Abstract
Discussion features in online communities can be effectively used to diagnose depression and allow other users or experts to provide self-help resources to those in need. Automatic emotion identification models can quickly and effectively highlight indicators of emotional stress in the text of such discussions. Such communities also provide patients with important knowledge to help better understand their condition. This study proposes a deep learning framework combining word embeddings, bi-directional long short-term memory (Bi-LSTM), and convolutional neural networks (CNN) to identify emotion labels from psychiatric social texts. The Bi-LSTM is a powerful mechanism for extracting features from sequential data in which a sentence consists of multiple words in a particular sequence. CNN is another powerful feature extractor which can convolute many blocks to capture important features. Our proposed deep learning framework also applies word representation techniques to represent semantic relationships between words. The paper thus combines two powerful feature extraction methods with word embedding to automatically identify indicators of emotional stress. Experimental results show that our proposed framework outperformed other models using traditional feature extraction such as bag-of-words (BOW), latent semantic analysis (LSA), independent component analysis (ICA), and LSA+ICA.
-
Author Keywords
- Multiple emotion labeling ,
- deep learning ,
- bi-directional recurrent neural network ,
- long short-term memory neural network ,
- convolutional neural network
-
IEEE Keywords
- Feature extraction ,
- Deep learning ,
- Logic gates ,
- Artificial neural networks ,
- Semantics ,
- Bidirectional control ,
- Analytical models
Introduction
Rather than seek professional help, people suffering from mental illness or emotional strain often turn to online communities in search of advice or a sense of human intimacy and understanding. In recent years, many online services have been developed to provide such people with a means for identifying and understanding the issues they face, and for finding helpful resources. Sufferers interact with these services through written texts and comments about their feelings, and qualified therapists who monitor these services then provide replies and appropriate suggestions.
However, such services suffer from an imbalance between “clients” and “providers”. Combined with the asynchronous nature of such communication, clients may wait for considerable lengths of time between replies, which not only reduces the potential benefit of engaging with the service but can also increase client anxiety. Excessive response delay increases the potential of self-harm or other negative behavior on the part of the client. A system that automatically parses client comments to identify particular emotions and their respective severity would allow providers to quickly identify clients in crisis, allowing them to prioritize responses, and thus, potentially avert undesirable outcomes. Such a system could also help provide a macro view of the relative prevalence of various emotional and psychological issues among service clients [1].
Healthcare-oriented web-based services draw many text-based queries related to depression. Psychiatrists and therapists reviewing and replying to these queries label them appropriately to represent the particular type of depression indicated. This paper seeks to automatically label such texts with appropriate emotion labels [2], thus reducing therapist workload and response latency. Table 1 shows an example text annotated with three emotion labels [3], depression, insomnia, and suicide. This example shows the difficulty of detecting and labeling multiple emotions in the field of sentiment analysis.
This paper uses deep learning models as neural networks (NNs) to resolve the issue of multiple label classification. Word embeddings, convolutional neural networks (CNN), and long short-term memory (LSTM) NNs are used to build a powerful classifier to identify emotion labels. These models have been successfully applied in a wide range of categorization tasks and are used here to develop a powerful classifying mechanism for multiple emotion labels within psychiatric social texts based on the classification performance of two factors: word embeddings and neural network architectures.
This paper has three main contributions. First, this study is the first work to use NN-based methods to address multiple emotion classification for psychiatric social texts, which can reduce the workload and response latency of psychiatrists and therapists. Second, the proposed framework outperforms other models using traditional feature extraction such as bag-of-words (BOW), latent semantic analysis (LSA), independent component analysis (ICA), and LSA+ICA. Third, different NN models are evaluated, providing a baseline result to facilitate future research on emotion classification for psychiatric social texts.
The remainder of this paper is organized as follows. Section II reviews the relevant literature. Section III describes the proposed model for multiple emotion classification. Section IV explains the generation of the psychiatric social text dataset and summarizes the experimental results. Conclusions and directions for future work are presented in Section V.
Conclusion
We proposed a deep learning model to assign emotion labels to psychiatric social texts. The proposed Bi-directional LSTM-CNN combines word embedding, long short-term memory networks, and convolutional neural networks to extract the hidden features. The major decision hidden features are obtained from the previous hidden features by CNN, and the word hidden features are obtained from the word embeddings. The abstractive hidden features are then successfully used to identify emotion labels. Therefore, the pipeline feature extraction processes, such as word embedding layer, bi-directional LSTM layer, and CNN layer extract many important features and provide high detection performance. Experimental results show that the proposed deep learning model significantly improved performance over other conventional models. Our proposed model using pretrained word embeddings through the GloVe model outperformed random initial word embeddings. Future work will focus on other deep learning approaches such as the attention-based model and tree-LSTM to improve performance and will include additional sentiment corpora to allow the system to capture more sentiment information.
About KSRA
The Kavian Scientific Research Association (KSRA) is a non-profit research organization to provide research / educational services in December 2013. The members of the community had formed a virtual group on the Viber social network. The core of the Kavian Scientific Association was formed with these members as founders. These individuals, led by Professor Siavosh Kaviani, decided to launch a scientific / research association with an emphasis on education.
KSRA research association, as a non-profit research firm, is committed to providing research services in the field of knowledge. The main beneficiaries of this association are public or private knowledge-based companies, students, researchers, researchers, professors, universities, and industrial and semi-industrial centers around the world.
Our main services Based on Education for all Spectrum people in the world. We want to make an integration between researches and educations. We believe education is the main right of Human beings. So our services should be concentrated on inclusive education.
The KSRA team partners with local under-served communities around the world to improve the access to and quality of knowledge based on education, amplify and augment learning programs where they exist, and create new opportunities for e-learning where traditional education systems are lacking or non-existent.
FULL Paper PDF file:
Identifying Emotion Labels From Psychiatric Social Texts Using a Bi-Directional LSTM-CNN ModelBibliography
author
Year
2020
Title
Identifying Emotion Labels From Psychiatric Social Texts Using a Bi-Directional LSTM-CNN Model,
Publish in
in IEEE Access, vol. 8, pp. 66638-66646, 2020,
Doi
10.1109/ACCESS.2020.2985228.
PDF reference and original file: Click here
Somayeh Nosrati was born in 1982 in Tehran. She holds a Master's degree in artificial intelligence from Khatam University of Tehran.
Professor Siavosh Kaviani was born in 1961 in Tehran. He had a professorship. He holds a Ph.D. in Software Engineering from the QL University of Software Development Methodology and an honorary Ph.D. from the University of Chelsea.
Nasim Gazerani was born in 1983 in Arak. She holds a Master's degree in Software Engineering from UM University of Malaysia.
