Imagine you are preparing a dataset to fine-tune a large language model to summarize complex scientific research papers into plain language for a general audience. The initial dataset consists of thousands of full-text academic articles scraped from various online repositories. Propose a set of three distinct rule-based filters (heuristics) you would apply to this dataset to improve its quality for the specified task. For each filter, you must:

1. Clearly state the rule.
2. Justify why this rule is necessary, explaining the specific data quality issue it targets and its potential impact on the model's final performance.