Title :
Keyword extraction from abstracts and titles
Author_Institution :
Sam Houston State Univ., Huntsville
Abstract :
Keywords are very important for any academic paper. We propose the Perceptron Training Rule for keyword extraction from titles and abstracts. We present a system for generating keywords which relies on weights of words in a sentence. The system generates keywords from academic research articles by selecting the most relevant keywords. We compare the keywords generated by our system and those generated by cluster analysis to the keywords given by the authors and analyze the results based on full-keyword matches, partial-keyword matches and no-keyword matches.
Keywords :
data mining; text analysis; abstract keyword extraction; academic research articles; cluster analysis; full-keyword matches; keyword generation; no-keyword matches; partial-keyword matches; perceptron training rule; relevant keyword selection; title keyword extraction; Abstracts; Data mining; Frequency; Humans; Learning systems; Machine learning; Natural language processing; Statistical analysis; Support vector machines; Web pages;
Conference_Titel :
Southeastcon, 2008. IEEE
Conference_Location :
Huntsville, AL
Print_ISBN :
978-1-4244-1883-1
Electronic_ISBN :
978-1-4244-1884-8
DOI :
10.1109/SECON.2008.4494366
