https://github.com/haydenshively/guardian-angel

LSTM for sentiment analysis and detection of domestic violence | LA Hacks 2019 (pair programming)
https://github.com/haydenshively/guardian-angel

android gcp-cloud-functions lstm sentiment-analysis

Last synced: 3 months ago
JSON representation

LSTM for sentiment analysis and detection of domestic violence | LA Hacks 2019 (pair programming)

• Host: GitHub
• URL: https://github.com/haydenshively/guardian-angel
• Owner: haydenshively
• License: unlicense
• Created: 2019-03-31T00:09:34.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2023-03-24T23:43:00.000Z (over 2 years ago)
• Last Synced: 2025-01-30T01:44:50.471Z (5 months ago)
• Topics: android, gcp-cloud-functions, lstm, sentiment-analysis
• Language: Java
• Homepage:
• Size: 881 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Guardian-Angel

LSTM for sentiment analysis and detection of domestic violence | LA Hacks 2019

This is a quick project that my roommates and I did for LA Hacks 2019. We wanted to use

a smartphone's microphone to detect arguments and domestic violence. Lacking a dataset,

we downloaded videos of Walmart fights from YouTube and trained and LSTM for sentiment

analysis.

[Adam Egyed](https://github.com/adamegyed) built an Android app that connects to a GCP

Cloud Function that runs the ML. Data analysis and model training was done by me

in Python.

## Results

Sadly, the results weren't great. The LSTM only occassionally identified violent phrases

successfully. We believe this is because the transcripts of the Walmart videos were too

low quality -- arguments are difficult to convert into organized sentences. In any case,

the potential for false positives is far too high for this app to be practical. Further

development would require a better dataset.