Natural Language Processing for Detecting Forward Reference in a Document  

Oleh:

Siahaan, Daniel ; Umami, Izzatul

Jenis: Article from Journal - ilmiah nasional - terakreditasi DIKTI
Dalam koleksi: IPTEK: The Journal for Technology and Science vol. 23 no. 4 (Nov. 2012), page 138-142.
Topik: Forward Reference; Natural Language Processing; Term; Forward Reference; Istilah; Pemrosesan Bahasa Alamiah
Fulltext: 99-192-1-PB.pdf (809.11KB)

Isi artikel

Meyer’s seven sins have been recognized as types of mistakes that a requirements specialist are often fallen to when specifying requirements. Such mistakes play a significant role in plunging a project into failure. Many researchers were focusing in ambiguity and contradiction type of mistakes. Other types of mistakes have been given less attentions. Those mistakes often happened in reality and may equally costly as the first two mistakes. This paper introduces an approach to detect forward reference. It traverses through a requirements document, extracts, and processes each statement. During the statement extraction, any terms that may reside in the statement is also extracted. Based on certain rules which utilize POS patterns, the statement is classified as a term definition or not. For each term definition, a term is added to a list of defined terms. At the same time, every time a new term is found in a statement, it is check against the list of defined terms. If it is not found, then the requirements statement is classified as statement with forward reference. The experimentation on 30 requirements documents from various domains of software project shows that the approach has considerably almost perfect agreement with domain expert in detecting forward reference, given 0.83 kappa index value.

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali