CLAIM leverages advanced natural language models to improve methods of data imputation in tabular datasets. It transforms datasets into natural language formats that align closely with the strengths of large language models. This innovative technique generates missing value descriptors and fine-tunes LLMs on the enriched datasets. Evaluations highlight CLAIM’s superior performance over conventional imputation methods, especially emphasizing the significance of contextual accuracy. Investigations reveal that context-specific descriptors lead to better outcomes, enhancing the reliability of data analysis and machine learning models when addressing missing data challenges.
CLAIM utilizes contextually relevant natural language descriptors to fill missing values, transforming datasets into formats that align with LLMs' capabilities for improved performance.
Evaluations show that CLAIM outperforms traditional imputation techniques and emphasizes the necessity of context-specific descriptors to enhance reliability in data analysis.
Collection
[
|
...
]