Yolobit Search [portable]

| Feature | Traditional Search (Inverted Index) | Yolobit Search (Bitmap Index) | | :--- | :--- | :--- | | | Postings List (Linked List of Doc IDs) | Bit-Vector (Array of 0s and 1s) | | Query Complexity | $O(N)$ (Depends on list length) | $O(1)$ (Fixed width bitwise ops) | | Memory Usage | Variable (grows with content) | Fixed (depends on corpus size $N$) | | Boolean Logic | Merging lists (Iterative) | Bitwise ops (Parallel/SIMD) | | Best Use Case | Large, dynamic web corpora | Static, high-filtering datasets |

Inspired by the YOLO object detection algorithm, the ranker does not re-score the entire result set. It applies a pre-computed static rank (weight) during the bitwise retrieval phase, outputting the top $K$ results immediately. yolobit search

This architecture is ideal for static datasets, embedded systems, or real-time filtering applications where millisecond latency is unacceptable and microsecond performance is required. | Feature | Traditional Search (Inverted Index) |

# Set the bit at doc_id position to 1 self.index[token][doc_id] = True # Set the bit at doc_id position to 1 self