CN112989199B - Cooperative network link prediction method based on multidimensional proximity attribute network - Google Patents
Cooperative network link prediction method based on multidimensional proximity attribute network Download PDFInfo
- Publication number
- CN112989199B CN112989199B CN202110343021.3A CN202110343021A CN112989199B CN 112989199 B CN112989199 B CN 112989199B CN 202110343021 A CN202110343021 A CN 202110343021A CN 112989199 B CN112989199 B CN 112989199B
- Authority
- CN
- China
- Prior art keywords
- network
- proximity
- features
- author
- loss function
- 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 26
- 230000006870 function Effects 0.000 claims abstract description 58
- 239000013598 vector Substances 0.000 claims abstract description 26
- 230000001149 cognitive effect Effects 0.000 claims abstract description 23
- 238000011478 gradient descent method Methods 0.000 claims abstract description 4
- 238000011160 research Methods 0.000 claims description 25
- 238000005295 random walk Methods 0.000 claims description 10
- 230000007704 transition Effects 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 4
- 230000002776 aggregation Effects 0.000 claims description 4
- 238000004220 aggregation Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 230000008878 coupling Effects 0.000 claims description 2
- 238000010168 coupling process Methods 0.000 claims description 2
- 238000005859 coupling reaction Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 8
- 238000005457 optimization Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000019771 cognition Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 239000008186 active pharmaceutical agent Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 231100000027 toxicology Toxicity 0.000 description 1
- 238000012549 training Methods 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
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/194—Calculation of difference between files
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a cooperative network link prediction method based on a multidimensional proximity attribute network, which belongs to the field of cooperative recommendation and comprises the following steps: respectively reserving multidimensional proximity features, local network features and global network features by using a self-encoder model, a joint probability model and an attribute Skip-Gram model; wherein the multidimensional proximity features include cognitive proximity features, geographic proximity features, and institutional proximity features; the loss function of the self-coding model, the loss function of the local network characteristic, the loss function of the global network characteristic and the loss function of the L2-norm are combined to serve as an overall objective function, and a random gradient descent method is adopted to optimize the overall objective function, so that the representation learning of the network nodes is realized; and carrying out cooperative network link prediction through the vector cosine similarity corresponding to the network node. According to the method, the network characteristics and the node attribute information are comprehensively considered, and the accuracy of the cooperative network link prediction is improved.
Description
Technical Field
The invention belongs to the field of collaborative recommendation, and in particular relates to a collaborative network link prediction method based on a multidimensional proximity attribute network.
Background
Partner recommendations are of great importance to promote scientific research cooperation. Existing literature research has focused mainly on recommending partnerships between all co-authors. The existing partner recommendation method is mainly based on a network model, a content model and a mixed model. Partner recommendation methods based on network models incorporate local network functions (e.g., public neighbors) or global network functions (e.g., random walk RWR with restart). Partner recommendation methods based on the content model recommend authors by extracting content features (e.g., LDA-based similarity). Partner recommendation methods based on hybrid models combine network features and content features. The attribute network embedded model combined with the network characteristics and the node attribute information shows good performance. The existing literature indicates five dimensions of scientific research collaboration proximity (proximity). However, existing partner recommendation methods only include social proximity (social proximity) that belongs to network features or cognitive proximity that belongs to text features.
Patent document CN104573103B proposes a partner recommendation method under a scientific literature heterogeneous network, but the method only considers the recommendation of a partner by a pair of authors and the willingness of the authors to cooperate with each other, and does not take the main author into consideration.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a cooperative network link prediction method based on a multidimensional proximity attribute network, which aims to solve the problems that the existing cooperative partner recommendation method is not considered from multidimensional proximity characteristics, and scientific research similarity and inter-author cooperation probability cannot be reflected after the multidimensional proximity characteristics are subjected to function processing, so that the prediction accuracy of author cooperation is lower.
