Evaluating Anti-Spam Feature Directions: The Server Routing Approach
Case context: Your machine learning team is brainstorming new features to improve an anti-spam system. The current system only looks at the text in the email body, but sophisticated spammers have started evading these content-based filters by constantly altering their wording.
Question: Based on the team's brainstorming directions, propose a specific non-content feature approach using the email's metadata. What information should you extract, and where do you find it?
Sample answer: The team should develop features from the email envelope or header. Specifically, they should extract information showing the set of internet servers the message passed through. This provides a strong signal based on the email's routing history rather than its text.
Key points:
- Propose using the email envelope or header.
- Extract data regarding the internet servers the message passed through.
- Recognize this as a non-content, metadata-driven anti-spam direction.
Rubric: A correct response must identify the email envelope or header as the source and specify that the target information is the set of internet servers the message went through.
0
1
Tags
Machine Learning
Deep Learning
Machine Learning Strategy
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Yearning @ DeepLearning.AI
Related
Which specific information does the email envelope/header feature direction use to help detect spam?
Developing features from the email envelope or header to identify which servers a message passed through is a possible anti-spam system direction.
One anti-spam direction is to develop features from the email _____ or header, which can show which internet servers the message passed through.
Match each email component to its description in the context of anti-spam envelope/header feature development.
Arrange the steps for building an anti-spam feature pipeline using email envelope and header routing information.
In Machine Learning Yearning, the email envelope/header anti-spam direction relies on what type of evidence?
In Machine Learning Yearning, the email envelope/header server-routing direction is the only recommended approach for building an anti-spam system.
Email envelope and header features can reveal which internet _____ the email message passed through, providing a signal useful for spam detection.
Match each anti-spam feature source to the type of information it captures in Machine Learning Yearning's anti-spam example.
Order the reasoning steps a team follows when evaluating whether email envelope/header routing features are a promising anti-spam direction.
Analyzing Email Routing Information for Spam Detection
Evaluating Anti-Spam Feature Directions: The Server Routing Approach
Information Derived from Email Envelopes and Headers