online social network, has important application in multiple fields of recommendation systems, human and object image completion and the like, is low in algorithm calculation complexity, can quickly align the same user in the network, and is high in applicability to real data.
Description
Technical Field
The invention belongs to the field of social media data mining, and particularly relates to a multi-source heterogeneous network user alignment method based on graph embedding.
Background
Online social networks have become an integral part of people's daily lives, with almost everyone using online social networks to connect friends, and a variety of social networks provide us with different types of services, such as Twitter, Facebook, YouTube, and so on. People may be attracted to different functions provided by different social networks, for example, a user logging in to an Instagram to share photos with his friends, and using Twitter to share opinions and emotions with others. To meet different levels of social needs, a user will register accounts on multiple social platforms, a survey in the united states shows that two thirds of online adults (66%) use the social media platform to keep in touch with friends, family and business partners.
Aligning users of multi-source heterogeneous networks has attracted tremendous attention in academia and industry because it can provide a comprehensive view of a user's profile that can benefit from a variety of applications. For example, if we can properly integrate different social networks together, one integrated personal information file can be created for each user and a better user interest model can be built. Linking the users of the e-commerce system and the social media platform correctly may then cause product recommendations to go from the e-commerce system to the social media platform.
Although many platforms provide users with interfaces with uniform account IDs, there is a large number of users that prefer to use isolated accounts for different platforms because many people may consider security or anonymity. How to judge the identity similarity of users in different social networks is a research focus.
Disclosure of Invention
In order to overcome the defects of the prior art and solve the problems of user identity virtualization and anonymity in the social network, the invention aims to provide a graph embedding-based multi-source heterogeneous network user alignment method, which can realize quick and accurate user identity alignment among a plurality of large-scale social networks.
In order to achieve the purpose, the invention adopts the technical scheme that:
a multi-source heterogeneous network user alignment method based on graph embedding comprises the following steps:
step 1: multi-source heterogeneous social network data pre-processing
Representing users of the social network as nodes, representing friends or concern relations among the users as edges, constructing a topological graph G (V, E) of the undirected weightless graph, wherein V is a set of nodes in the social network, E is a set of edges in the topological graph G, and counting the number of neighbors of each node.
Step 2: calculating user name similarity and social role similarity of users across social networks based on attribute information of the users, and combining and calculating the user name similarity and the social role similarity to serve as the user attribute similarity, specifically comprising the following steps:
step 2.1: the username similarity across social networks, i.e. the minimum number of editing operations required to transition from one to the other, is calculated by the levenstein distance. The allowed editing operations include replacing one character with another, inserting one character, and deleting one character.
Step 2.2: and calculating the score of each user by using a PageRank algorithm, and sorting the scores, wherein the top 1% of nodes are marked as large V users, the top 10% of other nodes are marked as active users, and the rest nodes are marked as normal users.
And 2.3, splicing and calculating the user name similarity and the social role similarity to serve as the user attribute similarity.
And step 3: executing a random walk algorithm on all nodes of the topological graph G to obtain the structure information of the network and the context information of each node, and training to obtain the embedded representation of the network, wherein the method specifically comprises the following steps:
step 3.1, performing multiple random walks with fixed length on all nodes in the topological graph G, and recording a traversal node sequence;
step 3.2, traversing the vertex sequence which is randomly walked in a sliding window mode, wherein in each window, a central vertex is used as input, and other nodes are used as the context of the central vertex;
step 3.3, for the current central vertex, updating the D-dimensional embedded representation of the current vertex using the context objective function, thereby learning the embedded representation of each central vertex.
And 4, step 4: the method comprises the following steps of splicing user attribute similarity of a user pair across networks and structural features of two users in the user pair to serve as input features, training a multilayer neural network to align users of a social network according to known anchor nodes, and specifically comprises the following steps:
step 4.1: constructing a neural network model, splicing the user attribute similarity of a user pair across networks and the structural features of two users in the user pair as input features, wherein the output of the neural network is a two-dimensional vector, [1,0] represents that the user pair is the same user, and [0,1] represents that the user pair is different users, the middle of the neural network contains a high-dimensional hidden layer, and the output is mapped into the range of [0,1] by using a softmax function to construct a loss function:
where y represents the true result of the sample,
prediction results of representative samples
Step 4.2: taking a known anchor node as a positive sample, constructing a negative sample user pair by a random extraction method, and performing the following steps according to a positive sample 1: 1 to train a multi-layer neural network. And (3) optimizing the target network by using a random gradient descent method, setting the learning rate lambda to be 0.001, setting 50 samples in each training batch, and fitting the data after multiple rounds of training.
