This paper reflects a thughful study about the use of natural language in automatic indexing systmes, from Data Bases and the Digital Management of Documents to full-text indexing of scientific electronic publications indexed in the Web. It offers a theoric approach to automatic intexing, pointing out the role that natural language (no controled) has demonstrated throughout its evolution. It also shows the last trends of automatic indexing founded in knowledge bases. The research finishes including the analysis of main features of new software for full-text automatic indexing, extrancting rigourous conclusions when comparing automatic indexing versus human indexing and planning new challenges for the future research in this field when the full-text electronic information is exponential and indexing by humans is really expensive.

Se plantea una aproximación teórica a la indización automática, poniendo de relieve el papel que ha desempeñado el lenguaje natural, no controlado, en su evolución, señalando sobre todo las últimas tendencias en indización automatizada fundamentadas en bases de conocimiento.