Abstract
BABOONS (BlAck BOx Optimization of Natural language data Summaries) optimizes text data summaries for an arbitrary, user-defined utility function. Primarily, it targets scenarios in which utility is evaluated via large language models. Users describe their utility function in natural language or provide a model, trained to score text summaries in a specific domain.
BABOONS uses reinforcement learning to explore the space of possible descriptions. In each iteration, BABOONS generates summaries and evaluates their utility. To reduce data processing overheads during summary generation, BABOONS uses a proactive processing strategy that dynamically merges current with likely future queries for efficient processing. Also, BABOONS supports scenario-specific sampling and batch processing strategies. These mechanisms allow to scale processing to large data and item sets. The experiments show that BABOONS scales significantly better than baselines. Also, they show that summaries generated by BABOONS receive higher average grades from users in a large survey.
Publisher
Association for Computing Machinery (ACM)
Subject
General Earth and Planetary Sciences,Water Science and Technology,Geography, Planning and Development
Reference72 articles.
1. 2021. https://www.upcounsel.com/what-is-a-product-description. 2021. https://www.upcounsel.com/what-is-a-product-description.
2. DIFF: a relational interface for large-scale data explanation
3. Data summarization: a survey
4. Detecting outlying properties of exceptional objects
5. Us vs. them: the minefield of comparative ads;Buchanan B.;Harvard Business Review,1989
Cited by
4 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献
1. A review of reinforcement learning for natural language processing and applications in healthcare;Journal of the American Medical Informatics Association;2024-08-29
2. Large Language Models: Principles and Practice;2024 IEEE 40th International Conference on Data Engineering (ICDE);2024-05-13
3. Demonstrating NaturalMiner: Searching Large Data Sets for Abstract Patterns Described in Natural Language;Companion of the 2023 International Conference on Management of Data;2023-06-04
4. From BERT to GPT-3 codex;Proceedings of the VLDB Endowment;2022-08