Dilated LSTM with attention for Classification of Suicide Notes

Schoene, Annika Marie; Lacey, George; Turner, Alexander P.; Dethlefs, Nina

doi:10.18653/v1/d19-6217

Cited by 12 publications

(7 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such research has been conducted in a range of different disciplines like psychology [18], linguistics [13] or healthcare [19]. Many experiments have also been conducted comparing different types of textual data with suicide notes such as depressed language or blog posts [20]. Overall there has been a growing interest in looking at content created online that may solicit need for help [21] or detecting mental health issues [22].…”

Section: Related Workmentioning

confidence: 99%

“…The main reason for this being that these tweets have mainly been collected using specific keywords such as 'suicide' to accumulate the data and there is no human verification that the person who wrote this tweet is indeed suicidal or has passed away. Due to the sparsity of genuine suicide notes that are publicly available, we have added new genuine suicide notes to the corpus provided by [20]. Other new additions to this corpus includes data from various sources (for a full list, see Appendix A).…”

Section: Genuine Suicide Note Datamentioning

confidence: 99%

See 1 more Smart Citation

Hierarchical Multiscale Recurrent Neural Networks for Detecting Suicide Notes

Schoene

Turner

Mel³

et al. 2023

IEEE Trans. Affective Comput.

Self Cite

View full text Add to dashboard Cite

Recent statistics in suicide prevention show that people are increasingly posting their last words online and with the unprecedented availability of textual data from social media platforms researchers have the opportunity to analyse such data. Furthermore, psychological studies have shown that our state of mind can manifest itself in the linguistic features we use to communicate. In this paper, we investigate whether it is possible to automatically identify suicide notes from other types of social media blogs in two document-level classification tasks. The first task aims to identify suicide notes from depressed and blog posts in a balanced dataset, whilst the second experiment looks at how well suicide notes can be classified when there is a vast amount of neutral text data, which makes the task more applicable to real-world scenarios. Furthermore we perform a linguistic analysis using LIWC (Linguistic Inquiry and Word Count). We present a learning model for modelling long sequences in two experiment series. We achieve an f1-score of 88.26% over the baselines of 0.60 in experiment 1 and 96.1% over the baseline in experiment 2. Finally, we show through visualisations which features the learning model identifies, these include emotions such as love and personal pronouns.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Genuine Suicide Note Datamentioning

confidence: 99%

Hierarchical Multiscale Recurrent Neural Networks for Detecting Suicide Notes

Schoene

Turner

Mel³

et al. 2023

IEEE Trans. Affective Comput.

Self Cite

View full text Add to dashboard Cite

show abstract

“…The biometric information (iris image)

{II}_{kj}^{UA}

about the user is provided to the AD‐LSTM‐AN model for user authentication. The dilated LSTM 29 differs from the general LSTM structure by the presence of dilated skip connections of recurrent layers. In ADLSTM‐AN, exponentially rising dilations are provided to the skip connections of recurrent layers.…”

Section: User Authentication and Trust Verification For Secured Healt...mentioning

confidence: 99%

HECC‐ABE: A novel blockchain‐based IoT healthcare data storage using hybrid cryptography schemes with key optimization by hybrid meta‐heuristic algorithm

Dubey,

Ramanjaneyulu,

Saraswat

et al. 2023

Trans Emerging Tel Tech

View full text Add to dashboard Cite

The most contemporary development in information technology is the Internet of Things (IoT), which integrates the digital world with the real world. It makes it possible for things and people to communicate via the Internet. Blockchain is one of the popular topics of interest right now, and it can be used in most IoT applications. The blockchain's salient characteristics, like decentralization, data integrity, confidentiality, protection, and openness, are the main reasons for implementing it in healthcare services. Hence, this paper introduces an effective approach for storing IoT‐based healthcare data storage in the blockchain. The procedures in this suggested paradigm are as follows: user authentication, user trust verification, and optimal key storage in the blockchain. Wearable sensors are inserted in or adhered to a patient's body and used by IoT networks to gather healthcare data. These details are supplied into the verification structure constructed by adaptive dilated long short term memory with attention network (AD‐LSTM‐AN). Once the user's authentication is confirmed, then the user's trust level is computed. The same AD‐LSTM‐AN‐based verification structure is used to verify the historical data that has been stored and retrieved by the user, together with their previous transactions, in this phase. The healthcare data are then sent to the hybrid elliptic curve cryptography with attribute‐based encryption (HECC‐ABE) cryptography schemes for data encryption once the authentication of the person who attempts to store the data is confirmed. An adaptive final position‐based golden eagle‐Harris hawks optimization (AFP‐GEHHO) algorithm is used for optimizing the keys of the encrypted data. The blockchain platform stores the encrypted data of the authorized user with a high level of trust. The same process is done in reverse order whenever a user needs to recover information that is saved in the blockchain to retrieve the initially stored healthcare data. The effectiveness of secure data storage in blockchain is tested through experimental simulations. Throughout the analysis, the designed model achieves 96% in terms of accuracy. Thus, the developed model shows more secure and effective, has less error and reduces the time and memory requirements compared to existing approaches.

show abstract

“…Then a knowledge-base will be build containing target-entity and emotion pairs that will later be used to add 'internal' knowledge into the RNN's cell. Furthermore, the learning model has been applied to document classification tasks to classify suicide notes from other types of discourse using both linguistic, cognitive and sentiment features (Schoene et al 2019).…”

Section: Progress and Research Planmentioning

confidence: 99%