Features
- Profanity software
- Accessible via an admin and API interaces;
- detect, identify, prevent and report profanity / insult at a rate of 80% and goes up to 90% including preprocessing stage.
- identify the originator and the source,
- Self-learning and eliminating human effort,
- Machine Learning (ML) based,
- Increases the accuracy rate by providing adaptation to new data over time,
- Analyze Streaming live data
- Via feedback forms, we detect our misclassifications and improve upon.
- Profanity filter features
- detecting swear, profane, bad words and beeping
Area of Use
- Chat,
- SMS,
- Blog(s),
- Feedback Form(s),
- Forum(s),
- Customer Message(s),
- Complaint system(s),
- Dating site(s)
Innovative Sides
- By the use of Asynchronous processing, we achieve great speed:
- 300 words : 250ms – rps: 10
- 500words : 341ms – rps: 7
- 1000words : 547ms – rps: 4
- 1500words : 798ms – rps: 2
- 3000words : 1.58s – rps: 0.8
- Using NLP we extract profane words and their positions
- We can eliminate special characters, URLs, emails, etc
- Profanity software has Easy API integration with other systems
- Finding both exact, similar, derived and suspicious word matches unlike other systems which rely on a single dictionary or exact word matches
- Self- Learning via ML algorithms that remodels itself according to feedback and new data increasing its accuracy and adaptiveness to the customer’s needs
- Sentence Analysis to detect insult besides profanity
- Whitelisting and blacklisting of words
