Skip to main content

​
Discovering Notebooks

Alph makes it easy to find relevant notebooks using AI-powered semantic search that understands meaning, not just keywords. Alph’s semantic search works at two levels:

Notebook-Level Search

Search by title, description, and overall purpose to find relevant notebooks

Cell-Level Search

Search within individual code cells to find specific implementations and techniques
This dual approach means you can find notebooks whether you’re looking for high-level concepts or specific code patterns. Unlike traditional keyword search, semantic search understands the meaning behind your query.

​
How It Works

1

Vector embeddings

Every public notebook in Alph is converted to a vector embedding using Google’s Gemini embedding-001 model (1536 dimensions).
2

Two-tier indexing

Metadata embeddings: Capture notebook title, description, and overall purposeCell-level embeddings: Index individual code cells for detailed search
3

Similarity search

When you search, your query is embedded and compared against all notebook embeddings using cosine similarity.
4

Ranked results

Results are ranked by relevance, not just keyword matches. Notebooks that conceptually match your query appear first.

​
Example Queries

  • By Technique
  • By Problem
  • By Data Type
  • By Library
Query: “sentiment analysis with transformers”Finds: Notebooks using BERT, RoBERTa, or other transformer models for sentiment classification, even if they don’t use those exact words.
Natural language queries work better than keywords. Describe what you want to accomplish!

​
Search Interface

1

Navigate to Notebooks

Click Notebooks in the main navigation (not within an organization)
2

Enter your query

Type your search in natural language:
  • “how to clean text data for nlp”
  • “exploratory data analysis sales data”
  • “train CNN for image classification”
3

Review results

Results show:
  • Notebook title and description
  • Relevant tags
  • Author and organization
  • View count and engagement metrics
  • Match score (how relevant)
4

Preview or open

  • Click title: Open full notebook
  • Hover: Quick preview of first few cells
  • Copy: Save to your organization

​
Search Filters

Refine results with filters:
Select tags to narrow results:
  • Language: python, r, julia
  • Domain: ml, data-viz, statistics
  • Level: beginner, intermediate, advanced
Search notebooks from specific users or organizations:
  • Your connections
  • Verified authors
  • Official tutorials
  • Published last week
  • Published last month
  • Published last year
  • All time
  • Relevance (default): Best semantic match
  • Trending: Most viewed recently
  • Popular: All-time views
  • Recent: Newest first

​
Browse by Tags

Explore notebooks by topic:

Machine Learning

  • classification
  • regression
  • clustering
  • deep-learning

Data Processing

  • data-cleaning
  • etl
  • pandas
  • sql

Visualization

  • matplotlib
  • plotly
  • seaborn
  • data-viz

NLP

  • text-processing
  • sentiment-analysis
  • transformers
  • nlp

Computer Vision

  • image-classification
  • object-detection
  • cnn
  • opencv

Time Series

  • forecasting
  • arima
  • lstm
  • time-series
Click any tag to see all related notebooks. Discover what’s popular in the community: Notebooks are ranked by:
  • Recent views: More weight to recent activity
  • Engagement: Copies, stars, shares
  • Velocity: Rapid growth in popularity
  • Quality: Complete execution, good documentation
Updated every hour.
  • This Week
  • This Month
  • All Time
  • Rising
Hottest notebooks from the past 7 days

​
Personalized Recommendations

Alph suggests notebooks based on your interests:

​
How Recommendations Work

  • Your notebooks: Analysis of topics you work on
  • Your views: Notebooks you’ve looked at
  • Your organization: What your team is interested in
  • Your searches: Past search queries

​
Where You See Recommendations

  • Organization dashboard
  • After viewing a notebook (“Related notebooks”)
  • Email digest (if enabled)
  • When creating new notebooks
Recommendations are privacy-preserving. Your private notebooks are never shared or used to train models.
Search within code cells for specific implementations:
  1. Enable Code Search toggle
  2. Enter specific code patterns or techniques
  3. Results show individual cells, not just notebooks
  4. Click to see cell in context
Example: Search for “RandomForestClassifier fit” to find specific sklearn usage.

​
Boolean Operators

Combine search terms: 
# AND (implicit)
machine learning python

# OR
(classification OR regression)

# NOT
neural networks -tensorflow

# Exact phrase
"exploratory data analysis"

​
Search Syntax

OperatorExampleMeaning
tag:tag:tutorialHas specific tag
author:author:janeBy specific author
org:org:acme-dataFrom organization
after:after:2024-01-01Published after date
lang:lang:pythonSpecific language

​
Saving Searches

Save frequently used searches:
  1. Perform a search with filters
  2. Click Save Search
  3. Name your saved search
  4. Access from sidebar under Saved Searches
Get notifications when new notebooks match saved searches (optional).

​
Collection​s

Organize discovered notebooks into collections:

​
Creating Collections

  1. Click New Collection from your profile
  2. Name and describe the collection
  3. Add notebooks by clicking Add to Collection
  4. Make public or keep private

​
Example Collections

  • “Best ML Tutorials”
  • “Time Series Analysis Resources”
  • “Team Onboarding Notebooks”
  • “Research Project References”
Collections can be shared with your organization or made public.

​
Discovery Best Practices

Better: “visualizing correlation matrix with seaborn” Worse: “plot”Specific queries yield better results.
Start broad with search, then narrow with filters:
  1. Search: “customer segmentation”
  2. Filter: tag:clustering, tag:beginner
  3. Sort: by popularity
Found something interesting but no time now?
  • Star the notebook
  • Access from Starred in your profile
  • Organize into collections later
To really learn, copy notebooks to your workspace:
  • Experiment with the code
  • Modify for your data
  • Learn by doing
Help the community:
  • Share on social media
  • Add to public collections
  • Mention in your notebooks
  • Credit original authors

​
What’s Searchable

  • ✅ Public notebooks only
  • ✅ Notebook metadata (title, description)
  • ✅ Tags and categories
  • ✅ Public user profiles

​
What’s Not Searchable

  • ❌ Private notebooks
  • ❌ Organization-internal notebooks (unless you’re a member)
  • ❌ Draft notebooks
  • ❌ Your search history (private to you)

​
Search Analytics

For public notebook authors, view search performance:
  • Search impressions: How often your notebook appears in results
  • Click-through rate: Percentage who click from search
  • Average position: Typical ranking in results
  • Top queries: What searches find your notebook
Use this to optimize titles, descriptions, and tags.

​
Improving Search Results

Help others find your notebooks:
1

Descriptive titles

Include key techniques and topics in the title
2

Detailed descriptions

1-2 sentences explaining what the notebook does and what techniques it uses
3

Relevant tags

Use specific, searchable tags (max 10)
4

Clear markdown

Well-documented notebooks rank higher in semantic search
5

Complete execution

Notebooks with outputs are preferred over empty ones

​
Next Steps

Publishing Notebooks

Make your notebooks discoverable

GitHub Integration

Import notebooks from repositories

Notebook Basics

Learn about working with notebooks

Organizations

Collaborate with your team