Skripsi

ANALISIS KINERJA MODEL KEYPHRASE GENERATION BERBASIS PARADIGMA ONE2SET PADA ARTIKEL BERITA

---

Penilaian

---

The rapid growth of digital content makes online news management increasingly challenging, as vast document streams demand compact representations of core information. Keyphrases provide concise lexical surrogates for main topics and support indexing, retrieval, and summarization, yet limited annotations motivate automatic systems that can predict both present and absent keyphrases. This study develops a keyphrase generation system using a Transformer architecture, the SETTRANS model within the ONE2SET paradigm, which casts prediction as a set through control codes, K-step assignment, and bipartite matching to reduce order bias and duplication. The model is trained on 30,000 KPTimes articles. Experimental factors include assign_steps at 1, 2, and 3, learning rates at 0.0001 and 0.00003, and decoding strategies using greedy and beam search with beam size 2 and 5. Evaluation with the F1-Score shows the best configuration at assign_steps 3, learning rate 0.00003, and beam size 2, yielding F1@5 of 0.24, F1@10 of 0.20, and F1@M of 0.25. The results indicate that the larger K strengthens set alignment and broadens coverage without increasing duplication, the smaller learning rate stabilizes score calibration and improves overall consistency, and beam size 2 offers a balanced search that sustains accuracy while remaining time efficient and limiting redundant generations.

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2507006211	T185547	T1855472025	Central Library (Reference)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T1855472025

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xviii, 187 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

004.07

Content Type

Text

Media Type

unmediated

Carrier Type

other (computer)

Edition

-

Subject(s)

Prodi Teknik Informatika
Teknik Pengelolaan Berita

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment