CN110442618B - Convolutional neural network review expert recommendation method fusing expert information association relation - Google Patents
Convolutional neural network review expert recommendation method fusing expert information association relation Download PDFInfo
- Publication number
- CN110442618B CN110442618B CN201910677191.8A CN201910677191A CN110442618B CN 110442618 B CN110442618 B CN 110442618B CN 201910677191 A CN201910677191 A CN 201910677191A CN 110442618 B CN110442618 B CN 110442618B
- Authority
- CN
- China
- Prior art keywords
- expert
- information
- expert information
- matrix
- experts
- 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
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
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Fuzzy Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a convolutional neural network review expert recommendation method fusing expert information association relations, and belongs to the technical field of data processing. The invention constructs an incidence relation matrix among a plurality of expert information, and splices the incidence relation matrix with the expert information vector converted into the vector to form an expert correlation matrix, and finally constructs an evaluation expert recommendation model fusing the expert incidence relation through a convolutional neural network. And training an expert recommendation model through the scoring relation between the items and the experts. The experimental result shows that the expert recommendation model provided by the invention achieves a better effect in an actual task, and the recommendation effect is improved to a certain extent compared with a method without considering the expert relation.
Description
Technical Field
The invention relates to a convolutional neural network review expert recommendation method fusing expert information association relations, and belongs to the technical field of data processing.
Background
In recent years, the activities of reporting and setting up scientific and technological projects in China are increasing day by day, the management of scientific and technological projects and talents at all levels in China also realizes an information system, and experts play a key role in review work all the time. In order to ensure the objectivity, fairness and fairness of the project evaluation work, the selection work of the evaluation experts is particularly important.
Expert Recommendation (Expert Recommendation) is based on rich Expert database information, and finds an Expert matched with a task to be recommended by using technologies such as data mining, machine learning and the like. From the task to be recommended, expert recommendation can be regarded as a special form of personalized recommendation based on content filtering. In the field of electronic commerce, recommendation algorithms and technologies are researched more, and therefore a sufficient theoretical basis is provided for expert recommendation.
Currently, in the research of recommendation methods, the following methods are mainly proposed, 1, a content-based method, 2, a collaborative filtering-based method, and 3, a deep learning-based method. In the content-based method, it is necessary to perform recommendation by mining the content characteristics of the item and the user and calculating the similarity using a machine learning method. The content-based recommendation method relies on characteristic information about user preference and items, does not need a large number of scoring records, and therefore does not have the problem of sparse scoring data. Meanwhile, for a new project, the user can be recommended only by performing feature extraction, the cold start problem of the new project is solved, and the problem of difficult feature extraction is often encountered. The main idea of the collaborative filtering-based method is to discover potential preferences of users for items by using a method with similar interest preferences among similar users. Collaborative filtering only requires the use of historical scoring data of the user and is therefore simple and effective. However, since the user has very little rating data for an item relative to the total number of items, the problem of data sparseness is often encountered, and in addition, for a new user or item, recommendation cannot be made because of no rating data, and there is a cold start problem. Deep learning does not need to design features manually like the traditional method, but how to integrate the association relationship between experts into a deep learning model is the second key problem.
In order to solve the problems, the invention provides a convolutional neural network review expert recommendation method for researching and fusing expert association relations.
Disclosure of Invention
The invention provides a convolutional neural network review expert recommendation method fusing expert information incidence relation, wherein a model constructed in the invention realizes matching degree score calculation between an expert and a project by means of vector representation of review project information and expert attribute information in an existing database through a convolutional neural network. In particular, information association relations among multiple experts are merged into the model so as to improve the recommendation performance of the model.
The technical scheme of the invention is as follows: the convolutional neural network review expert recommendation method fusing the expert information association relationship comprises the following specific steps:
step1, collecting expert information and project information, and converting the information of the experts and the project information into an expert information implicit vector and a project information implicit vector respectively through embedding;
in the Step1, as a preferred scheme of the present invention, the expert and project structured data information is represented as vector representation which can be identified and processed by a computer through one-hot coding, and is respectively mapped into dense hidden vectors through an embedded Embedding layer.
Step2, expressing the relation between the expert information into a correlation relation matrix between the expert information by using the thought of a Markov network;
as a preferred embodiment of the present invention, in Step2, the undirected edges of the markov network are used to construct correlations between expert information; firstly, determining the incidence relation between experts by using partial attribute information of the experts, and then calculating the correlation between the expert information by using a plurality of different incidence relations between the expert information so as to generate an expert information incidence relation matrix.
As a preferable aspect of the present invention, a calculation formula of the correlation between the expert information is as follows:
wherein if such relational features exist between the expert information, h m (e i ,e j ) =1, otherwise h m (e i ,e j ) =0; where M is the mth characteristic function and M is the total number of associated characteristics, M =4, λ in the present invention m The weight representing the corresponding characteristic function is,
denotes all and nodes e i Set of connected edges, eigenfunction weight λ m Estimating the data by adopting a maximum likelihood estimation method; when Sim (e) i ,e j ) Is greater than a given threshold value beta, the expert information e is considered i And expert information e j Is relevant, i.e. there is a undirected edge between the two nodes in the expert Markov network, the weight of the edge being Sim (e) i ,e j )。
Step3, merging the expert information hidden vector obtained in Step1 and the expert information incidence relation matrix in Step2, and combining the merged expert information hidden vector and the expert information incidence relation matrix into an expert information correlation matrix;
as a preferred embodiment of the present invention, in Step3, in order to fully utilize the association relationship between the attribute information of the expert and the expert to solve the expert recommendation problem for project review, it is considered to fuse the expert information association relationship matrix and the expert information vector, and to prepare for extracting the expert association features by convolution in the next Step.
And Step4, performing convolution and pooling operation on the expert information correlation matrix in Step3 through a convolution neural network, extracting relation characteristic hidden vectors in a plurality of experts, merging the output hidden vectors with the project information hidden vectors in Step1, inputting the merged matrix vectors into an MLP layer of the multilayer perceptron, and learning to obtain the matching degree score of the experts on the evaluation projects, thereby training the model.
As a preferred scheme of the present invention, in Step4, in order to fully utilize the association relationship between the attribute information of the experts and the experts to solve the problem of multi-expert recommendation for project review, a convolutional neural network is used to extract features from an expert correlation matrix in which expert information is fused. The main work of the MLP layer is to perform softmax operation on a vector obtained by combining an expert information hidden vector obtained by convolution and pooling and a project information hidden vector obtained only by Embedding (Embedding) layer, so that the model scores recommended experts.
The invention has the beneficial effects that:
1. the invention fuses the expert information incidence relation matrix in the network,
2. the CNN convolutional neural network is used for extracting the characteristics of the expert information relation matrix, so that the hidden relation existing among experts can be mined, and the accuracy of expert recommendation is improved.
3. The experimental result shows that the expert recommendation model provided by the invention achieves a better effect in an actual task, and the recommendation effect is improved to a certain extent compared with a method without considering the expert relation.
Drawings
FIG. 1 is a diagram of the correlation between the review expert data based on the Markov network in the invention;
fig. 2 is a flow chart of the convolutional neural network of the invention.
Detailed Description
Example 1: as shown in fig. 1-2, the convolutional neural network review expert recommendation method fusing the expert information association relationship specifically comprises the following steps:
step1, extracting expert structured data information in a talent expert management system of a science and technology hall in Yunnan province, converting project structured data information to be reviewed into binary sparse vectors through one-hot coding, and respectively mapping the binary sparse vectors into dense hidden vectors through Embedding an Embedding layer.
Step2, expressing the relation between the expert information into a correlation relation matrix between the expert information by using the thought of a Markov network;
before constructing the expert information incidence relation matrix, a Markov (Markov) network is used for constructing a relation graph between experts, and then the relation graph is used for generating the expert incidence relation matrix. As can be seen from FIG. 2, the expert information in the expert Markov network is represented by node e j It means that the correlation between the expert information constitutes a undirected edge whose weight of the edge depends on the correlation between the expert information. In the expert Markov network, the correlation between experts needs to be calculated by fusing the association relation of the experts. In order to fuse the expert information incidence relations of the types, the expert information incidence relations are defined as a characteristic function h m (e i ,e j ) And assigning a weight λ to each eigenfunction m And fusing the correlation characteristics by using a logarithmic linear model. Assume correlation between expert information Sim (e i ,e j ) Expressed, the correlation calculation formula is:
wherein if such relational features exist between the expert information, h m (e i ,e j ) =1, otherwise h m (e i ,e j ) And =0. Where M is expressed as the mth characteristic function and M is the total number of associated characteristic features, M =4, λ m The weight representing the corresponding characteristic function is,
denotes all and nodes e i A set of connected edges. Characteristic function weight lambda m And estimating the data by adopting a maximum likelihood estimation method. When Sim (e) i ,e j ) Is greater than a given threshold value beta, the expert information e is considered i And expert information e j Is relevant, i.e. in expert Markov networksA non-directional edge exists between the two nodes, and the weight of the edge is Sim (e) i ,e j ). An expert relationship matrix is constructed according to the above method.
TABLE 1 characteristics of expert information associations
Step3, merging the expert information hidden vector obtained in Step1 and the expert information incidence relation matrix in Step2, and combining the merged expert information hidden vector and the expert information incidence relation matrix into an expert information correlation matrix;
as a preferred scheme of the present invention, in Step3, in order to fully utilize the association relationship between the attribute information of the expert and the expert to solve the expert recommendation problem for project review, the expert information association relationship matrix and the expert information vector are considered to be fused, so as to prepare for extracting the expert association features in the next convolution.
Specifically, n expert vectors corresponding to the project are mapped through an Embedding layer and then are vertically connected into a matrix of n × k, wherein k is the dimension of the hidden expert vector. And then horizontally connecting and combining the obtained matrix with the size of n multiplied by k and the expert information correlation matrix with the size of n multiplied by n of the experts to form a new expert joint matrix.
And Step4, performing convolution and pooling operation on the expert information correlation matrix in Step3 through a convolution neural network, extracting relation characteristic hidden vectors in a plurality of experts, merging the output hidden vectors with the project information hidden vectors in Step1, inputting the merged matrix vectors into an MLP layer of the multilayer perceptron, and learning to obtain the matching degree score of the experts on the evaluation projects, thereby training the model.
The construction process is as follows:
a first layer: an input layer: the expert association matrix fusing the expert correlation information obtained in Step2 and the item information vector obtained in Step1 are used as input.
A second layer: convolutional layer whose main task is to extract from expert associative matrices using convolutional neural networksAnd (5) characterizing. Let x i ∈R n An n-dimensional vector representing the ith expert in the expert joined matrix, the matrix containing m experts can be represented as:
wherein it is present>
Indicating a connection operation, e.g. </or >>
The convolution operation refers to a sliding window of length h from x 1:m And intercepting the vector as the input of the activation function of the convolution layer, and calculating the activation function to obtain the output which is the feature to be extracted. As shown in the following formula: c. C i ＝f(W·x i:i+h-1 + b). Where the function f represents an activation function, the ReLU function is chosen herein as the activation function. W and b represent a weight matrix and a bias unit, are parameters of the neural network and can be obtained by training the neural network. c. C i Representing a function according to an activation function f and an input vector x i:i+h-1 The calculated features. The above formula describes the sliding window of length h at position x i:i+h-1 The above-described operations, when the sliding window slides from the beginning to the end of a sentence, result in a set of inputs, denoted as (x) 1:h ,x 2:1+h ,...,x m-h+1:m ) Further, a set of outputs, denoted as (c), may be derived from the input 1 ,c 2 ,...,c m-h+1 ) This set of outputs is called a feature map.
And a third layer: and a max-posing layer, wherein the main purpose of the max-posing layer is to process the feature mapping of the previous layer to obtain a final feature vector. In particular, the maximum values in the feature maps are extracted, the idea being to consider that the maximum value in each feature map represents the most important feature in the map. Furthermore, one of the main advantages of the max-posing layer is that a fixed-length feature vector can be obtained by means of the operation of this layer without concern for the length of the original input vector.
Fourth step ofLayer (b): and the output layer is mainly used for connecting and learning the expert information hidden vectors and the project information hidden vectors obtained through convolutional neural network learning and outputting a score result. The layer is an MLP layer, a Relu function is used as an activation function, and the probability is obtained by performing softmax processing on the result. Assuming that the number of softmax layer classes, i.e. the number of neurons, is k, the expression is
Wherein z represents an input vector of the softmax layer, W represents a parameter of the softmax layer network, y i And the output value of the ith neuron of the output layer is represented, and the final output result is a probability value that the k-dimensional vector represents k categories. The output layer of the model determines the number of neurons according to the number of classes of the classification task. In order to obtain the probability values of the scores of the n experts for the same item, n softmax are used in the last layer of the MLP layer to output the score results corresponding to the n experts.
In addition, in order to verify the effectiveness of the expert information association, and in view of the particularity of the experimental study problem, a matrix not including the expert information association is used as an input of a model as a comparison experiment, the model is denoted as M1, and the convolutional neural network recommendation model fusing the expert information association, which is proposed herein, is denoted as M2. The two models are respectively trained by using the same training set, and the average NDCG values of the same five experts recommended to a single project by the two different models are respectively obtained by using the same test set, as shown in the following table 2, wherein K is the size of each type of attribute after Embedding in the representation of the expert or the attribute of the content of the project
Table 2 shows the results of comparative tests
| K | M1 | M2 |
| 8 | 0.931 | 0.934 |
| 16 | 0.935 | 0.945 |
| 32 | 0.942 | 0.948 |
As can be seen from table 2, the model using the expert information association as the auxiliary input achieves a better effect than the model not using the expert information association as the auxiliary input. Experimental results prove that the method achieves a good effect on the NDCG index sorted by the recommendation system.
While the present invention has been described in detail with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, and various changes can be made without departing from the spirit of the present invention within the knowledge of those skilled in the art.
Claims (3)
1. The convolutional neural network review expert recommendation method fusing the expert information association relationship is characterized by comprising the following steps of: the method comprises the following specific steps:
step1, collecting expert information and project information, and converting the information of the expert information and the project information into an expert information hidden vector and a project information hidden vector respectively through embedding;
step2, expressing the relation between the expert information into a correlation relation matrix between the expert information by using the thought of a Markov network;
step3, merging the expert information hidden vectors obtained in Step1 and the expert information incidence relation matrix in Step2, and combining the merged expert information hidden vectors into an expert information correlation matrix;
step4, performing convolution and pooling operation on the expert information correlation matrix in Step3 through a convolution neural network, extracting relation characteristic hidden vectors in a plurality of experts, merging the output hidden vectors with the project information hidden vectors in Step1, inputting the merged matrix vectors into an MLP layer of the multilayer perceptron, and learning to obtain the matching degree score of the experts on the evaluation projects, thereby training a model;
in the Step1, the expert and project structured data information is represented as vector representation which can be identified and processed by a computer through one-hot coding, and the vector representation is respectively mapped into dense hidden vectors through an Embedding layer;
in Step2, the undirected edges of the Markov network are used for constructing the correlation between the expert information; firstly, determining the incidence relation between experts by using partial attribute information of the experts, and then calculating the correlation between the expert information by using various different incidence relations between the expert information so as to generate an expert information incidence relation matrix;
the calculation formula of the correlation between the expert information is as follows:
wherein if such a relational feature exists between the expert information, h m (e i ，e j ) =1, otherwise h m (e i ，e j ) =0; wherein M is represented as the mth characteristic function, M is the total number of the associated relation characteristics, M =4, λ m A weight representing a function of the corresponding feature,
denotes all and nodes e i Set of connected edges, eigenfunction weight λ m Estimating the data by adopting a maximum likelihood estimation method; when Sim (e) i ，e j ) Is greater than a given threshold value beta, then the system is considered to be specializedFamily information e i And expert information e j Is relevant, i.e. there is a undirected edge between the two nodes in the expert Markov network, the weight of the edge being Sim (e) i ，e j )。
2. The expert information association fused convolutional neural network review expert recommendation method of claim 1, wherein: in Step3, in order to fully utilize the association relationship between the attribute information of the expert and the expert to solve the expert recommendation problem aiming at project review, the expert information association relationship matrix and the expert information vector are considered to be fused, and preparation is made for extracting the expert association characteristics in the next convolution.
3. The expert information association fused convolutional neural network review expert recommendation method of claim 1, wherein: in Step4, in order to fully utilize the incidence relation between the attribute information of the experts and the experts to solve the problem of multi-expert recommendation for project review, a convolutional neural network is utilized to extract features from an expert correlation matrix integrating the expert information, and the main work of the MLP layer is to perform softmax operation on the merged vector of the expert information hidden vectors obtained by convolution and pooling operation and the vector of the project information hidden vectors obtained only by Embedding the Embedding layer, so that the model scores recommendation experts.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910677191.8A CN110442618B (en) | 2019-07-25 | 2019-07-25 | Convolutional neural network review expert recommendation method fusing expert information association relation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910677191.8A CN110442618B (en) | 2019-07-25 | 2019-07-25 | Convolutional neural network review expert recommendation method fusing expert information association relation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110442618A CN110442618A (en) | 2019-11-12 |
| CN110442618B true CN110442618B (en) | 2023-04-18 |
Family
ID=68431499
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910677191.8A Active CN110442618B (en) | 2019-07-25 | 2019-07-25 | Convolutional neural network review expert recommendation method fusing expert information association relation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110442618B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111782797A (en) * | 2020-07-13 | 2020-10-16 | 贵州省科技信息中心 | Automatic matching method for scientific and technological project review experts and storage medium |
| CN112100370B (en) * | 2020-08-10 | 2023-07-25 | 淮阴工学院 | Picture-trial expert combination recommendation method based on text volume and similarity algorithm |
| CN113343666B (en) * | 2021-06-29 | 2023-07-14 | 深圳前海微众银行股份有限公司 | Method, device, equipment and storage medium for determining confidence of score |
| CN113516094B (en) * | 2021-07-28 | 2024-03-08 | 中国科学院计算技术研究所 | System and method for matching and evaluating expert for document |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103631859A (en) * | 2013-10-24 | 2014-03-12 | 杭州电子科技大学 | Intelligent review expert recommending method for science and technology projects |
| CN107330115A (en) * | 2017-07-12 | 2017-11-07 | 广东工业大学 | A kind of information recommendation method and device |
| CN109062961A (en) * | 2018-06-27 | 2018-12-21 | 淮阴工学院 | A kind of expert's combination recommended method of knowledge based map |
| CN109241424A (en) * | 2018-08-29 | 2019-01-18 | 陕西师范大学 | A kind of recommended method |
| CN109308315A (en) * | 2018-10-19 | 2019-02-05 | 南京理工大学 | A kind of collaborative recommendation method based on specialist field similarity and incidence relation |
| CN109657140A (en) * | 2018-12-13 | 2019-04-19 | 平安普惠企业管理有限公司 | Information-pushing method, device, computer equipment and storage medium |
| CN109902201A (en) * | 2019-03-08 | 2019-06-18 | 天津理工大学 | A kind of recommended method based on CNN and BP neural network |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3480714A1 (en) * | 2017-11-03 | 2019-05-08 | Tata Consultancy Services Limited | Signal analysis systems and methods for features extraction and interpretation thereof |
| CN108647206B (en) * | 2018-05-04 | 2021-11-12 | 重庆邮电大学 | Chinese junk mail identification method based on chaos particle swarm optimization CNN network |
| CN108804689B (en) * | 2018-06-14 | 2020-10-16 | 合肥工业大学 | Question-answering platform-oriented label recommendation method integrating user hidden connection relation |
-
2019
- 2019-07-25 CN CN201910677191.8A patent/CN110442618B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103631859A (en) * | 2013-10-24 | 2014-03-12 | 杭州电子科技大学 | Intelligent review expert recommending method for science and technology projects |
| CN107330115A (en) * | 2017-07-12 | 2017-11-07 | 广东工业大学 | A kind of information recommendation method and device |
| CN109062961A (en) * | 2018-06-27 | 2018-12-21 | 淮阴工学院 | A kind of expert's combination recommended method of knowledge based map |
| CN109241424A (en) * | 2018-08-29 | 2019-01-18 | 陕西师范大学 | A kind of recommended method |
| CN109308315A (en) * | 2018-10-19 | 2019-02-05 | 南京理工大学 | A kind of collaborative recommendation method based on specialist field similarity and incidence relation |
| CN109657140A (en) * | 2018-12-13 | 2019-04-19 | 平安普惠企业管理有限公司 | Information-pushing method, device, computer equipment and storage medium |
| CN109902201A (en) * | 2019-03-08 | 2019-06-18 | 天津理工大学 | A kind of recommended method based on CNN and BP neural network |
Non-Patent Citations (4)
| Title |
|---|
| A Method to Review Expert Recommendation Using Topic Relevance and Expert Relationship;Shengxiang Gao et al.;《International Journal of Cooperative Information Systems》;20170526;第26卷(第2期);正文第5页第2.2.2节 * |
| Review Expert Collaborative Recommendation Algorithm Based on Topic Relationship;Shengxiang Gao et al.;《IEEE/CAA JOURNAL OF AUTOMATICA SINICA》;20151012;第2卷(第4期);403-411 * |
| 基于主题信息的项目评审专家推荐方法;余峰 等;《计算机工程》;20140615;第40卷(第6期);201-205 * |
| 融合关联关系的卷积神经网络项目评审专家推荐方法研究;何孝胥;《中国优秀硕士学位论文全文数据库 信息科技辑》;20210815(第8期);I138-649 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110442618A (en) | 2019-11-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110442618B (en) | Convolutional neural network review expert recommendation method fusing expert information association relation | |
| CN108920720B (en) | Large-scale image retrieval method based on depth hash and GPU acceleration | |
| CN105912611B (en) | A kind of fast image retrieval method based on CNN | |
| CN108665323B (en) | Integration method for financial product recommendation system | |
| CN110390017B (en) | Target emotion analysis method and system based on attention gating convolutional network | |
| JP6378855B1 (en) | Image search system, image search method and program | |
| CN110222218B (en) | Image retrieval method based on multi-scale NetVLAD and depth hash | |
| CN112819023B (en) | Sample set acquisition method, device, computer equipment and storage medium | |
| CN109948125B (en) | Method and system for improved Simhash algorithm in text deduplication | |
| CN105809473B (en) | Training method for matching model parameters, service recommendation method and corresponding device | |
| CN108897791B (en) | Image retrieval method based on depth convolution characteristics and semantic similarity measurement | |
| WO2018133596A1 (en) | Continuous feature construction method based on nominal attribute | |
| CN104239553A (en) | Entity recognition method based on Map-Reduce framework | |
| Ding et al. | A Hybrid Feature Selection Algorithm Based on Information Gain and Sequential Forward Floating Search① | |
| CN110851176B (en) | Clone code detection method capable of automatically constructing and utilizing pseudo-clone corpus | |
| CN107145519B (en) | Image retrieval and annotation method based on hypergraph | |
| CN110443574B (en) | Recommendation method for multi-project convolutional neural network review experts | |
| CN110737805B (en) | Method and device for processing graph model data and terminal equipment | |
| CN112396492A (en) | Conversation recommendation method based on graph attention network and bidirectional long-short term memory network | |
| CN111539444A (en) | Gaussian mixture model method for modified mode recognition and statistical modeling | |
| CN114332519A (en) | Image description generation method based on external triple and abstract relation | |
| CN110569355B (en) | Viewpoint target extraction and target emotion classification combined method and system based on word blocks | |
| CN116680363A (en) | Emotion analysis method based on multi-mode comment data | |
| CN115204967A (en) | Recommendation method integrating implicit feedback of long-term and short-term interest representation of user | |
| JP2019086979A (en) | Information processing device, information processing method, and program |
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 |