Step 4.3: and inputting the unknown sample into the trained model to obtain the user pair with aligned identities, and aligning the two networks.
Further, in the step 3.1, all nodes in the topological graph G are subjected to random walk γ times, the order of the nodes in G is disturbed every time, t steps of random walk are performed starting from the initial node, a target node of each random walk is a node selected with a medium probability from neighbor nodes of the current node, and the initial node and the target node of each walk are recorded to obtain a traversal node sequence.
In the step 3.2, the size of the window is W, each node is sequentially used as a central vertex, and the front and rear W nodes are used as the context of the central vertex;
in step 3.3, the embedded representation of the central vertex is updated by taking the central vertex and the context node as input, so as to obtain the embedded representation of each node in the topological graph G.
In the present invention, the preset length d of the embedded vector representation is 64, and the preset length w of the window size is 5.
Compared with the prior art, the invention has the beneficial effects that:
(1) according to the invention, cross-network user alignment is carried out only through the attribute information and the network topology structure information of the users in the social network, data is easy to obtain, privacy-related user information is not required additionally, and the method is suitable for user alignment in most social networks.
(2) The invention compares the attribute similarity of the users through the aspects of user names, social roles and the like, extracts network topology information through a random walk graph embedding method, and fully considers the characteristics of the users in the social network, thereby enabling the extracted characteristics of the invention to be more effective.
(3) The invention uses multilayer neural networks to align users across networks, and can learn the special relationship between any two social networks through training, so that the method can be expanded to any two networks and obtain higher accuracy.
Drawings
FIG. 1 is a schematic flow chart of a multi-source heterogeneous network user alignment method based on graph embedding according to the present invention
Fig. 2 is a schematic diagram of the steps of graph embedding according to random walks.
Detailed Description
The embodiments of the present invention will be described in detail below with reference to the drawings and examples.
The invention relates to a multi-source heterogeneous network user alignment method based on graph embedding.
Graph embedding is the conversion of an attribute graph into a vector or a vector set, and the embedded vector can capture the topology of the graph, the vertex-to-vertex relationship and other relevant information about the graph, the subgraph and the vertex. More attribute-embedded coding can achieve better results in later tasks.
The multi-source heterogeneous network user alignment method based on graph embedding realizes user alignment in a plurality of social networks through a plurality of methods such as graph embedding, neural networks and the like, and has important significance in a plurality of applications such as a recommendation system, portrait completion, link prediction and the like.
As shown in fig. 1, the present invention specifically includes the following steps:
step S1: preprocessing online social network data;
the online social network data set in the embodiment is derived from http:// www.apexlab.org/datasets/8, a microblog-broad bean data set in the data set is selected for experiment, the data set comprises 6,280,561 relations of 141614 microblog users and attention or attention among the users, 141614 users of broad bean and 2,700,602 friend relations, and users with few friend relations and isolated scatter points are removed. The bean delete friend relation is less than 5 nodes, the microblog delete fan and the focus is less than 30 nodes, 59,033 edges of the bean nodes 1,277,548 are obtained, 46,208 edges of the microblog nodes 2,081,109 are obtained, and 30,095 aligned users are obtained.
Representing users of the social network as nodes, representing friends or attention relations among the users as edges, and constructing an undirected topology graph G ═ V, E, wherein V is a set of nodes in the social network, and E is a set of edges in the topology graph G.
Step S2: calculating the user similarity among the cross-social networks based on the attribute information of the users, wherein the calculation of the user attribute similarity specifically comprises the following steps:
step S201: the username similarity, i.e. the minimum number of editing operations required between two username strings to switch from one to the other, is calculated by the levenstein distance. The editing operations allowed include replacing one character with another, inserting one character, deleting one character, and the transformation process of the two user name character strings is shown in table 1. The levenstein distance statistics for aligned users are shown in table 2, where the user names of the same user in different networks are very similar.
TABLE 1
TABLE 2
Step S202: and respectively calculating the score of each user in the two social networks by using a PageRank algorithm, and ranking the scores, wherein the first 1% of nodes are marked as large V users, the first 10% of other nodes are marked as active users, and the rest nodes are marked as common users.
