Seekr is a high-performance, scalable search engine built with Java that simulates core features of modern search engines, including crawling, indexing, ranking, and advanced query handling. It features a web interface built with React and a backend powered by Spring Boot.
The system is composed of several tightly integrated modules, each with a distinct role in the search pipeline:
Purpose: Automatically traverses the web starting from seed URLs, fetching and parsing HTML documents.
Features:
- URL normalization to avoid duplicate visits.
- Recursively extracts links and restricts crawl to HTML-only content.
- Respects
robots.txtdirectives to avoid unauthorized access. - Designed with multithreading for performance scalability.
- Crawl state persistence enables stopping/resuming without data loss.
Purpose: Constructs an efficient, searchable index from crawled documents.
Capabilities:
- Extracts terms with contextual weighting (title, headers, body).
- Stores the index persistently in an SQLite database for quick lookups.
- Supports incremental updates for newly fetched documents.
- Optimized to handle large document volumes (6000+ documents in under 2 minutes).
Purpose: Processes user queries and retrieves relevant documents from the index.
Highlights:
- Stems query terms to support partial and morphological matches (e.g., travel, traveling, traveler).
- Filters out noise/stopwords to improve precision.
- Prepares tokens for downstream phrase and ranking modules.
Phrase Matching:
- Supports exact-phrase retrieval using quotation marks.
Example:"machine learning"
Boolean Logic:
- Handles
AND,OR, andNOToperators with up to two operations per query.
Example:"machine learning" AND pythonorclimate NOT "climate change"
Purpose: Sorts search results based on relevance and document authority.
Ranking Factors:
- Relevance: Utilizes TF-IDF and structural context (e.g., title matches).
- Popularity: Implements a variant of PageRank to prioritize highly-linked pages.
Speed:
- First result rendered within 20–50 ms.
- Subsequent results in under 5 ms.
Purpose: Presents search results in a user-friendly, responsive format.
User Experience:
- Developed using React for a dynamic, single-page interface.
- Clean layout similar to popular search engines.
- Displays result snippets with highlighted query terms.
- Pagination support (e.g., 10 results per page across 10+ pages).
- Smart auto-suggestions as users type, based on popular queries.
Homepage:
Search Results Page:
Seekr is built using a modern and lightweight technology stack:
| 🔧 Layer | 🛠️ Technology | 📋 Description |
|---|---|---|
| Backend | Java | Core logic for crawling, indexing, and ranking |
| Framework | Spring Boot | REST API development and modular backend structure |
| Database | SQLite | Lightweight, file-based storage for indexing and metadata |
| Frontend | React | Interactive and responsive web interface for user search experience |
 |
 |
 |
 |
|---|---|---|---|
| Saleh Ahmed | Shady Mohamed | Saif Eldin | Mohamed Bahgat |