In order to achieve the above object, the present invention provides a method for predicting a cooperative network link based on a multidimensional proximity attribute network, comprising the steps of:
respectively reserving multidimensional proximity features, local network features and global network features by using a self-encoder model, a joint probability model and an attribute Skip-Gram model; wherein the multidimensional proximity features include cognitive proximity features, geographic proximity features, and institutional proximity features;
the loss function of the self-coding model, the loss function of the local network characteristic, the loss function of the global network characteristic and the loss function of the L2-norm are combined to serve as an overall objective function, and a random gradient descent method is adopted to optimize the overall objective function, so that the representation learning of the network nodes is realized;
carrying out cooperative network link prediction through vector cosine similarity corresponding to the network node;
wherein the network node represents an author; cognitive proximity features characterize the cognitive level of authors in the scientific domain; the geographic proximity features characterize the position relationship of each author; the system proximity feature represents the similarity of the language of the position where the author is located; the local network characteristics represent the probability representation of the cooperation of each author; global network features represent scientific similarity by likelihood values of author proximity vectors.
Preferably, the cognitive proximity feature is expressed as:
wherein Cp is i,y For author a i The thesis cognitive vector accumulated sum published in y years; y is 0 Is the base year; y is an annual interval;
the geographic proximity feature is denoted gg= (VG, EG), where VG is the set of geographic nodes; EG is a collection of geographic edges;
preferably, the system proximity is measured in terms of a continuous aggregation index in a common language.
Preferably, the self-encoder model is:
h i ＝σ 1 (W (1) x i +b (1) )
wherein x is i For the author's neighboring feature vector, h i Is a hidden layer representation of the encoder;
is a reconstruction of the decoder; θ= { W (1) ,b (1) ,W (2) ,b (2) -model parameters; sigma (sigma) 1 (. Cndot.) is the tanh function in the activation function;
the loss function from the encoder model is:
where n is the total number of authors.
Preferably, the loss function of the local network feature is:
wherein p is ij Is author a i And author a j Is a joint probability of (2); e, e ij For author a i And author a j The connecting edges between the two.
Preferably, the loss function of the global network feature is:
wherein a is i+j For the node context in the generated node sequence, w is the window size; p (a) i+j |x i ) The conditional probability of (a) is context a i+j Giving likelihood values of the node i proximity vector; g is a set of all network nodes; c is all random walk sequences; a, a i Representing the author.
In general, the above technical solutions conceived by the present invention have the following beneficial effects compared with the prior art:
according to the invention, from the multi-dimensional proximity perspective, the attribute characteristics of scientific research collaborators are comprehensively covered; the cognition adjacency, the geographic adjacency and the system adjacency are pre-trained to be expressed into low-dimensional vectors, so that the attribute specific to each author can be comprehensively considered on the premise of not damaging the network characteristics; meanwhile, the reserved network characteristics comprise a local network and a global network, wherein the local network can accurately reflect the cooperation wish of each author, and the global network can more prominently reflect the similarity of scientific research; on the basis, the minimum value of the sum of the network characteristics and the node attribute characteristic loss function is used as a target to optimize, and the accuracy of the prediction of the cooperative network link is improved.
Drawings
Fig. 1 (a) is a schematic diagram of a distance weighted network of GPNs provided by an embodiment of the present invention;
fig. 1 (b) is a schematic diagram of a geographic neighboring network of a GPN according to an embodiment of the present invention;
fig. 1 (c) is a schematic diagram of transition probability of a GPN node D according to an embodiment of the present invention;
FIG. 2 (a) is a schematic diagram of an institutional adjacent network provided by an embodiment of the present invention;
fig. 2 (b) is a schematic diagram of transition probability of the IPN node d according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a scientific research cooperation network construction provided by an embodiment of the present invention;
FIG. 4 is a schematic diagram of a joint optimization framework provided by an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The invention provides a cooperative network link prediction method based on a multidimensional proximity attribute network, which comprises the following steps:
respectively reserving multidimensional proximity features, local network features and global network features by using a self-encoder model, a joint probability model and an attribute Skip-Gram model; wherein the multidimensional proximity features include cognitive proximity features, geographic proximity features, and institutional proximity features;
the loss function of the self-coding model, the loss function of the local network characteristic, the loss function of the global network characteristic and the loss function of the L2-norm are combined to serve as an overall objective function, and a random gradient descent method is adopted to optimize the overall objective function, so that the representation learning of the network nodes is realized;
carrying out cooperative network link prediction through vector cosine similarity corresponding to the network node;
wherein the network node represents an author; cognitive proximity features characterize the cognitive level of authors in the scientific domain; the geographic proximity features characterize the position relationship of each author; the system proximity feature represents the similarity of the language of the position where the author is located; the local network characteristics represent the cooperative relationship existing in each author; global network characteristics pass through neighborhood similarity among authors.
