Winda Kurnia Sari | Universitas Sriwijaya (original) (raw)

Uploads

Papers by Winda Kurnia Sari

Abstrak Fakultas Ilmu Komputer Universitas Sriwijaya adalah salah satu perguruan tinggi yang meny... more Abstrak Fakultas Ilmu Komputer Universitas Sriwijaya adalah salah satu perguruan tinggi yang menyadari pentingnya sebuah pendokumentasian dari data dan informasi bagi keberlangsungan kegiatan perguruan tinggi. Saat ini pendokumentasian pengetahuan tentang teknis komputer dan jaringan yang ada di FASILKOM belum terstruktur sehingga berdampak pada kegiatan fakultas yang terasa tidak efektif. Berdasarkan hasil analisa, terdapat banyak knowledge penting dibagian teknisi komputer dan jaringan yang fungsinya untuk menunjang kegiatan perguruan tinggi. Metodologi yang digunakan pada penelitian ini merujuk ke metodologi knowledge management yang dikembangkan oleh Amrit Tiwana. Pada metodologi ini terdapat 4 tahap utama, yaitu: persiapan dan evaluasi infrastruktur, analisis dan desain knowledge management, pengembangan knowledge management dan evaluasi. Knowledge management terasa sangat dibutuhkan pada saat ini untuk memfasilitasi masalah pendokumentasian dan penggunaannya serta meningkatkan...

Neural network-based applications are recently shown promising results for text classification. H... more Neural network-based applications are recently shown promising results for text classification. However, it is still challenging for the model to contemplate local features and word contingent on the information of the sentence. This work proposed a deep learning approach to generate a more precise sentence that leverages the preceding texts when classifying a subsequent one. One of the deep learning methods used is Recurrent Neural Network (RNN) with the architecture Long Short-Term Memory (LSTM). By training four variant models of 1-layer LSTM for each balance dataset in pre-processing process with 20,000, 25,000, 30,000, 35,000, 40,000, and 45,000 using optimizer Adam and RMSProp. The results show that; first, the more data input, the higher the accuracy it gets and the second is Adam can perform better as optimizer than RMSProp in this research. The highest Precision, Recall, and F1-score obtain are 97.

2019 International Conference on Electrical Engineering and Computer Science (ICECOS)

Text classification usually has the basic problem of presenting data with very high dimension, so... more Text classification usually has the basic problem of presenting data with very high dimension, so that the model formed is usually hampered because (1) training time increases exponentially appropriate with the number of feature used and (2) Model has an increased risk of overfitting with a growing number of features. Recurrent Neural Network (RNN) is one of the most popular architectures used in natural language processing (NLP) since the recurrent structure is very suitable for long variable text processing. One of the deep learning methods proposed in this study is RNN with the application of the Long Short-Term Memory (LSTM) architecture. Long Short-Term Memory (LSTM) offers element which is expected to be able to record a feature of input, such as in natural language processing for English, an element record gender of the subject, other element records whether single or plural subject. These features will be found by LSTM itself in the training process. With variant word sequence features, the results of this study have the highest accuracy of 82.13%.

Jurnal Sistem Informasi, Dec 11, 2014

Jurnal Ilmiah Teknik Elektro Komputer dan Informatika

In the classification of traditional algorithms, problems of high features dimension and data spa... more In the classification of traditional algorithms, problems of high features dimension and data sparseness often occur when classifying text. Classifying text with traditional machine learning algorithms has high efficiency and stability characteristics. However, it has certain limitations concerning largescale dataset training. In this case, a multi-label text classification technique is needed to be able to group four labels from the news article dataset. Deep Learning is a proposed method for solving problems in text classification techniques. This experiment was conducted using one of the methods of Deep Learning Recurrent Neural Network with the application of the architecture of Long Short-Term Memory (LSTM). In this study, the model is based on trial and error experiments using LSTM and 300-dimensional word embedding features with Global Vector (GloVe). By tuning the parameters and comparing the eight proposed LSTM models with a largescale dataset, to show that LSTM with features GloVe can achieve good performance in text classification. The results show that text classification using LSTM with GloVe obtain the highest accuracy is in the sixth model with 95.17, the average precision, recall, and F1-score are 95. Besides, LSTM with the GloVe feature gets graphic results that are close to good-fit on average.

Abstrak Fakultas Ilmu Komputer Universitas Sriwijaya adalah salah satu perguruan tinggi yang meny... more Abstrak Fakultas Ilmu Komputer Universitas Sriwijaya adalah salah satu perguruan tinggi yang menyadari pentingnya sebuah pendokumentasian dari data dan informasi bagi keberlangsungan kegiatan perguruan tinggi. Saat ini pendokumentasian pengetahuan tentang teknis komputer dan jaringan yang ada di FASILKOM belum terstruktur sehingga berdampak pada kegiatan fakultas yang terasa tidak efektif. Berdasarkan hasil analisa, terdapat banyak knowledge penting dibagian teknisi komputer dan jaringan yang fungsinya untuk menunjang kegiatan perguruan tinggi. Metodologi yang digunakan pada penelitian ini merujuk ke metodologi knowledge management yang dikembangkan oleh Amrit Tiwana. Pada metodologi ini terdapat 4 tahap utama, yaitu: persiapan dan evaluasi infrastruktur, analisis dan desain knowledge management, pengembangan knowledge management dan evaluasi. Knowledge management terasa sangat dibutuhkan pada saat ini untuk memfasilitasi masalah pendokumentasian dan penggunaannya serta meningkatkan...

Neural network-based applications are recently shown promising results for text classification. H... more Neural network-based applications are recently shown promising results for text classification. However, it is still challenging for the model to contemplate local features and word contingent on the information of the sentence. This work proposed a deep learning approach to generate a more precise sentence that leverages the preceding texts when classifying a subsequent one. One of the deep learning methods used is Recurrent Neural Network (RNN) with the architecture Long Short-Term Memory (LSTM). By training four variant models of 1-layer LSTM for each balance dataset in pre-processing process with 20,000, 25,000, 30,000, 35,000, 40,000, and 45,000 using optimizer Adam and RMSProp. The results show that; first, the more data input, the higher the accuracy it gets and the second is Adam can perform better as optimizer than RMSProp in this research. The highest Precision, Recall, and F1-score obtain are 97.

2019 International Conference on Electrical Engineering and Computer Science (ICECOS)

Text classification usually has the basic problem of presenting data with very high dimension, so... more Text classification usually has the basic problem of presenting data with very high dimension, so that the model formed is usually hampered because (1) training time increases exponentially appropriate with the number of feature used and (2) Model has an increased risk of overfitting with a growing number of features. Recurrent Neural Network (RNN) is one of the most popular architectures used in natural language processing (NLP) since the recurrent structure is very suitable for long variable text processing. One of the deep learning methods proposed in this study is RNN with the application of the Long Short-Term Memory (LSTM) architecture. Long Short-Term Memory (LSTM) offers element which is expected to be able to record a feature of input, such as in natural language processing for English, an element record gender of the subject, other element records whether single or plural subject. These features will be found by LSTM itself in the training process. With variant word sequence features, the results of this study have the highest accuracy of 82.13%.

Jurnal Sistem Informasi, Dec 11, 2014

Jurnal Ilmiah Teknik Elektro Komputer dan Informatika

In the classification of traditional algorithms, problems of high features dimension and data spa... more In the classification of traditional algorithms, problems of high features dimension and data sparseness often occur when classifying text. Classifying text with traditional machine learning algorithms has high efficiency and stability characteristics. However, it has certain limitations concerning largescale dataset training. In this case, a multi-label text classification technique is needed to be able to group four labels from the news article dataset. Deep Learning is a proposed method for solving problems in text classification techniques. This experiment was conducted using one of the methods of Deep Learning Recurrent Neural Network with the application of the architecture of Long Short-Term Memory (LSTM). In this study, the model is based on trial and error experiments using LSTM and 300-dimensional word embedding features with Global Vector (GloVe). By tuning the parameters and comparing the eight proposed LSTM models with a largescale dataset, to show that LSTM with features GloVe can achieve good performance in text classification. The results show that text classification using LSTM with GloVe obtain the highest accuracy is in the sixth model with 95.17, the average precision, recall, and F1-score are 95. Besides, LSTM with the GloVe feature gets graphic results that are close to good-fit on average.