ICACT20230130 Question.1
Questioner: liuyuya@ucas.ac.cn 2023-02-21 오후 3:11:31 |
ICACT20230130 Answer.1
Answer by Auhor liufan@nus.edu.sg 2023-02-21 오후 3:11:31
Chrome  Click!! |
| After data integration, the recommendation strategy adopted in this paper is traditional collaborative filtering method. Will the recommendation accuracy increase if using deep learning-based methods? |
Thanks for your question.
At present, our primary focus of research is to implement a hybrid recommendation algorithm to manage a large amount of data. However, incorporating deep learning techniques into the recommendation system is an interesting idea, and we are exploring ways to merge DNN with our existing algorithm. |
ICACT20230130 Question.11
Questioner: yangkoon@gmail.com 2023-02-22 오후 12:58:10 |
ICACT20230130 Answer.11
Answer by Auhor liufan@nus.edu.sg 2023-02-22 오후 12:58:10
Chrome Click!! |
| I heard that the Lucene score was calculated to determine the popularity of the book, but how can I judge the accuracy of this? |
Thanks for your question.
We compared the hybrid algorithm with popularity (including both item-based and ALS) with the algorithm without popularity. The former is slightly better than the latter.
|
ICACT20230130 Question.2
Questioner: jianms@nfu.edu.tw 2023-02-21 오후 5:39:21 |
ICACT20230130 Answer.2
Answer by Auhor liufan@nus.edu.sg 2023-02-21 오후 5:39:21
Chrome Click!! |
| From the various social media the data would be diverse.
Is the clustering or grouping the individuals from the social media needed in advance in your research?
Or any possible tool is used for finding the feature of each person in your research? |
Thanks for your question.
We don't group different users in advance. As the proposed recommendation system is focused on the item-based recommendation algorithm.
However, in the future, if we can collect more user data such as user demographic, we can explore how to utilize such data in our research work. |
ICACT20230130 Question.3
Questioner: quandh13@fe.edu.vn 2023-02-23 오전 10:18:53 |
ICACT20230130 Answer.3
Answer by Auhor liufan@nus.edu.sg 2023-02-23 오전 10:18:53
Chrome Click!! |
| Hello, I would like to know if and how your proposed method can handle the item cold start problem. Thanks! |
Thanks for your question.
For new items with little or no historical data, the system can make recommendations based on their popularity. |
