Keyword Extraction Using Graph
Centrality and WordNet
Chhavi Sharma, Minni Jain and Ayush Aggarwal
1 Introduction
Keywords are a short representation of a document which summarizes the important
content of the document into certain selective words which are either picked from
the document or constructed using the underlying sense of the document. Keyword
extraction deals with automatically picking up the underlying text from the document
which can best describe the senses of it.
Keyword extraction is a very important field of study for NLP purposes because it
serves great many functionality. The amount of text being produced daily is increas-
ing exponentially and it becomes impossible for the reader to sift through it, and
hence, keyword presents an alternative of understanding for the user to determine
whether the document deserves the time or not. Keywords are also used in informa-
tion retrieval systems [1] such as search engines to better index the documents and
provide document management and categorization.
A lot of research has been done in the field of keyword extraction from supervised
to graph-based approach, but not much work has been done exploring the semantic
relatedness of the term and ranking them on the language tree to determine the central
words of the documents. In this paper, we present an unsupervised learning approach
to extract keywords from the document which are based on the semantic strength
of words in a particular document. We use WordNet [2] as our knowledge base to
C. Sharma · M. Jain (B ) · A. Aggarwal
Delhi Technological University, Shahbad Daulatpur, Bawana Road,
New Delhi 110042, Delhi, India
e-mail: minnijain91@gmail.com; minnijain@dtu.ac.in
C. Sharma
e-mail: sharma.chhavi96@gmail.com
A. Aggarwal
e-mail: aggarwal96ayush@gmail.com
© Springer Nature Singapore Pte Ltd. 2018
S. Chakraverty et al. (eds.), Towards Extensible and Adaptable
Methods in Computing, https://doi.org/10.1007/978-981-13-2348-5_27
363