2019 Fiscal Year Annual Research Report
Research on Influence Maximization Algorithm and Recommendation Application based on Social Big Data
| Project/Area Number |
19F19704
|
|---|
| Research Institution | Tokyo University of Technology |
|---|
Principal Investigator |
大山 恭弘 東京工科大学, 工学部, 教授 (00233289)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
WANG FENG 東京工科大学, 工学部, 外国人特別研究員
|
|---|
| Project Period (FY) |
2019-04-25 – 2021-03-31
|
| Keywords | Influence Maximization / Recommendation Algorithm / Social Big Data / User Influence Eval. / Network Analysis |
|---|
| Outline of Annual Research Achievements |
We devised a preprocess method for user influence information in social big data. First, we presented methods to identify the relevant influence information including user social relationships, hobbies, influence relationships, etc. Then, we designed flexible and efficient multi-dimensional influence information collection models and methods. Finally, we proposed an implicit representation learning approach for user influence features to improve the performance of influence measurement.
We also devised an influence deep learning (IDL) model to learn users’ latent feature representation for predicting influence spread. The IDL model is fully data-driven, and it uses sampling subnetworks as inputs to deep neural networks for learning users’ latent vector representation.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The study is progressed as schedule. We prepared the study very smoothly. So, It made it possible for me to start and focus on my study. As a result, I submitted a paper to a very famous international conference and successfully presented my paper.
|
| Strategy for Future Research Activity |
Influence maximization algorithm (2020/04~2020/10)
(1) To establish a preprocess method for user influence information in online social networks. (2) To propose a user influence evaluation approach based on the social big data. Then, we plan to devise a novel influence maximization algorithm to find a balance point between effectiveness and efficiency.
Recommendation algorithm based on the top-k influential nodes (2020/10~2021/03)
(1) To propose a new method of analyzing the important role of influential users in the process of education resources recommendation. (2) To devise a group recommendation algorithm of education resources based on the top-k influential nodes. (3) To attempt to develop algorithms for the seed set detection and resource recommendation.
|
