--- author: Leso_KN mat-no: 000000 sources: - Docker Logo - https://www.docker.com/sites/default/files/vertical.png - Tree Graphics - Amulya Gupta 2018, https://www.aclweb.org/anthology/attachments/P18-1197.Presentation.pdf ref: https://www.aclweb.org/anthology/P18-1197/ --- Demo # To Attend Or Not To Attend ## Anaylsis and replication attempt of the case study An original presentation by Leso_KN. Translated from German by Leso_KN. ### Agenda * Introduction * Technical Approach * Field Test * Evaluation * Prospects ## Introduction ## `____________________________________________` ### The Project * Detection of similarities between two sentences * Detection of paraphrases * Case study * Which model generates the best results? * Scope for this analysis:
Similarity of two sentences
## Technical Approach ## > python main.py ### Technical Approach * Classical approach: Linear * Alternative approach: Trees * Motivation:
The assumption, that trees contain more linguistic information
than linear models
* Libraries: Python 3, pytorch, scipy and Stanford Parser
* Stanford Constituency Parser * T-LSTM * Comparison: MDA ⬌ PA * Best behaviour: PA-model
### Technical Approach ## Field Test ## > Test() ### Evaluation * **Replicability** * Was it possible to just run the code?
out-of-the-box experience? * No.. * **Traceability** * Were the results the same? * Yes! * But: The uploaded code was incomplete!
## Prospects ## `____________________________________________` # Many thanks! ### Sources Docker Logo: https://www.docker.com/sites/default/files/vertical.png Tree Graphics: Amulya Gupta 2018, https://www.aclweb.org/anthology/attachments/P18-1197.Presentation.pdf **See also** https://www.aclweb.org/anthology/P18-1197/ Autor: Leso_KN (000000) – Semester: 0
WiSe '19/20 – Applied Methods of Natural Language Processing, Prof. Doc. John Doe
This is a DEMO – Copyright (c) Lesosoftware 2020