Using lexical chains for keyword extraction Presenter : Jian-Ren Chen Authors : Gonenc Ercan, Ilyas Cicekli * 2007.IPM

Outlines Motivation Objectives Methodology Experiments Conclusions Comments

Motivation It is a hard and time consuming task to assign keywords to documents. The automatic keyphrase extraction algorithms are limited with phrases that appear in the text.

Objectives We present a keyword extraction method such that it uses the features based on lexical chains in the selection of keywords for a text.

Methodology first occurrence position frequency last occurrence position

Methodology – Lexical chains WordNet - synonym、hyponym/hypernym more than one relationship up to 3 levels of depth identify nouns only consider individual words

Methodology – lexical chains Lexical chain score of a word: Direct Lexical chain score of a word: Lexical chain span score of a word: Direct lexical chain span score of a word: 59 (=7 · 7 + 10) 28 (=4 · 7)

Experiments - Datasets Corpus Name Texts in corpus Number of Training Documents Testing Total Journal Articles full texts 55 20 75 Global corpus abstracts 110 45 155 - extract 1, 5, 10, 15 keywords for each document

Experiments Lexical chain score of a word: Direct Lexical chain score of a word: Lexical chain span score of a word: Direct lexical chain span score of a word:

Experiments

Experiments

Conclusions According to the results that are obtained, the lexical chain features improve the precision significantly in the keyword extraction process.

Comments Advantages Lexical chains will more consider the semantic relationships Applications keyword extraction