<p>Detecting duplicate bug reports is an active area of research in software testing, but not often deployed. Most developers have experienced duplicated bug reports, however, only few consider them as a serious issue. In this paper we addressed this gap by performing exploratory study on the duplication of bug reports by preliminary proposing a prototype i.e., Siamese Subnet Signal Compression Model (SSSC) that uses two connected LSTM layers to predict the duplicates in bug reports. Our analysis tested three thresholds of 0.3, 0.5 and 0.7, claiming that threshold 0.3 performed better than two other thresholds. We reviewed the F1-score, which shows that the score for Level 0 is high (90%), and the score for Level 1 is also relatively high compared to other thresholds. Finally, we demonstrated that SSSC Manhattan Distance Classifier has shown promising results compared to other baseline techniques with an accuracy of 84%, indicating that better results can be attained through a much more rigorous study. We evaluated our model for cross-project bug classification; while performance was suboptimal, the results are promising at this stage, indicating that further research on duplicate bug report detection is warranted.</p>