Early identification of high attention content for online mental health community users based on multi-level fusion model
| Date | 11 July 2024 |
| Pages | 838-857 |
| DOI | https://doi.org/10.1108/DTA-06-2023-0230 |
| Published date | 11 July 2024 |
| Author | Song Wang,Ying Luo,Xinmin Liu |
Early identification of high
attention content for online mental
health community users based
on multi-level fusion model
Song Wang and Ying Luo
Shandong University of Science and Technology, Qingdao, China, and
Xinmin Liu
Qingdao Agricultural University, Qingdao, China
Abstract
Purpose –The overload of user-generated content in online mental health community makes the focus and
resonance tendencies of the participating groups less clear. Thus, the purpose of this paper is to build an early
identification mechanism for users’high attention content to promote early intervention and effective
dissemination of professional medical guidance.
Design/methodology/approach –We decouple the identification mechanism from two processes: early
feature combing and algorithmic model construction. Firstly, based on the differentiatedneeds and concerns of
the participant groups, the multiple features of “information content þsource users”are refined. Secondly, a
multi-level fusion model is constructed for features processing. Specifically, Bidirectional Encoder
Representation from Transformers (BERT)-Bi-directional Long-Short Term Memory (BiLSTM)-Linear are
used to refine the semantic features, while Graph Attention Networks (GAT) is used to capture the entity
attributes and relation features. Finally, the Convolutional Neural Network (CNN) is used to optimize the multi-
level fusion features.
Findings –The results show that the ACC of the multi-level fusion model is 84.42%, F1 is 79.43% and R is
76.71%. Compared with other baseline models and single feature elements, the ACC and F1 values are
improved to different degrees.
Originality/value –The originality of this paper lies in analyzing multiple features based on early stages and
constructing a new multi-level fusion model for processing. Further, the study is valuable for the orientation of
psychological patients’needs and early guidance of professional medical care.
Keywords Online mental health community, High attention content, Early identification,
Deep learning model, BERT-BiLSTM-linear, GAT
Paper type Research paper
1. Introduction
In recent years, as the public’s demand for health awareness and health information has
increased significantly, virtualized online platforms are becoming an important way for
people to search for health knowledge and share medical experiences (Tala and Pouyan,
2020). In particular, mental health communities established to alleviate depression, anxiety
and many other disturbances have received particular attention (Gu et al., 2023). In these
communities, users express their personal emotions and requests for help through postings,
and patients with similar experiences respond based on information interactions such as likes
and comments (Choudhury et al., 2016;Wang et al., 2021). However, with the increase in the
number of participating users, the content volume of the community has long reached an
overload state. The large amount of content flooding the community makes the focus and
resonance tendencies of the participating groups less clear. The above phenomenon tends to
DTA
58,5
838
Competing interest: The authors declare that they have no conflict of interest.
Funding: Not applicable.
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/2514-9288.htm
Received 4 June 2023
Revised 14 December 2023
3 April 2024
Accepted 20 June 2024
Data Technologies and
Applications
Vol. 58 No. 5, 2024
pp. 838-857
© Emerald Publishing Limited
2514-9288
DOI 10.1108/DTA-06-2023-0230
weaken the efficiency of targeting users’health needs and hinders the early intervention of
professional medical guidance (Swar et al., 2017). Therefore, there is an urgent need to build
an early identification mechanism around online mental health communities targeted at
content of high attention to users. While assisting scientific medical treatment to intervene in
a timely manner, it is also possible to rely on the high attention potential of this content so that
more patients with similar claims can receive timely help.
The exploration of refinements around users, information and behaviors in mental health
communities has been a hot topic of scholarly interest (Zhou et al., 2019;Chen and Xu, 2021).
Most of the current studies provide insights into the characteristics of doctor-patient users,
engagement behaviors, user needs mining and mental health prediction. At the same time,
they recognize the importance of content generated by doctor-patient users, which provides
support value for the successive exploration of this paper (Liu and Gao, 2022;Kabir et al.,
2023). However, few studies have paid attention to the hot spots and resonance tendencies of
user groups in mental health communities, and the early identification studies conducted on
this need to be further explored. It is worth noting that the content of high attention to users in
the community is often discussed from the perspective of information content quality, user
commenting behavior, etc. for which a more complete index system and identification
framework has been formed (Guo et al., 2022). But, the above features are mostly focused on
the mature stage of content dissemination, and the identification mechanism has obvious
lagging effects. Only based on the content distribution stage, according to the different needs
of different participant groups, we can comprehensively sort out the motivation factors that
trigger users’attention behavior and form a multi-faceted feature system to assist content
identification. In addition, with the development of artificial intelligence technology, machine
learning and deep learning methods are widely favored by scholars and have been validated
as well as applied in many ways (Cheng and Chen, 2022). Specifically, Convolutional Neural
Network (CNN), which better captures local semantic features; Recurrent Neural Network
(RNN), which is good at dealing with long and short-term dependencies; Attention, which
focuses on key semantics; Graph Neural Network (GCN) and Graph Attention Neural
Network (GAT), which optimize “node þrelation”data, provide strong technical support for
the value extraction and comprehensive processing of multi-feature elements in this paper
(Schmidhuber, 2015;Wu et al., 2021).
In view of this, this paper will focus on the early identification of users’high attention
content in online mental health community. Specifically, using a health forum for depression
patients as a research scenario, we will focus on the early stages of content distribution, and
target different user groups’differentiated needs, so as to refine the multiple features that
influence content attention; Following that, Bidirectional Encoder Representation from
Transformers (BERT), Bi-directional Long-Short Term Memory (BiLSTM) and GAT are used
as baseline models to build a multi-level fused deep learning architecture for processing.
Finally, we achieve the identification of the target content by the purpose of text
classification. The main contributions of this work are summarized as follows:
(1)We have developed a multiple feature system of “information content and source
user”for early identification, based on the differentiated information needs of
diagnosed patients, potential demanders of medical knowledge and professionals
who provide medical advice.
(2)We design a multi-level fusion model for early identification of users’high attention
content, which enables integrated processing of text semantics and entity relation
networks.
(3)We conducted experiments on the data of “Sunshine Project Psychological Support
Forum”. It is verified that the integrated capture of multiple features is better than the
Data
Technologies and
Applications
839
Get this document and AI-powered insights with a free trial of vLex and Vincent AI
Get Started for FreeStart Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting