Learn how to build and deploy a semantic search engine, first for text documents and later for images. This series of hands-on sessions will touch upon Python programming, need-to-know concepts about embeddings, vector DBs, machine learning, and cloud deployment.
| Delivery Mode | Online & In Person |
| Duration | 4 hours |
| Prerequisites | Containerization 101, Serverless Computing, Embeddings & Vector DBs |
What Is Semantic Search?
Semantic search is a technique for searching information by understanding the meaning, or semantics, behind the words in queries and documents rather than just keyword matching.
A Few Simple Examples
A Simple Example
🏘️🌳 “houses near a green area”
Imagine a prospective home buyer is looking for properties to purchase near a park. They might search for “houses near a green area”. A traditional keyword-based search might return very few relevant results, as it only matches those exact words.
A semantic search engine, on the other hand, would understand the concepts behind the words. It knows that a “green area” is related to terms like “park”, “recreational area”, and “playground” 🌳 + 🛝 + 🏘️. It also knows that “houses” are a type of dwelling, so it would expand the search to include related terms like “homes” or “properties”.
By mapping the concepts in the query to associated ideas in its knowledge base, a semantic search engine can return listings that mention houses or properties near parks, trails, recreational facilities, and other related terms. This provides a complete picture of available options that meet the user’s needs rather than just literal keyword matches.
Another Example
🥗🍉🚴♂️🏃🩸🧁 “links between diet, exercise and blood sugar levels”
Imagine a person is curious about the link between lifestyle and diabetes rates. S/he uses a semantic search engine to query for “links between diet, exercise, and blood sugar levels”.
The semantic search engine understands that the medical term for blood sugar levels is “glycemia”. It finds articles that mention glucose, insulin, or HbA1c, all semantically related terms.
The system also includes articles about diabetes, a condition semantically associated with high blood sugar levels. It recognizes that diet and exercise are lifestyle factors that can affect diabetes risk.
How Do Semantic Search Engines Work?
Semantic search engines are powered by vector search, thus enabling to deliver relevant search results that are based on the context, as well as the and intent. Vector search encodes details of searchable information into fields of related terms or items, or vectors, and then compares vectors to determine which are most similar.
Target Audience
This session is suitable for:
- Data Engineers
- Machine Learning Engineers
- Other data professionals who are interested in learning about semantic search
- University students who are interested in building their portfolio in support of their job search
Upcoming Sessions
| When | Where | |||
|---|---|---|---|---|
| Sunday, 18 Feb 2024 | 1 - 5 PM 🌏 | Online | Register |
Please note:
- All times above are expressed in SGT. Click on the 🌏 icon next to sessions of interest to get your location’s corresponding date and time.
- For reference, 1 PM SGT is 10:30 AM (Bangalore), 12 Noon (Jakarta), 10 AM (Lahore)
Other Dates & Times
👋 I am interested but don’t see a date or time that works for me.
You can indicate your preferred date & time when you register.
What’s Next?
What’s next after attending this session?
This session covers content that is part of the following tracks:
Check out the complete list of upcoming sessions.
Alternatively, if you are a hands-on creator, check out the upcoming Hands-On Learning Sessions.