Step S203: the user name similarity and the social role similarity are spliced to be used as the user attribute similarity, and the user attribute similarity occupies 64-dimensional features.
Step S3: executing a random walk algorithm on all nodes of the topology graph G to obtain structure information of the network and context information of each node, and training to obtain an embedded representation of the network, as shown in fig. 2, specifically including the following steps:
step S301: and (3) carrying out random walk on all nodes in the topological graph G for 10 times, for each time, disturbing the sequence of the nodes in the topological graph G, starting from the initial node, carrying out 40-step random walk, wherein the target node of each random walk is a node selected from neighbor nodes of the current node at a medium probability, and recording the initial node and the target node of each walk to obtain a traversal node sequence.
Step S302: and traversing the vertex sequence which is randomly walked in a sliding window mode, wherein the size of the window is 5, and in each window, the central vertex is used as input, and other nodes are used as the context of the central vertex.
Step S303: and taking the central vertex and the context node as input, updating the 64-dimensional embedded representation of the node, and obtaining the embedded representation of each node in the topological graph G after all the nodes are trained.
Step S4: the method comprises the following steps of training a multilayer neural network to align social network users according to known anchor nodes by using attribute similarity of a user pair across networks and structural feature splicing of two users as input features, wherein the method specifically comprises the following steps:
step S401: constructing a neural network model, using 64-dimensional attribute similarity of a user pair across the network and 192-dimensional characteristics of structural characteristic splicing of two users as input characteristics, wherein the output of the neural network is a two-dimensional vector, [1,0] represents that the user pair is the same user, and [0,1] represents that the user pair is different users, the middle of the neural network contains a 320-dimensional hidden layer, the RELU is used as an activation function, the output is mapped into the range of [0,1] by using a softmax function, and a loss function is constructed:
step S402: taking the known 30095 anchor nodes as positive samples, constructing the same number of negative sample user pairs by a random extraction method, and training set: the test set was 5: 1, positive and negative samples are in a ratio of 1: 1 to train a multi-layer neural network. And (3) optimizing the target network by using a random gradient descent method, setting the learning rate lambda to be 0.001, setting 50 samples in each training batch, and fitting the data after multiple rounds of training.
In order to check the effect of the user alignment method based on the multi-source heterogeneous network provided by the present invention in this embodiment, the test set in this embodiment is tested, the whole data set includes 2500 positive samples and 2500 negative samples, and precision and recall are calculated on the test set as evaluation indexes.
precision is for the prediction result, which indicates how many of the samples predicted to be positive are true positive samples. There are two possibilities, one is to predict the positive class as positive (TP) and the other is to predict the negative class as positive (FP), i.e.:
recall is for the original sample and indicates how many of the positive cases in the sample are predicted correctly. There are two possibilities, one is to predict the original positive class as a positive class (TP) and the other is to predict the original positive class as a negative class (FN), i.e.:
the experimental results of this example are as follows:
the accuracy values for the test set stabilized at 0.8943 and the recall stabilized at 0.8454, as shown in Table 3.
TABLE 3
| Evaluation index | Results of the experiment |
| precision | 0.8943 |
| recall | 0.8454 |
The experimental result shows that the user alignment method based on the multi-source heterogeneous network can achieve user alignment in an online social network and achieve a good effect. The method can be used for user alignment of the online social network, has important application in multiple fields of recommendation systems, human and figure image completion and the like, is low in calculation complexity of an algorithm, can quickly align the same user in the network, and is high in applicability to real data.
Although illustrative embodiments of the present invention have been described above to facilitate the understanding of the present invention by those skilled in the art, it should be understood that the present invention is not limited to the scope of the embodiments, and various changes may be made apparent to those skilled in the art as long as they are within the spirit and scope of the present invention as defined and defined by the appended claims, and all matters of the invention which utilize the inventive concepts are protected.
