Rangkohärenz

Rangkohärenz is a concept in information retrieval and natural language processing that measures the consistency of a set of documents with respect to the distribution of ranked topics or query terms. It quantifies how coherently a collection of top‑retrieved documents reflect a particular thematic or semantic grouping expressed by a query. The term is derived from the German words Rank (ranking) and Kohärenz (coherence), and has been used in computational linguistics research to evaluate search engines, clustering algorithms, and query expansion techniques.

In practice, rangkohärenz is computed by comparing the ranked list of documents returned by a retrieval system to a reference set that is known to be topic‑related. Statistical methods such as Jensen–Shannon divergence, Kullback–Leibler divergence, or correlation coefficients are applied to the probability distributions of term usage or latent topics in the two lists. A low divergence or high correlation indicates that the retrieval algorithm preserves topic coherence in its ranking; a high divergence suggests that the ranking is scattered across unrelated subjects.

The metric has been employed to benchmark search engines in academic studies, to validate entity‑recognition pipelines, and to assess the effectiveness of query reformulation strategies. By providing a quantitative measure of thematic integrity in ranked outputs, rangkohärenz helps researchers and practitioners fine‑tune retrieval models and evaluate the relevance of results in terms of conceptual consistency rather than individual document scores alone.