CN111915216B - Open source software project developer recommendation method based on secondary attention mechanism - Google Patents

Open source software project developer recommendation method based on secondary attention mechanism Download PDF

Info

Publication number
CN111915216B
CN111915216B CN202010818089.8A CN202010818089A CN111915216B CN 111915216 B CN111915216 B CN 111915216B CN 202010818089 A CN202010818089 A CN 202010818089A CN 111915216 B CN111915216 B CN 111915216B
Authority
CN
China
Prior art keywords
project
team
developers
developer
attention mechanism
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
Application number
CN202010818089.8A
Other languages
Chinese (zh)
Other versions
CN111915216A (en
Inventor
潘国盛
姚远
徐锋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University
Original Assignee
Nanjing University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing University filed Critical Nanjing University
Priority to CN202010818089.8A priority Critical patent/CN111915216B/en
Publication of CN111915216A publication Critical patent/CN111915216A/en
Application granted granted Critical
Publication of CN111915216B publication Critical patent/CN111915216B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06311—Scheduling, planning or task assignment for a person or group
    • 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
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06311—Scheduling, planning or task assignment for a person or group
    • G06Q10/063112—Skill-based matching of a person or a group to a task
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06313—Resource planning in a project environment

Abstract

The application discloses an open source software project developer recommendation method based on a secondary attention mechanism, which is used for modeling project team features and project features based on relationships among project team members, relationships between project teams and projects, text description information of the projects and the like. First, text features of developers and projects are obtained through web representation learning and text representation learning. The relative weights of the team's existing developers about the project team are then learned using a first tier of attention mechanism, thereby obtaining features of the project team. The relative weights of the project team and the project document with respect to the project are then learned using a second level of attention mechanism, thereby obtaining characteristics of the project population. And finally, calculating the similarity between the overall characteristics of the project and the characteristics of the developers, and recommending proper developers for the open-source software project according to the similarity sequence.

Description

Open source software project developer recommendation method based on secondary attention mechanism
Technical Field
The application relates to an open source software project developer recommendation method based on a secondary attention mechanism.
Background
The software projects which are developed by depending on the open source platform are more and more, more and more developers begin to pay attention to the software projects in the open source platform, and a plurality of high-quality and large-scale open source software projects such as VSCode, flutter, tensorflow are generated. According to the 2019 report published by Github, there are currently over 4000 thousands of developer users worldwide, and 1000 thousands of new developers are added in 2019 one year, and 130 ten thousands of developers make the first contribution to open sources.
For an open source software project, some existing team members exit the current project development for various reasons during the project development process, and due to the evolution of related dependency technologies, the addition of developers familiar with related fields is also required. Therefore, it is important to continuously add new developers suitable for the current project and team, and to continuously add new developers of good quality to the project. However, recommending proper development novice for open source software projects presents a significant challenge due to the large number of developers in the open source community and the great complexity of the areas, programming languages, and projects experienced by the developers. Many studies have thus been made of recommendation tools for recommending new members to an open source software project team.
The research of the recommendation problem of the open source software project needs to consider the following two problems:
(1) Interactive modeling of project teams with candidate developers,
(2) Modeling the interaction of the project document and the candidate developer;
for interactive modeling between candidate developers and open source software projects, due to the characteristics of team development of the open source software projects, modeling is not only needed for matching degree between project tasks and attributes of the developers in the process of selecting the candidate developers, but also similarity modeling between the candidate developers and existing team members is needed to be added in the process of recommending the candidate developers in order to reduce the cost of running-in and difficulty of communication after the developers are added. When recommending candidate developers, team documents and project teams may have different impact weights, and similarly when taking a project team as a selection basis, the impact weights of team members on candidate developer similarity analysis are also different. Existing methods are based on relatively fixed patterns when dealing with both types of interactions. In processing project team-to-project document relationships, stitching is typically employed, thus omitting modeling of the project document and the inter-team impact weights. In dealing with project team and candidate developer relationships, a member of the team is typically taken as a proxy, ignoring modeling of all members of the entire team and the relative weights between members.
Disclosure of Invention
The application aims to: the existing developer recommendation method often does not consider or does not consider the relative weights between candidate developers and existing team members of the open source software project and the relative relationships between project documents and project teams when the interactive modeling between the candidate developers and the existing open source software project is considered. Aiming at the problems and defects existing in the prior art, the application provides an open source software project developer recommendation method based on a secondary attention mechanism, which comprehensively considers the relationship among project team members, the relationship between project team and project, text description information of the project and the like, and solves the problem of developer recommendation according to the relationship structural characteristics of an open source platform. The following problems need to be considered in the technical scheme:
(1) How the relative relationships between project team members are considered in constructing the project team features;
(2) How to consider the relative relationship between project documents and project teams when building project ensemble features;
for problem (1), because there may be different branches in a single project team, the candidate developer does not have to have a high match with all developers in the team, and the degree of matching of some team members will affect the similarity between the final team and the candidate developer, and thus the matching between the final project and the candidate developer. The application utilizes a first-layer attention mechanism to automatically learn a weight for different developers in the team when interacting with the current candidate developer for influencing and calculating the characteristic representation of the project team.
To solve problem (2), then the feature representation of the candidate developer needs to interact with both the project itself document feature representation and the feature representation of the project team derived from (1). Because the project itself and the project team can possibly have important influence on candidate developer selection, the application adopts a second-layer attention mechanism to learn a weight coefficient for the project characteristic representation and the team characteristic representation, and carries out weighted summation on the project characteristic representation and the team characteristic representation, and the obtained characteristic representation is used as a final characteristic representation to interact with candidate developers to obtain a matching score.
The application utilizes a multi-layer attention mechanism and provides a recommendation model DETEX (Developer Team Expansion in Open Source Platforms, a team expansion model of an open source platform developer) based on a secondary attention mechanism. Based on the relationship among the project team members, the relationship between the project team and the project, text description information of the project and the like, modeling project team features and project features, and considering the similarity between candidate developers and open source software projects. Based on the assumption that the current team member is more suitable for the current development project than other developers, the team member in the current project is regarded as a positive example, the other developers are regarded as a negative example, the problem is converted into a predicted problem for the current member in the team, the DETEX model parameters are trained on the basis, and the model output obtained according to the parameters is used as the matching degree to be ordered.
Experiments on real data show that the method provided by the application has a remarkable improvement on the accuracy of matching team expansion with candidate developers compared with the existing method.
The technical scheme is as follows: a project developer recommendation method based on a secondary attention mechanism is disclosed, which comprises training a project developer recommendation model by utilizing attribute network data composed of existing open source platform projects and developers, recommending and sorting given candidate developers according to project document information and project member information, modeling project team features and project features by utilizing the secondary attention mechanism based on the relation between project team members, the relation between project team and projects and text description information of the projects, and finally recommending according to the calculated matching degree ordering of the developers and the projects, wherein the recommendation model adopted by the method mainly comprises the following contents:
1) Modeling the relative weight relationship between the existing members of the project team through a first layer of attention mechanism to obtain project team features;
2) The relative weight relationship between the project team and the project document is modeled by a second level of attention mechanism to obtain project overall features.
Using P to represent open source software project set, D to represent developer set, P E P to represent project to be expanded, T for project P p Representing a current set of development team members for a software project. The goal of the method is to find a developer D e D corresponding to the suitable joining of the open source software project p,
(1) Modeling the relative relationship between the existing members of the team through a first layer of attention mechanism to obtain the characteristics of the project team;
and simultaneously, taking the characteristic representation of the candidate developer, the characteristic representation of the open source software project and the skill attribute characteristic representation of the current project team member as inputs, and inputting the inputs into the DETEX model. Firstly, inputting a characteristic representation of a developer into a nonlinear layer of the DETEX model, and inputting a characteristic representation of a software item into the nonlinear layer of the DETEX model, wherein the formula is as follows
v d ＝μ(W d d+b d )
Wherein W is d 、b d For the network weight parameters and bias terms, d represents the feature representation obtained through network representation learning, v d Representing the output of the nonlinear layer, μ represents the activation function, here using a leakage correction linear element (LeakyReLU) with a negative slope of 0.01, the activation function curve is shown in fig. 1.
For the feature representation of the team, a convergence operation is performed on the feature representation of the developer in the team, and the feature representation is represented as follows:
wherein a is T Is a network parameter, t i Representing the current team of software projects T p The developer's members of the family,representing developer t i Feature representation output in non-linear layer, v d Representing the characteristic representation of the candidate developer output in the nonlinear layer, +.. In the formula, the interaction between the candidate developer and each member in the existing team of the open source software project is modeled, and different influence is given to different members in the team by a concentration mechanism +.>That is, the degree of influence that members within a team have on it is different for different candidate developers.
(2) The relative relationship between the team and the project document is modeled by a second level of attention mechanism to obtain the overall characteristics of the project.
For the project document, obtaining a characteristic representation of the project document by a word vector average method, and inputting the characteristic of the project document into a nonlinear layer:
v p ＝μ(W p p+b p )
wherein W is p 、b p For the weight parameters and bias terms, p represents a feature representation of the project document obtained by word vector means. Interactions between candidate developers and project teams will be modeled next. To model the preference relationships between the current team of open source project documents and projects for different candidate developers at the time of selection, the attention mechanism described by the following formula is employed:
wherein, the liquid crystal display device comprises a liquid crystal display device,v p 、/>v t respectively different neural network layer parameters alpha p And alpha t Representing the weights of the open source project and team, and thus obtaining a characteristic representation v representing the entirety of the project and team comb ：
v comb ＝α p v p +α t v t
According to v comb The overall similarity between candidate developers and open source software projects is predicted through a multi-layer perceptron structure, and the formula is as follows:
wherein MLP stands for multi-layer perceptron,after the last layer of MLP, performing a sigmoid operation to make the final output value range be [0,1 ]]Is a similarity of (3).
Drawings
FIG. 1 is a graph of an activation function;
FIG. 2 is a block diagram of the DETEX model in an embodiment of the application;
FIG. 3 is the results of DETEX under the HR criteria;
FIG. 4 is the result of DETEX under the nDCG index;
fig. 5 is a flow chart of the method of the present application.
Detailed Description
The present application is further illustrated below in conjunction with specific embodiments, it being understood that these embodiments are meant to be illustrative of the application and not limiting the scope of the application, and that modifications of the application, which are equivalent to those skilled in the art to which the application pertains, fall within the scope of the application defined in the appended claims after reading the application.
The open source software project developer recommendation method based on the secondary attention mechanism introduces the relative relation between project team members and between project team and project document based on the pre-training developer characteristic representation and the project document characteristic representation. Wherein the first layer of attention mechanisms models the relative relationships between members of the project team to make up a feature representation of the project team and the second layer of attention mechanisms is used to model the relative relationships between the project team and the project document to make up a feature representation of the project. And finally, obtaining the final similarity and the recommendation result through the interaction between the project overall feature representation and the candidate developer.
Using P to represent open source software project set, D to represent developer set, P E P to represent project to be expanded, T for project P p Representing a current set of development team members for a software project. The goal of the method is to find a developer D e D corresponding to the suitable joining of the open source software project p,
(1) Modeling the relative relationship between the existing members of the team through a first layer of attention mechanism to obtain the characteristics of the project team;
as shown in fig. 2, to model existing relationships, to calculate the similarity between candidate developers and a given open source software project, it is necessary to input the candidate developer's representation, the open source software project's representation, and the skill attribute representation of the current project team member as inputs simultaneously to the DETEX model. In order for DETEX to possess the ability to better express complex interactions, the ability to encode nonlinear relationships, a developer's feature representation of a software project is first entered into a nonlinear layer (E in FIG. 2 p And E is d ) The formula is as follows
v d ＝μ(W d d+b d )
Where d represents the feature representation obtained by network representation learning and μ represents the activation function, here we use a leak relu with a negative slope of 0.01.
For team feature representation, since multiple developers are included in each open project team, the feature representations of the developers in the team are put together (E in FIG. 2 t ) The expression is as follows:
wherein t is i Representing the current team of software projects T p The developer's members of the family,representing the developer t i The characteristics output in the nonlinear layer represent, and the element-wise multiplication is indicated. In this formula, interactions of candidate developers with each member in the existing team of the open source software project are modeled, and different influence is given to different members in the team through a concentration mechanism, namely, the matching degree with the same team member is different for different candidate developers, and the final influence on the matching degree of the team is different.
(2) The relative relationship between the team and the project document is modeled by a second level of attention mechanism to obtain the overall characteristics of the project.
For the project document, the characteristic representation of the project document is obtained by a word vector average method, and the characteristic of the project document is input into a nonlinear layer:
v p ＝μ(W p p+b p )