The following detailed description firstly describes a multi-dimensional proximity representation learning method related to the invention, and secondly describes a framework of an ARCR model (scientific cooperation recommendation model: attribute-aware research recommendation) provided by the invention.
1. The multi-dimensional proximity feature representation provided by the invention comprises a cognitive proximity feature representation, a geographic proximity feature representation and a system proximity representation.
(1) Cognitive proximity features represent:
scientific papers are linked text that contains not only text, but also quotation links. Both textual content information and linking information are essential to measure similarity of scientific papers. In the invention, text content characteristics and quotation link characteristics are represented by P2V (paper-to-vector) to obtain the cognitive proximity representation of the scientific research subject. Taking into account the dynamic change of the cognitive basis of the subject of scientific research, adopting a time weight attenuation factor for an author a i Based on the cognitive vectors of the annual published papers, the cognitive proximity characteristic representation is obtained:
wherein Cp is i,y For author a i The cumulative sum of the cognitive vectors of papers published in y years; y is 0 Is the base year; y is an annual interval.
(2) The geographic proximity feature represents:
a geographic proximity network (GPN: geographical Proximity Network) is defined as gg= (VG, EG), where VG is a set of geographic nodes; v 1 E VG, represents a city; EG is a collection of geographic edges; e, e 1 EG, represents the geographic relationship between two cities, e 1 ＝(u 1 ,v 1 ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein u is 1 Is different from v 1 Is a city of another city; e, e 1 Weighting of edges
Correlation;
Is city u 1 And city v 1 Distance between them.
FIG. 1 (a) is a distance weighted network of GPNs; FIG. 1 (b) is a geographic proximity network of GPNs; fig. 1 (c) is a transition probability of the network node D; in the figure, A, B, C and D represent different cities; and (3) performing bias random walk on the GPN, wherein the transition probability is proportional to the weight of the edge, generating a city node sequence (shown in fig. 1 (c)) according to the transition probability, and finally executing an attribute Skip-Gram model. Two cities with closer geographic distances are more likely to co-occur in the sampling result, resulting in a similar vector representation.
(3) The system proximity feature represents:
scientific research cooperation can be easier when subjects of research have similar cultural backgrounds. Language is the core of culture. Jacque Mei Lici (jacques. Melitz) integrates common native, common spoken, common official languages and language similarity, thereby proposing a continuous aggregate index (continual aggregate index) for common languages that is superior to the traditional dummy variable model. Therefore, the index is used for measuring the system adjacency of the scientific research subjects.
The institutional neighbor network (IPN: institutional Proximity Network) is defined as gi= (VI, EI), VI is the institutional node set, v 2 Epsilon VI represents a specific country; EI is the set of institutional edges; each system side e 2 E EI, the system adjacency between two countries, e 2 ＝(u 2 ,v 2 ) Weights to the side of the system
Related (I)>
Is a continuous aggregation index of common language between country pairs; wherein u is 2 Is different from v 2 Is a country of another country.
Fig. 2 (a) is a institutional neighboring network, and fig. 2 (b) is a transition probability of network node d; a node sequence is generated based on random walks (random walks) at the IPN, and then the attribute Skip-Gram algorithm is performed on the node sequence. In the sampling result, two countries with greater system proximity are more likely to occur simultaneously, resulting in a similar representation.
