Zamzami, Moh. Ilham Fakhri (2024) Ekstraksi Kebutuhan Nonfungsional dari Ulasan Pengguna ke Atribut Kualitas Perangkat Lunak Menggunakan Pendekatan Model Ensemble. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5025201275-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5025201275-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (3MB) | Request a copy |
Abstract
Rekayasa kebutuhan merupakan tahapan penting yang menghasilkan daftar kebutuhan sebagai acuan dalam pengembangan proyek perangkat lunak. Kebutuhan non-fungsional mencakup aspek-aspek yang tidak berkaitan langsung dengan fungsi spesifik perangkat lunak, tetapi mempengaruhi kinerjanya secara keseluruhan. Kebutuhan non-fungsional sulit untuk dianalisis, karena kebutuhan non-fungsional sering ditemukan tidak lengkap, tersembunyi, dan tercampur di dalam kalimat kebutuhan lain pada dokumen. Permasalahan tersebut menyebabkan aspek kebutuhan non-fungsional sering terlewatkan selama proses pengembangan sistem perangkat lunak. Atribut kebutuhan non-fungsional terbagi menjadi banyak kategori. Hal ini menjadi sebuah masalah dalam pengklasifikasian pada pembelajaran mesin.
Tugas akhir ini mengangkat permasalahan tersebut dan menawarkan solusi berupa aplikasi web untuk mengklasifikasikan teks ulasan pengguna berdasarkan atribut kebutuhan non-fungsional menggunakan model ensemble. Proses ekstrakasi melewati praproses data, vektorisasi data teks, dan klasifikasi dengan model ensemble. Praproses data terdiri dari case folding, numeric to text, remove special character, tokenisasi, dan penghapusan stopwords. vektorisasi data teks dilakukan oleh pretrained fasttext model. Proses klasifikasi menggunakan model ensemble yang dibandingkan diantara algoritma Random Forrest, Adaptive Boosting, Categorical Boosting, dan Extreme Gradient Boosting. Pengukuran pembandingan dilakukan berdasarkan nilai confusion matrix yang didapat dari masing-masing kombinasi. Model dengan kombinasi terbaik akan digunakan untuk pembuatan aplikasi web.
Kombinasi model ekstraksi terbaik didapatkan oleh model ensemble categorical boosting dengan kombinasi pretrained fasttext model dan teknik praproses tanpa lemmatisasi yang terdiri dari case folding, change numeric to text, remove special character, tokenisasi, dan penghapusan stopwords. Hasil evaluasi menyatakan nilai accuracy 83.18%, precision 77.17%, recall 60.6%, dan f1-score 67.93%. Hasil serta manfaat dari Tugas Akhir ini berupa aplikasi ekstraksi ulasan pengguna pada kebutuhan nonfungsional yang diharapkan dapat memperbesar peluang kesuksesan proyek rekayasa perangkat lunak.
========================================================================================================================
Requirements engineering is an important stage that produces a list of requirements as a reference in software project development. Non-functional requirements include aspects that are not directly related to the specific functions of the software, but affect its overall performance. Non-functional requirements are difficult to analyze, because non-functional requirements are often found incomplete, hidden, and mixed in other requirement sentences in the document. These problems cause non-functional requirement aspects to be overlooked during the software system development process. Non-functional requirement attributes are divided into many categories. This becomes a problem in machine learning classification.
This final project addresses these issues and offers a solution in the form of a web application to classify user review text based on non-functional requirement attributes using an ensemble model. The extraction process goes through data preprocessing, text data vectorization, and classification with ensemble models. Data preprocessing consists of case folding, numeric to text, remove special characters, tokenization, and stopwords removal. Text data vectorization is performed by a pretrained fasttext model. The classification process uses an ensemble model that is compared among Random Forrest, Adaptive Boosting, Categorical Boosting, and Extreme Gradient Boosting Algorithms. The comparison measurement is based on the confusion matrix value obtained from each combination. The model with the best combination will be used for web application development.
The best extraction model combination is obtained by the Categorical Boosting ensemble model with a combination of pretrained fasttext model and preprocessing without lemmatization techniques consisting of case folding, change numeric to text, remove special characters, tokenization, and stopwords removal. The evaluation results stated the accuracy value of 83.18%, precision 77.17%, recall 60.6%and f1-score 67.93%. The results and benefits of this Final Project are in the form of an application for extracting user reviews on non-functional requirements which are expected to increase the chances of success of software engineering projects.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Ensemble, Kebutuhan Non fungsional, Klasifikasi, Kualitas Perangkat Lunak, Ulasan Pengguna, Classification, Ensemble, Nonfunctional Requirement, Software Requirements, User Reviews |
| Subjects: | T Technology > T Technology (General) > T57.5 Data Processing |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Moh. Ilham Fakhri Zamzami |
| Date Deposited: | 31 Jul 2024 12:56 |
| Last Modified: | 31 Jul 2024 12:56 |
| URI: | http://repository.its.ac.id/id/eprint/110759 |
Actions (login required)
 |
View Item |