Teamwork Representation Learning With Multi-channel Mechanism

Recently,the vigorous development of economy and technology has spawned a number of larger and more complex projects.At the same time,more and more people have begun to choose the teamwork in order to achieve the goals required by projects,such as in academic cooperation,legislative systems and other scenes with fierce com-petition.However,how to quickly,efficiently and intelligently build a team and meet the requirements of projects is still full of challenges.Specifically,in order to increase the probability of teamwork to be succeed,the construction of the team usually needs to consider multiple factors.On the one hand,the team needs to bring in members with different professions,qualifications and backgrounds to enhance the team's profession-alism and authority.On the other hand,the team members who use different strategies to attract supporters with common interests can also enable projects to receive support and help from multiple channels.Therefore,in-depth analysis of the team proposal process centered on projects,that is,the process by which individuals make team pro-posals which representing their common interests successful through teamwork,is an important topic in the field of quantitative proposal analysis and teamwork learning.However,the existing quantitative analysis methods for team proposals still have much room for improvement.First,the existing technology does not fully excavate and utilize the complex structure information in the team proposal process,such as in-terpersonal relationships,interest relationships,and support relationships.Most of the current mainstream team proposal process analysis only uses individual-level resume information and calculates the match degree between the individual and the team pro-posal to predict the success of the team proposal.This type of method greatly ignores team-level information,making it unable to accurately reflect the complete process of team proposals,which in turn leads to certain limitations in predictive performance and interpretability.Second,the existing quantitative proposal analysis and teamwork learn-ing methods are mostly based on traditional statistical learning methods.This type of method often requires manual construction of a large number of features and its non-end-to-end learning method makes them inefficient and not easy to apply.Therefore,in order to better model and quantify the team proposal,this article starts from the individual and team levels of the team proposal process to solve the disadvan-tages of poor performance and interpretability of traditional methods.Furthermore,this article solves the efficiency problem of traditional methods by using complex structure modeling methods such as deep learning and graph neural networks.The main research contents and contributions of this paper are as follows:(1)In the field of quantitative proposal analysis and teamwork representation,dif-ferent from the existing methods that only focus on personal resume information,this article proposes a complete modeling of the team proposal process from the perspective of the individual and the team.(2)A team proposal process modeling framework based on the individual-level and team-level is proposed.The framework uses deep learning,graph neural network and other complex structure processing methods to fully integrate personal professionalism,personal interest appeals,and potential team supporters information,which can more accurately model the team proposal process.(3)Based on the above framework,this paper conducts a lot of data analysis and comparison experiments on real-world data.The experimental results show the effec-ti veness of the team proposal representation method proposed in this paper.At the same time,based on the interpretability of the model,related findings can be used for guidance and suggestions for team building.