Prompt de Resumen en Cadena de Densidad

[ { "Missing_Entities": "article; entities; summary", "Denser_Summary": "This article discusses the process of generating summaries. It identifies missing entities from an article. These entities are relevant, specific, novel, and faithful. The process involves creating denser summaries of identical length. The goal is to produce concise, self-contained summaries." }, { "Missing_Entities": "verbose language; fillers; ~80 words", "Denser_Summary": "The article describes a method for iterative summarization. It starts with a long, non-specific summary using verbose language and fillers like 'this article discusses' to reach ~80 words. Missing entities are identified and added. Each new summary must be denser but same length, covering all previous entities." }, { "Missing_Entities": "fusion; compression; removal", "Denser_Summary": "The method involves rewriting previous summaries to improve flow. Techniques include fusion, compression, and removal of uninformative phrases. This makes space for additional entities. The summaries become highly dense yet self-contained. They must be easily understood without the original article." }, { "Missing_Entities": "never drop; add fewer; 5 steps", "Denser_Summary": "The process repeats 5 times. Missing entities are 1-3 informative items, delimited by semicolons. They must be specific (5 words or fewer) and anywhere in the article. Never drop entities from previous summaries. If space is tight, add fewer new entities." }, { "Missing_Entities": "JSON; list of dictionaries; keys", "Denser_Summary": "The final output is in JSON format. It is a list of 5 dictionaries. Each dictionary has keys 'Missing_Entities' and 'Denser_Summary'. The summaries must be identical in word count. The process ensures entity density and faithfulness to the article." } ]