Skripsi

ANALISIS SENTIMEN ULASAN APLIKASI SHOPEE DI GOOGLE PLAY STORE MENGGUNAKAN METODE BI-LSTM DAN WORD2VEC

---

Shopee is one of the largest e-commerce platforms in Indonesia that provides various products and services. User reviews of the Shopee application are an important source of information for companies to understand user needs and experiences, in order to improve service quality. This research aims to conduct sentiment analysis on user reviews of the Shopee application with the Bidirectional Long Short-Term Memory (Bi-LSTM) and Word2Vec methods. Word2Vec is used to convert the words in the review into numerical vectors that reflect the meaning and context of each word, so that the model can understand the meaning and relationship between words more precisely and accurately. Meanwhile, Bidirectional Long Short-Term Memory (Bi-LSTM), as a recurrent neural network, analyzes the word order from two directions, so that a thorough understanding of the sentence context can be achieved. The dataset used consists of 50,000 reviews from the Google Play Store. Tests were conducted with 8 configuration scenarios to obtain optimal results. The best configuration involved 64 LSTM units, LSTM dropout 0.5, recurrent dropout 0.2, dense layer 64 neurons, layer dropout 0.5, learning rate 0.001, batch size 32, and 10 training epochs. The best model showed an accuracy of 86.62% on training data and 85.75% on validation data, while evaluation on test data resulted in an accuracy of 86.15%. The model also showed average values of macro precision, recall, and F1-Score of 77.42%, 67.93%, and 70.44%, as well as weighted precision, recall, and F1-Score of 85.05%, 86.15%, and 85.07% respectively.

---

Availability

Detail Information

Series Title

-

Call Number

T1630002024

Publisher

Indralaya : Prodi Sistem Informasi, Fakultas Ilmu Komputer Universitas Sriwijaya., 2024

Collation

xxv, 156 hlm.; ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

005.360 7

Content Type

Text

Media Type

unmediated

Carrier Type

other (computer)

Edition

-

Subject(s)

Program untuk Komputer Personal
Prodi Sistem Informasi

Specific Detail Info

-

Statement of Responsibility

MURZ

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment