Dataset Overview

Dataset Information

• Name: Long Term Memorability of Advertisements (LAMBDA)
• License: MIT
• Task Categories:
  • Text Classification
  • Text Generation
  • Question Answering
• Tags:
  • Memorability
  • Long‑Term Memorability
  • Advertising Memorability

Dataset Structure

• Features:
  • video_id: Identifier of the sample, type int64
  • recall_score: Memorability score of the video, range 0–1, type float64
  • youtube_id: YouTube ID of the video, type string
  • ad_details: Scene features for each video, containing:
    • Audio: Audio, type string
    • Brand: Brand, type string
    • Duration: Duration, type string
    • Orientation: Orientation, type string
    • Pace: Pace, type string
    • Scenes: List of scene items, each with sub‑features:
      • Colors: Colors, type string
      • Description: Description, type string
      • Emotions: Emotions, type string
      • Number: Number, type string
      • Photography Style: Photography Style, type string
      • Tags: Tags, type string
      • Text Shown: Text Shown, type string
      • Tone: Tone, type string
      • Visual Complexity: Visual Complexity, type string
    • Title: Title, type string

Data Splits

• Training Set:
  • Sample count: 1,964
  • Bytes: 5,490,622.457169034
• Test Set:
  • Sample count: 219
  • Bytes: 612,243.5428309665

Dataset Sizes

• Download Size: 2,551,503 bytes
• Total Size: 6,102,866 bytes

Configuration

• Default Configuration:
  • Training path: data/train-*
  • Test path: data/test-*

Citation

plaintext @misc{s2024longtermadmemorabilityunderstanding, title={Long‑Term Ad Memorability: Understanding and Generating Memorable Ads}, author={Harini S I au2 and Somesh Singh and Yaman K Singla and Aanisha Bhattacharyya and Veeky Baths and Changyou Chen and Rajiv Ratn Shah and Balaji Krishnamurthy}, year={2024}, eprint={2309.00378}, archivePrefix={arXiv}, primaryClass={cs.CL}, url={https://arxiv.org/abs/2309.00378} }