2. Scientific research network (Coauthorship network)
The scientific research cooperation network reflects the social adjacency, and the scientific research main body composes the cooperation information of the published papers to construct the scientific research cooperation network. FIG. 3 is a schematic diagram of a scientific research collaboration network construction; when two subjects have a treatise, the two subjects have two sides with a weight of 1; when two subjects have k treatises, then both have a continuous edge with weight k.
3. Statement of problem
G= (a, E, X) is an attribute scientific network, where a= { a 1 ,a 2 ,...,a n -is the author set; each edge e ij ＝(a i ,a j ) E represents author a i And a j Scientific research cooperation relation between the two; x epsilon R n×m Representing a node attribute matrix; x is x i Is author a i Is a proximity feature vector of (1); node attribute information is a union of cognitive, geographic and institutional proximity vectors; the aim is to map the function f by learning: a, a i →h i ∈R d Each author a i Expressed as a low-dimensional vector h i Wherein d is less than n, and network characteristics and node attribute information are reserved; the network features include local network features and global network features. The local network characteristics indicate whether an edge exists between two authors; global network characteristics represent high-order neighborhood similarities of nodes.
4. Self-encoder based on proximity features
The proximity feature belongs to the attribute feature, and the self-encoder is adopted to store node attribute information. The self-encoder model comprises an input layer, a hidden layer and an output layer; the representation function of the self-encoder is:
h i ＝σ 1 (W (1) x i +b (1) )
wherein x is i For the author's neighboring feature vector, h i ∈R d Is a hidden layer representation of the encoder;
is a reconstruction of the decoder; θ= { W (1) ,b (1) ,W (2) ,b (2) -model parameters; sigma (sigma) 1 (. Cndot.) is the tanh function in the activation function;
model parameters were learned by minimizing the following loss functions.
In order to preserve high nonlinearity in the attribute information, a K-layer hidden layer is adopted in the encoder;
…
wherein,,
representation author a i The desired low-dimensional hidden representation. Correspondingly, a K-layer hidden layer is also used in the decoder.
5. Local network features
The network features include local network features and global network features; the local network characteristics are as follows:
the following likelihood estimates are maximized to preserve local network characteristics: l (L) f ＝∏e ij>0 p ij ；
Wherein p is ij Is a i And a j Is a joint probability of (a):
thus, the negative likelihood can be minimized, specifically as follows:
6. global network features
To preserve global network characteristics, a Skip-Gram model based on attributes is employed. By providing the current node a for all random walk sequences C e C i And its proximity feature x i The following negative log likelihood is minimized:
wherein a is i+j For the node context in the generated node sequence, w is the window size; p (a) i+j |x i ) The conditional probability of (a) is context a i+j Giving likelihood values of the node i proximity vector; g is a set of all network nodes; c is all random walk sequences;
wherein f (·) is a function of the encoder portion of the automatic encoder model;
is a context node->
Corresponding representation of (a). However, this formula is computationally expensive, and therefore, p (a) i+j |x i ) The substitution is as follows:
wherein sigma 2 (. Cndot.) is the sigmoid function in the activation function, |neg| is the number of negative examples samples;
as a desired function;
d a Is the degree of node a;
7. joint optimization framework
FIG. 4 is a joint optimization framework. Since the self-encoder model, the joint probability model, and the attribute-aware skip-gram model share the same encoder layer, the models are closely connected.
The final representation of (a) captures network characteristics and node attribute information.
Combining the three objective functions to obtain a total objective function of the joint model:
optimizing the loss function in the above using a random gradient descent algorithm, iteratively optimizing the two coupling components (αl f +βL ae +γL reg ,and L h )；L reg Is that
8. Scientific research collaboration recommendation
Based on author a i Low-dimensional hidden representation of (a)
Author a j Is a low-dimensional hidden representation->
Calculate h i And h j Cosine similarity; finally, the first k similar authors are recommended to the target authors as potential scientific research cooperation objects.