Claims (6)
1. A multi-source heterogeneous network user alignment method based on graph embedding is characterized by comprising the following steps:
step 1, preprocessing multi-source heterogeneous social network data, representing users of a social network as nodes, representing friends or concern relations among the users as edges, constructing an undirected and unweighted topological graph G (V, E), wherein V is a set of the nodes in the social network, E is a set of the edges in the topological graph G, and counting the number of neighbors of each node;
step 2, calculating user name similarity and social role similarity of the users across the social network based on the attribute information of the users, and splicing and calculating the user name similarity and the social role similarity to serve as the user attribute similarity;
step 3, executing a random walk algorithm on all nodes of the topological graph G to obtain the structure information of the network and the context information of each node, and training to obtain the embedded representation of the network;
step 4, splicing the user attribute similarity of the user pairs across the network and the structural features of the two users in the user pairs to serve as input features, and training the multilayer neural network to align the social network users according to the known anchor nodes;
the specific method of the step 4 comprises the following steps:
step 4.1, constructing a neural network model, splicing the user attribute similarity of a user pair across the network and the structural characteristics of two users in the user pair as input characteristics, wherein the output of the neural network is a two-dimensional vector, [1,0]]Representing that the user pair is the same, [0,1]]Then the user pair is represented as different users, a high-dimensional hidden layer is arranged in the middle, and the output is mapped to 0,1 by using the softmax function]Within the range, a loss function L is constructed,
where y represents the true result of the sample,
a prediction result representing the sample;
step 4.2, taking the known anchor node as a positive sample, constructing a negative sample user pair by a random extraction method, and performing the following steps according to the positive sample and the negative sample 1: 1, training a multilayer neural network, optimizing a target network by using a random gradient descent method, setting a learning rate lambda to be 0.001, setting 50 samples in each training batch, and fitting data after multiple rounds of training;
and 4.3, inputting the unknown sample into the trained model to obtain the user pair with aligned identities, and aligning the two networks.
2. The graph embedding-based multi-source heterogeneous network user alignment method according to claim 1, wherein the specific method in the step 2 is as follows:
step 2.1, calculating the user name similarity between the social networks through the Levensan distance, namely, the minimum number of editing operations required for converting one user name character string into the other user name character string, wherein the allowed editing operations comprise replacing one character with another character, inserting one character and deleting one character;
step 2.2, calculating the score of each user by using a PageRank algorithm, and sequencing the scores, wherein the first 1% of nodes are marked as large-V users, the first 10% of other nodes are marked as active users, and the rest nodes are marked as common users;
and 2.3, splicing and calculating the user name similarity and the social role similarity to serve as the user attribute similarity.
3. The graph embedding-based multi-source heterogeneous network user alignment method according to claim 1, wherein the specific method in step 3 is as follows:
step 3.1, performing multiple random walks with fixed length on all nodes in the topological graph G, and recording a traversal node sequence;
step 3.2, traversing the vertex sequence which is randomly walked in a sliding window mode, wherein in each window, a central vertex is used as input, and other nodes are used as the context of the central vertex;
step 3.3, for the current central vertex, updating the D-dimensional embedded representation of the current vertex using the context objective function, thereby learning the embedded representation of each central vertex.
4. The multi-source heterogeneous network user alignment method based on graph embedding according to claim 3, wherein in the step 3.1, all nodes in the topological graph G are subjected to gamma random walks, for each time, the sequence of the nodes in the G is disturbed, t steps of random walks are performed from an initial node, a target node of each random walk is a node selected with a medium probability from neighbor nodes of a current node, the initial node and the target node of each walk are recorded, and a traversal node sequence is obtained;
in the step 3.2, the size of the window is W, each node is sequentially used as a central vertex, and the front and rear W nodes are used as the context of the central vertex;
in step 3.3, the embedded representation of the central vertex is updated by taking the central vertex and the context node as input, so as to obtain the embedded representation of each node in the topological graph G.
5. The graph embedding-based multi-source heterogeneous network user alignment method according to claim 3 or 4, wherein the preset length of the window size W is 5.
