A Method for Advanced Analysis of Item Browsing Data Considering User Interest Durations Using Parameters Estimated from Hidden Semi-Markov Model
Lists
Only you see this
Abstract
Recently, competition among video streaming services for customers has intensified, so that it has become important to introduce appropriate marketing measure considering users’ preference and their purchasing behavior. In general, users’ purchasing actions for video content (items), unlike daily necessities, have a strong influence on their real-time transition of interests while viewing items (consumption). In other words, the user’s next purchasing intention after the consumption of an item is influenced by whether their interest is continued (probability of continued interest under the item). Therefore, it is important to select and evaluate items based on the probability of continued interest under items to allow users to use the service for a long time. For this purpose, the hidden semi-Markov model (HSMM), proposed as a model to predict the next item to be consumed by a user, can be a solution considering the user’s interest persistence. If the probabilities of continuing interest under each item can be calculated and analyzed by using the HSMM, new insights can be expected to lead to marketing strategies. In this study, we propose an analysis process based on item clustering using the probabilistic distribution of continued interests, using the characteristics of HSMM. We demonstrate the effectiveness of our proposed method by applying it to an actual dataset. The experimental results show that we can obtain the characteristics of the probability of continued interest under item for each class of items, and that we can evaluate items from a new perspective of the probability of continued interest under item.
BibTeX
@inproceedings{tsuchiya2023method,
author = {Kirin Tsuchiya and Yuki Tsuboi and Ryotaro Shimizu and Masayuki Goto},
title = {{A Method for Advanced Analysis of Item Browsing Data Considering User Interest Durations Using Parameters Estimated from Hidden Semi-Markov Model}},
booktitle = {Industrial},
year = {2023},
doi = {10.7232/iems.2023.22.4.437},
}Loading related content...