where p represents the project document representation obtained by word vector mean. We will next model interactions between candidate developers and projects and project teams. To model the preference relationships between the current team of open source project documents and projects for different candidate developers at the time of selection, we employ the attention mechanism described by the following formula:
wherein alpha is p And alpha t Representing the weights of the open source project and team, so we can get a representation v representing the project and team as a whole comb ：
v comb ＝α p v p +α t v t
According to v comb The overall similarity between candidate developers and open source software projects can be predicted through a multi-layer perceptron structure, and the formula is as follows:
wherein MLP stands for multi-layer perceptron,after the last layer of MLP, performing a sigmoid operation to make the final output value range be [0,1 ]]Is a similarity of (3).
The implementation flow of the method is shown in fig. 5. The method firstly inputs a pattern network of { project-developer } of the open source software project as input, and trains a recommendation model of the open source software project developer. And after model training is completed, the recommended project document information and the existing team member information are input into the model for prediction. The model will calculate the relative relationships between project members and the relative relationships of the project team to the project document according to the secondary attentiveness mechanism proposed by the present application. And calculating the matching degree ranking of the candidate developers about the recommended items according to the weight coefficient of the relative relation.
The DETEX model training method comprises the following steps: the DETEX model training method treats members of the current project team as positive examples and other developers not in the team as negative examples based on the assumption that the current team members are better suited for the current development project than other developers. For the assumption to be more consistent with the real situation of the data, only software projects of at least 5 star numbers and 5 developers and developers participating in at least 5 projects are reserved in the initial screening of training data, so that the projects of the projects and the superiority of the developers are ensured. The trained model parameters are more reasonable.
On model training, the optimization problem is converted into a classification problem, and the following cross entropy objective function is optimized:
where p is the set of software items in the training set, T p Project p the set of developers in the current project team, s p.d And (3) withRepresenting the similarity of the true matching tag and the prediction, respectively, and σ represents the sigmoid function. By adjusting network parametersThe numbers are used to maximize the cross entropy function for training purposes.
To train the DETEX model using the above formula, each existing developer in each project development team is set to be positive with the label for that project p.d =1. Meanwhile, for each positive example, randomly selecting a negative example, and setting the label as s p.d ＝0。
Experiment setting: in the aspect of setting set generation, a leave-one-out method is adopted, namely, one developer is removed from the development team member set of each open source software project to serve as a test set positive example. Meanwhile, because sequencing tests are very time-consuming in all developers, 100 developers which are not in the project development team are randomly selected for each positive example to serve as negative examples on the premise of not losing generality. The remaining developers will compose a training set for training the model.
Evaluation index: two indexes of Hit Rate (HR) and nDCG are selected to evaluate the DETEX performance, and the calculation formulas of HR and nDCG are as follows:
wherein hit t E {0,1} is 1 when the similarity rank of the positive case developer in the test set is less than or equal to K, is 0 when the similarity rank is greater than K, and is r t E {1, 2..k } represents the ranking of positive example developers in the test set. When the ranking of the positive example developer is higher, the two indexes are larger, which represents that the performance of the tested method is better, and K is set to be 1,5, 10 and 20 in the experiment.
Experimental data: the open source software platform Github data and the programming questioning and answering community StackOverflow data are adopted. For the Giuhub data, in order to meet experimental assumptions, the training quality is guaranteed, and fewer than 5 developers participate in the project and fewer than 5 stars and software projects in which 5 developers participate are filtered out. Duplicate entries for fork are also deduplicated. For the StackOverflow data, 400 high frequency occurrence skill tags were manually screened. The statistics associated with the other data are shown in table 1.
The comparison method comprises the following steps: according to the modeling of the application, the expansion of a software development team is a one-class recommendation problem, so the following four methods are chosen to compare with the method of the application:
(1) BPR, a pair-wise recommendation method;
(2) NCF, a method for modeling interaction between an article and a user by using a neural network;
(3) The TECE adds interactive modeling between the candidate developer and the team leader on the basis of NCF;
(4) tBPR, on the basis of BPR, adds modeling for project team.
Wherein BPR and NCF are used for team expansion modeling by using a recommendation system related method, and TECE and tBPR are added with modeling for the current project team on the basis of traditional user-item interaction.
Experimental results: first, we directly compare the DETEX model method with these comparison methods, and fig. 3 and fig. 4 show the experimental results of HR and nccg, respectively. Compared with the comparison methods, the method provided by the application has obvious improvement under two indexes. For example, compared with the comparison method TECE with the best experimental results, the HR and nDCG are respectively improved by 20.2% -77.2% and 43.0% -77.2%. There are two main reasons for the significant improvement. First, from a network representation learning perspective, we build and use a representation containing skill information for each developer, compared to existing methods; second, from a model level, the model modeling candidate developer of the present application interacts with each member of the software team's existing development team and uses the attention mechanism to give relative weight between the developer and project and team.
We have also tried to investigate the improvement that our DETEX model alone brings to the final effect. Since the model employed by the existing method has no step of acquiring the pre-trained feature representation based on the network representation learning, the skill representation obtained by the network representation learning method is input to the model of the existing method, and the results of hr@10 and ncdg@10 are shown in table 2. First, it can be seen that, with the addition of skill representation, the results of all comparison methods are significantly improved, with NCF being improved by 14.4% and 19.3% over the original hr@10 and nDCG@10, respectively. In addition, it can be observed that the DETEX model employed in the present application still has better performance than other comparison methods. For example, there is still an 11.6% improvement over TECE with skill representation added, at nccg@10. Such results indicate that both the representation of skill information and modeling for teams in DETEX bring a useful boost to the final recommendation.
Table 1 data statistics for examples
Number of open source software items 6599
Number of developers 11931
Average developer number per project 8.94
Average number of participating projects per developer 5.10
Number of StackOverflow developer 123214
Number of co-developers 879
Number of StackOverflow problem skill labels 400
Number of StackOverflow questions 53566
Table 2 addition shows experimental results after learning