6. The graph embedding-based multi-source heterogeneous network user alignment method according to claim 3 or 4, wherein the preset length d of the embedded representation is 64.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010778910.8A CN112084373B (en) | 2020-08-05 | 2020-08-05 | Graph embedding-based multi-source heterogeneous network user alignment method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010778910.8A CN112084373B (en) | 2020-08-05 | 2020-08-05 | Graph embedding-based multi-source heterogeneous network user alignment method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112084373A CN112084373A (en) | 2020-12-15 |
| CN112084373B true CN112084373B (en) | 2022-06-03 |
Family
ID=73736033
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010778910.8A Active CN112084373B (en) | 2020-08-05 | 2020-08-05 | Graph embedding-based multi-source heterogeneous network user alignment method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112084373B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112953825A (en) * | 2021-01-26 | 2021-06-11 | 中山大学 | Attribute heterogeneous network embedding method, device, equipment and medium |
| CN113095948B (en) * | 2021-03-24 | 2023-06-06 | 西安交通大学 | Multi-source heterogeneous network user alignment method based on graph neural network |
| CN113409157B (en) * | 2021-05-19 | 2022-06-28 | 桂林电子科技大学 | Cross-social network user alignment method and device |
| CN113255895B (en) * | 2021-06-07 | 2023-06-16 | 之江实验室 | Structure diagram alignment method and multi-diagram joint data mining method based on diagram neural network representation learning |
| CN113779520B (en) * | 2021-09-07 | 2023-06-13 | 中国船舶重工集团公司第七0九研究所 | Cross-space target virtual identity association method based on multi-layer attribute analysis |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097125A (en) * | 2019-05-07 | 2019-08-06 | 郑州轻工业学院 | A kind of across a network account correlating method indicated based on insertion |
| CN110347932A (en) * | 2019-06-04 | 2019-10-18 | 中国科学院信息工程研究所 | A kind of across a network user's alignment schemes based on deep learning |
| CN111476673A (en) * | 2020-04-02 | 2020-07-31 | 中国人民解放军国防科技大学 | Method, device and medium for aligning users among social networks based on neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9195777B2 (en) * | 2012-03-07 | 2015-11-24 | Avira B.V. | System, method and computer program product for normalizing data obtained from a plurality of social networks |
-
2020
- 2020-08-05 CN CN202010778910.8A patent/CN112084373B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097125A (en) * | 2019-05-07 | 2019-08-06 | 郑州轻工业学院 | A kind of across a network account correlating method indicated based on insertion |
| CN110347932A (en) * | 2019-06-04 | 2019-10-18 | 中国科学院信息工程研究所 | A kind of across a network user's alignment schemes based on deep learning |
| CN111476673A (en) * | 2020-04-02 | 2020-07-31 | 中国人民解放军国防科技大学 | Method, device and medium for aligning users among social networks based on neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112084373A (en) | 2020-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112084373B (en) | Graph embedding-based multi-source heterogeneous network user alignment method | |
| Jenkins et al. | Unsupervised representation learning of spatial data via multimodal embedding | |
| CN107562812A (en) | A kind of cross-module state similarity-based learning method based on the modeling of modality-specific semantic space | |
| Wang et al. | Retweet wars: Tweet popularity prediction via dynamic multimodal regression | |
| CN113095948B (en) | Multi-source heterogeneous network user alignment method based on graph neural network | |
| CN112036445B (en) | User identity recognition method across social networks based on neural tensor network | |
| CN113139140B (en) | Tourist attraction recommendation method based on space-time perception GRU and combined with user relationship preference | |
| KR101224312B1 (en) | Friend recommendation method for SNS user, recording medium for the same, and SNS and server using the same | |
| CN113434782B (en) | Cross-social network user identity recognition method based on joint embedded learning model | |
| CN113422761A (en) | Malicious social user detection method based on counterstudy | |
| CN115470994B (en) | Information popularity prediction method and system based on explicit time and cascade attention | |
| CN109949174A (en) | A kind of isomery social network user entity anchor chain connects recognition methods | |
| CN112559747A (en) | Event classification processing method and device, electronic equipment and storage medium | |
| CN105512301A (en) | User grouping method based on social content | |
| CN109960755A (en) | A kind of privacy of user guard method based on Dynamic iterations Fast Field | |
| CN115687760A (en) | User learning interest label prediction method based on graph neural network | |
| CN109885797B (en) | Relational network construction method based on multi-identity space mapping | |
| CN114942998A (en) | Entity alignment method for sparse knowledge graph neighborhood structure fusing multi-source data | |
| Yuan et al. | User naming conventions mapping learning for social network alignment | |
| CN112380455A (en) | Method for directionally and covertly acquiring data of international and foreign internet based on backtracking security controlled network access channel | |
| US20240086731A1 (en) | Knowledge-graph extrapolating method and system based on multi-layer perception | |
| Sun et al. | Mapping users across social media platforms by integrating text and structure information | |
| CN115080871A (en) | Cross-social network social user alignment method | |
| Ma et al. | Friend closeness based user matching cross social networks | |
| CN115063251A (en) | Social communication propagation dynamic network representation method based on relationship strength and feedback mechanism |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |