The objective of automatic keyphrase extraction is
to generate keyphrases for large number of documents. A
weakness of earlier keyphrase extraction algorithms is that
occasionally they have lesser coherence among the extracted
keyphrases. This paper examines the effect of injecting the
domain information of the document to the ranking phase of
automatic keyphrase extraction. The proposed method utilizes
the statistical similarity of the domain between the document
and the automatically extracted keyphrases as the criteria for
ranking the keyphrases. The method is evaluated on baseline
as well as advanced methods like KEA and resulted in a
considerable amount of growth in accuracy. To demonstrate
the feasibility of this approach, a naive implementation is also
provided. The method has the potential to be widely applicable
in all Keyphrase extraction algorithms.