Claims (1)

1. A method for recommending open source software project developers based on a secondary attention mechanism is characterized by comprising the following steps: training a project developer recommendation model by utilizing attribute network data formed by the existing open source platform project and the developer, recommending and sorting given candidate developers according to project document information and project member information, modeling project team features and project features by utilizing a secondary attention mechanism based on the relation among project team members, the relation between the project team and the project and the text description information of the project, and finally recommending according to the calculated matching degree sorting of the developers and the project, wherein the recommendation model adopted by the method mainly comprises the following contents:
1) Modeling the relative weight relationship between the existing members of the project team through a first layer of attention mechanism to obtain project team features;
2) Modeling a relative weight relationship between the project team and the project document through a second layer of attention mechanism to obtain a project overall feature;
introducing developer characteristic representation by using a network representation learning method, then learning a relative weight relation between the existing members of the project team through a first layer of attention mechanism, and obtaining the representation of the project team characteristic by using the weight; firstly, inputting characteristic representations of developers and software items into a nonlinear layer, wherein the formulas are as follows:
v d ＝μ(W d d+b d )
wherein W is d 、b d Respectively representing a network weight parameter and a bias term, d represents a characteristic representation obtained through network representation learning, and mu represents an activation function;
for the feature representation of the team, a convergence operation is performed on the feature representation of the developer in the team, and the feature representation is represented as follows:
wherein t is i Representing the current team of software projects T p Developer member in v d Representing the feature representations output by the candidate developer in the non-linear layer,representing developer t i The characteristic representation output in the nonlinear layer, +.; in the formula, interactions between candidate developers and each member in the existing team of the open source software project are modeled, and different influence is given to different members in the team through an attention mechanism, namely, the influence degree of the members in the team on the different candidate developers is different;
modeling the relative weight relationship between the team and the project document through a second layer of attention mechanism to obtain the overall characteristics of the project;
for the project document, obtaining a characteristic representation of the project document by a word vector average method, and inputting the characteristic of the project document into a nonlinear layer:
v p ＝μ(W p p+b p )
wherein W is p And b p Respectively representing weight parameters and bias items of a network, wherein p represents characteristic representation of a project document obtained through word vector average; interactions between candidate developers and projects and project teams are then to be modeled; to model the preference relationships between the current team of open source project documents and projects for different candidate developers at the time of selection, the attention mechanism described by the following formula is employed:
wherein alpha is p And alpha t Representing the weights of the open source project and team, and thus obtaining a characteristic representation v representing the entirety of the project and team comb ：
v comb ＝α p v p +α t v t
According to v comb The overall similarity between candidate developers and open source software projects is predicted through a multi-layer perceptron structure, and the formula is as follows:
wherein MLP stands for multi-layer perceptron,after the last layer of MLP, performing a sigmoid operation to make the final output value range be [0,1 ]]Is a similarity of (3).
CN202010818089.8A 2020-08-14 2020-08-14 Open source software project developer recommendation method based on secondary attention mechanism Active CN111915216B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010818089.8A CN111915216B (en) 2020-08-14 2020-08-14 Open source software project developer recommendation method based on secondary attention mechanism

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010818089.8A CN111915216B (en) 2020-08-14 2020-08-14 Open source software project developer recommendation method based on secondary attention mechanism

Publications (2)

Publication Number Publication Date
CN111915216A CN111915216A (en) 2020-11-10
CN111915216B true CN111915216B (en) 2023-09-12

Family

ID=73284073

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010818089.8A Active CN111915216B (en) 2020-08-14 2020-08-14 Open source software project developer recommendation method based on secondary attention mechanism

Country Status (1)

Country Link
CN (1) CN111915216B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117495064A (en) * 2024-01-03 2024-02-02 安徽思高智能科技有限公司 Code reviewer recommendation method oriented to diversity, storage medium and equipment

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108549979A (en) * 2018-04-02 2018-09-18 南京大学 A kind of Open Source Software team extended method based on the embedded expression of precision
CN108830478A (en) * 2018-06-12 2018-11-16 北京航空航天大学 A kind of team's recommended method towards the processing of crowdsourcing task
CN110619082A (en) * 2019-09-20 2019-12-27 苏州市职业大学 Project recommendation method based on repeated search mechanism
CN110851713A (en) * 2019-11-06 2020-02-28 腾讯科技（北京）有限公司 Information processing method, recommendation method and related equipment
CN111444428A (en) * 2020-03-27 2020-07-24 腾讯科技（深圳）有限公司 Information recommendation method and device based on artificial intelligence, electronic equipment and storage medium

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108549979A (en) * 2018-04-02 2018-09-18 南京大学 A kind of Open Source Software team extended method based on the embedded expression of precision
CN108830478A (en) * 2018-06-12 2018-11-16 北京航空航天大学 A kind of team's recommended method towards the processing of crowdsourcing task
CN110619082A (en) * 2019-09-20 2019-12-27 苏州市职业大学 Project recommendation method based on repeated search mechanism
CN110851713A (en) * 2019-11-06 2020-02-28 腾讯科技（北京）有限公司 Information processing method, recommendation method and related equipment
CN111444428A (en) * 2020-03-27 2020-07-24 腾讯科技（深圳）有限公司 Information recommendation method and device based on artificial intelligence, electronic equipment and storage medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于改进的宽深度模型的推荐方法研究;王艺平;冯旭鹏;刘利军;黄青松;;计算机应用与软件(11);55-60 *

Also Published As

Publication number Publication date
CN111915216A (en) 2020-11-10

Similar Documents

Publication Publication Date Title
Abdelrahman et al. Knowledge tracing: A survey
CN110046952B (en) Recommendation model training method and device, and recommendation method and device
Kučak et al. MACHINE LEARNING IN EDUCATION-A SURVEY OF CURRENT RESEARCH TRENDS.
Peña-Ayala Educational data mining: A survey and a data mining-based analysis of recent works
US20170220928A1 (en) Method and System for Innovation Management and Optimization under Uncertainty
Chai et al. Predicting the Risk of Attrition for Undergraduate Students with Time Based Modelling.
Liu et al. Collaborative learning team formation: a cognitive modeling perspective
Al Breiki et al. Using educational data mining techniques to predict student performance
Sokkhey et al. Multi-models of educational data mining for predicting student performance in mathematics: A case study on high schools in Cambodia
Doko et al. Sequential Pattern Mining Model to Identify the Most Important or Difficult Learning Topics via Mobile Technologies.
Pejić et al. Math proficiency prediction in computer-based international large-scale assessments using a multi-class machine learning model
Ali et al. A novel hybrid classification approach for predict performance student in e-learning
CN111915216B (en) Open source software project developer recommendation method based on secondary attention mechanism
He et al. Exercise recommendation method based on knowledge tracing and concept prerequisite relations
Li et al. Research on personalized learning path planning model based on knowledge network
Jayanthi et al. Research contemplate on educational data mining
Kaensar et al. Analysis and Prediction of Student Performance Based on Moodle Log Data using Machine Learning Techniques
Daniati et al. Evaluation Framework for Decision Making Based on Sentiment Analysis in Social Media
Abdel-Aal et al. Constructing optimal educational tests using GMDH-based item ranking and selection
Subahi Data Collection for Career Path Prediction Based on Analysing Body of Knowledge of Computer Science Degrees.
Pejic et al. Relationship Mining in PISA CBA 2012 Problem Solving Dataset using Association Rules
Ma et al. Towards FCA-facilitated ontology-supported recruitment systems
Ogunkunle et al. A data mining based optimization of selecting learning material in an intelligent tutoring system for advancing STEM education
Kim et al. Classification of mathematical test questions using machine learning on datasets of learning management system questions
Raj et al. An Approach for Early Prediction of Academic Procrastination in e-Learning Environment

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