Examples
Data sources, using the Web of Science core quotation database to collect papers of the medical category of 2010-2019 ("biochemistry AND molecular biology", "medicine, research AND experiment", "pharmacology AND pharmacy" AND "toxicology"), search queries were "wc=apy=b AND language= 'engish'", where a is the pharmaceutical domain AND B is 2010-2019. Excluding independent authors' papers and non-journal papers, 528118 papers were eventually retrieved. The authors who published more than 5 papers in 162196 were screened after the name of the authors disambiguated. Geographic information is obtained using the Google map API. And acquiring language information by using CEPII language. The dataset was divided into two parts by year of publication: data before 2018 is used as a training set, and data in 2018-2019 is used as a test set.
The implementation process comprises the following steps: establishing a three-layer automatic encoder; the hidden dimensions of the first, second and third layers are d (1) =600, d (2) =512, d (3) =256, respectively; the dimension of both the geographic proximity vector and the institutional proximity vector is 64, and the dimension of the cognitive proximity vector is 256. The weight of the local network feature loss function is set to α=1, the weight of the auto encoder loss function is set to β=10, and the weight of the L2-norm regularization γ is set to 10. 100 authors were randomly selected as target nodes and ARCR was run.
In summary, the present invention has the following advantages:
according to the invention, from the multi-dimensional proximity perspective, the attribute characteristics of scientific research collaborators are comprehensively covered; the cognition adjacency, the geographic adjacency and the system adjacency are pre-trained to be expressed into low-dimensional vectors, so that the attribute specific to each author can be comprehensively considered on the premise of not damaging the network characteristics; meanwhile, the reserved network characteristics comprise a local network and a global network, wherein the local network can accurately reflect the cooperation wish of each author, and the global network can more prominently reflect the similarity of scientific research; on the basis, the minimum value of the sum of the network characteristics and the node attribute characteristic loss function is used as a target to optimize, and the accuracy of the prediction of the cooperative network link is improved.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (6)
1. A cooperative network link prediction method based on a multidimensional proximity attribute network is characterized by comprising the following steps:
respectively reserving multidimensional proximity features, local network features and global network features by using a self-encoder model, a joint probability model and an attribute Skip-Gram model; wherein the multidimensional proximity features include cognitive proximity features, geographic proximity features, and institutional proximity features;
the loss function of the self-coding model, the loss function of the local network characteristic, the loss function of the global network characteristic and the loss function of the L2-norm are combined to serve as an overall objective function, and a random gradient descent method is adopted to optimize the overall objective function, so that the representation learning of the network nodes is realized;
carrying out cooperative network link prediction through vector cosine similarity corresponding to the network node;
wherein the network node represents an author; the multi-dimensional proximity feature representation includes a cognitive proximity feature representation, a geographic proximity feature representation, and a institutional proximity feature representation; cognitive proximity features characterize the cognitive level of authors in the scientific domain; the geographic proximity features characterize the position relationship of each author; the system proximity feature represents the similarity of the language of the position where the author is located; the local network characteristics represent the cooperative relationship existing in each author; the global network features represent scientific research similarity through likelihood values of author proximity vectors;
wherein the geographic proximity network in the geographic proximity feature is defined as gg= (VG, EG), where VG is a set of geographic nodes; v 1 E VG, represents a city; EG is a collection of geographic edges; e, e 1 EG, represents the geographic relationship between two cities, e 1 ＝(u 1 ,v 1 ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein u is 1 Is different from v 1 Is a city of another city; e, e 1 Weighting of edges
Correlation;
Is city u 1 And city v 1 A distance therebetween; performing biased random walk on a geographic proximity network, wherein the transition probability is in direct proportion to the weight of the edge, generating a city node sequence according to the transition probability, and finally executing an attribute Skip-Gram model; the more likely two cities with closer geographic distances are co-present in the sampling result, and similar vector representations are obtained; wherein the system proximity network in the system proximity feature is defined as gi= (VI, EI), VI is the system node set, v 2 Epsilon VI represents a specific country; EI is the set of institutional edges; each system side e 2 E EI, the system adjacency between two countries, e 2 ＝(u 2 ,v 2 ) Weights with the system side->
Related (I)>
Is a continuous aggregation index of common language between country pairs; wherein u is 2 Is different from v 2 Is a country of another country; performing random walk on a system proximity network to generate a national node sequence, and performing an attribute Skip-Gram algorithm on the national node sequence; in the sampling result, the more likely two countries with greater system proximity are to appear simultaneously, resulting in a similar representation;
the self-encoder model, the joint probability model and the attribute aware skip-gram model share the same encoder layer;
the overall objective function is:
L＝L h +αL f +βL ae +γL reg
optimizing the whole objective function by adopting a random gradient descent algorithm, and iteratively optimizing the two coupling components alpha L f +βL ae +γL reg And L h ；L f A loss function that is a local network feature; l (L) h Is a global network feature; l (L) ae A loss function that is a self-coding model; l (L) reg Is a loss function of the L2-norm.
2. The cooperative network link prediction method according to claim 1, wherein the institutional adjacency is measured in terms of a continuous aggregation index in a common language.
3. The cooperative network link prediction method according to claim 2, wherein the self-encoder model is: h is a i ＝σ 1 (W (1) x i +b (1) )，
Wherein x is i For the author's neighboring feature vector, h i Is a hidden layer representation of the encoder;
is a reconstruction of the decoder; θ= { W (1) ,b (1) ,W (2) ,b (2) -model parameters; sigma (sigma) 1 (. Cndot.) is the tanh function in the activation function. />
4. A method of collaborative network link prediction according to claim 3, wherein the loss function of the self-encoder model is:
where n is the total number of authors.
5. The cooperative network link prediction method according to claim 2, wherein the loss function of the local network characteristic is:
wherein p is ij Is author a i And author a j Is a joint probability of (2); e, e ij For author a i And author a j The connecting edges between the two.
6. The cooperative network link prediction method according to claim 2, wherein the loss function of the global network feature is:
wherein a is i+j For the node context in the generated node sequence, w is the window size; p (a) i+j |x i ) The conditional probability of (a) is context a i+j Giving likelihood values of the node i proximity vector; g is a set of all network nodes; c is all random walk sequences; a, a i Representing the author.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110343021.3A CN112989199B (en) | 2021-03-30 | 2021-03-30 | Cooperative network link prediction method based on multidimensional proximity attribute network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110343021.3A CN112989199B (en) | 2021-03-30 | 2021-03-30 | Cooperative network link prediction method based on multidimensional proximity attribute network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112989199A CN112989199A (en) | 2021-06-18 |
| CN112989199B true CN112989199B (en) | 2023-05-30 |
Family
ID=76338483
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110343021.3A Active CN112989199B (en) | 2021-03-30 | 2021-03-30 | Cooperative network link prediction method based on multidimensional proximity attribute network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112989199B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8732101B1 (en) * | 2013-03-15 | 2014-05-20 | Nara Logics, Inc. | Apparatus and method for providing harmonized recommendations based on an integrated user profile |
| CN109145087A (en) * | 2018-07-30 | 2019-01-04 | 大连理工大学 | A kind of scholar's recommendation and collaborative forecasting method based on expression study and competition theory |
| CN111368074A (en) * | 2020-02-24 | 2020-07-03 | 西安电子科技大学 | Link prediction method based on network structure and text information |
| CN112069306A (en) * | 2020-07-22 | 2020-12-11 | 中国科学院计算机网络信息中心 | Paper partner recommendation method based on author writing tree and graph neural network |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8170971B1 (en) * | 2011-09-28 | 2012-05-01 | Ava, Inc. | Systems and methods for providing recommendations based on collaborative and/or content-based nodal interrelationships |
| CN109101629A (en) * | 2018-08-14 | 2018-12-28 | 合肥工业大学 | A kind of network representation method based on depth network structure and nodal community |
| US11544530B2 (en) * | 2018-10-29 | 2023-01-03 | Nec Corporation | Self-attentive attributed network embedding |
| CN110245682B (en) * | 2019-05-13 | 2021-07-27 | 华中科技大学 | Topic-based network representation learning method |
| CN111709474A (en) * | 2020-06-16 | 2020-09-25 | 重庆大学 | Graph embedding link prediction method fusing topological structure and node attributes |
| CN112256870A (en) * | 2020-10-15 | 2021-01-22 | 大连理工大学 | Attribute network representation learning method based on self-adaptive random walk |
-
2021
- 2021-03-30 CN CN202110343021.3A patent/CN112989199B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8732101B1 (en) * | 2013-03-15 | 2014-05-20 | Nara Logics, Inc. | Apparatus and method for providing harmonized recommendations based on an integrated user profile |
| CN109145087A (en) * | 2018-07-30 | 2019-01-04 | 大连理工大学 | A kind of scholar's recommendation and collaborative forecasting method based on expression study and competition theory |
| CN111368074A (en) * | 2020-02-24 | 2020-07-03 | 西安电子科技大学 | Link prediction method based on network structure and text information |
| CN112069306A (en) * | 2020-07-22 | 2020-12-11 | 中国科学院计算机网络信息中心 | Paper partner recommendation method based on author writing tree and graph neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112989199A (en) | 2021-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| McCabe et al. | A review of research into social tourism: Launching the Annals of Tourism Research Curated Collection on Social Tourism. | |
| Huai et al. | Which environmental features contribute to positive and negative perceptions of urban parks? A cross-cultural comparison using online reviews and Natural Language Processing methods | |
| Hu et al. | A natural language processing and geospatial clustering framework for harvesting local place names from geotagged housing advertisements | |
| CN107679661B (en) | Personalized tour route planning method based on knowledge graph | |
| Jochem et al. | Classifying settlement types from multi-scale spatial patterns of building footprints | |
| CN111125453B (en) | Opinion leader role identification method in social network based on subgraph isomorphism and storage medium | |
| CN113535974B (en) | Diagnostic recommendation method and related device, electronic equipment and storage medium | |
| CN106960044B (en) | Time perception personalized POI recommendation method based on tensor decomposition and weighted HITS | |
| CN116738066B (en) | Rural travel service recommendation method and device, electronic equipment and storage medium | |
| Hoyle et al. | Are neural topic models broken? | |
| Liu et al. | Recommending attractive thematic regions by semantic community detection with multi-sourced VGI data | |
| Ma et al. | Exploring the spatial distribution characteristics of emotions of weibo users in wuhan waterfront based on gender differences using social media texts | |
| CN116521936B (en) | Course recommendation method and device based on user behavior analysis and storage medium | |
| Westerholt et al. | On the way to platial analysis: Can geosocial media provide the necessary impetus? | |
| Zhou et al. | Familiarity-dependent computational modelling of indoor landmark selection for route communication: a ranking approach | |
| CN117312680A (en) | Resource recommendation method based on user-entity sub-graph comparison learning | |
| Ye et al. | Analyzing asymmetric city connectivity by toponym on social media in China | |
| Noorian | A BERT-based sequential POI recommender system in social media | |
| Díaz-Pacheco et al. | A comprehensive deep learning approach for topic discovering and sentiment analysis of textual information in tourism | |
| Li et al. | A knowledge-based approach for estimating the distribution of urban mixed land use | |
| BouMjahed et al. | Wired at birth: Childhood, technology engagement, and travel behavior | |
| Yao et al. | Unsupervised land-use change detection using multi-temporal POI embedding | |
| CN116910364A (en) | Tourist attraction recommendation method and system based on knowledge graph | |
| CN112989199B (en) | Cooperative network link prediction method based on multidimensional proximity attribute network | |
| Liang et al. | Enhancing scenic recommendation and tour route personalization in tourism using ugc text mining |
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 |