A Siamese Neural Network for Learning Semantically-Informed Sentence Embeddings

Citation

Bölücü N, Can B & Artuner H (2023) A Siamese Neural Network for Learning Semantically-Informed Sentence Embeddings. Expert Systems with Applications, 214, Art. No.: 119103. https://doi.org/10.1016/j.eswa.2022.119103

Abstract
In 2014, 2018 and 2021, we measured the vertical distributions of several water quality indicators in Lake Toba, a representative large tropical lake. This lake has a north basin (NB) and south basin (SB), connected by a strait. Similar water temperature profiles were observed in both basins, showing increasing trends. Shoaling of hypolimnetic DO (dissolved oxygen)-deficient waters was clearly observed in both basins except in the period from 2018 to 2021 during which the zero DO layer deepened in the SB. In 2014 and 2018, the middle-layer maximums (or minimums) of DO were found in the NB while the SB showed a monotonously downward decreasing tendency. Middle-layer minimums of electric conductivity adjusted to 25 °C (EC25) corresponded to the middle-layer DO maximums in the NB; significantly negative correlations between DO and EC25 were found in both basins. Based on horizontal distributions of EC25, water quality difference between the basins using satellite imagery and gradual change in the DO-EC25 relation, we consider the flow of hypolimnetic water from SB to NB and/or influence of worse water quality near the bottom of the strait with reference to the different behaviors of DO and EC25.

Keywords
Semantic parsing; UCCA; Self-attention; Semantic textual similarity; Siamese Network; Recursive Neural Network

Journal
Expert Systems with Applications: Volume 214

People

Dr Burcu Can Buglalilar

Lecturer in Computing Science, Computing Science