Linguistically summarizing hierarchical data: The case of statistical territorial data

Linguistic summaries are suitable for summing up information from the large amount of data by simulating human reasoning. Summarized information helps users to focus on a relevant higher hierarchical level by asking short questions about data on respective lower levels. Territorial units are organized in a hierarchy from the top level to the lowest level. Two main parts of linguistic summaries are predicate (consisting of summarizer and restriction) and relative quantifier, both expressed as fuzzy sets. Therefore, summarized information is expressed as a quantified sentence of natural language. Quality measures ensure that only summaries which bear relevant information and are not calculated on outliers are valuable. Furthermore, benefits of linguistic summaries are demonstrated on illustrative examples on real territorial units' data. Finally, perspectives for further research are touched.