Summarizing Agent Using Recurrent Neural Network
Mohd Nazri Syah Sani
Summarizing Agent Using Recurrent Neural Network / Mohd Nazri Syah Bin Sani - 2017
Abstract in English "A report project submitted in partial fulfillment of the requirements for the award of Bachelor of Computer Science (Hons)." -- On t. p.
Project paper (Bachelor of Computer Science ) - University Malaysia of Computer Science and Engineering, 2017.
Text summarization has become an important tool for the past few years to minimize the amount of storage for text document in the database. Due to that, many researches has been done within this domain to solve the issue. Moreover, there are not many researches of text summarization for Malay text. Most of this researches focussing on extractive method. In this research, a model under recurrent neural network approach called sequence-to-sequence model with the combination of attention model from a TensorFlow library has been proposed to generate a summary from a single document in abstractive manner. This research is focussing on Malay text. This research has done a literature review of different abstractive summarization methods from previous researches. The benefits and limitations for each methods has been done too. This research conducts three different experimental setups to check for the accuracy of the model. Results for the three experimental setups are shown in graph. Each graph were compared to determine which experimental setup has the more accuracy compared to others. At the end, with the amount of datasets collected for this research, it is concluded that the accuracy is higher with a larger amount of datasets used to train the sequence-to-sequence model. It is hope in this research will give an idea of a bigger step to the researchers who are interested in this domain.
Summarizing Agent Using Recurrent Neural Network / Mohd Nazri Syah Bin Sani - 2017
Abstract in English "A report project submitted in partial fulfillment of the requirements for the award of Bachelor of Computer Science (Hons)." -- On t. p.
Project paper (Bachelor of Computer Science ) - University Malaysia of Computer Science and Engineering, 2017.
Text summarization has become an important tool for the past few years to minimize the amount of storage for text document in the database. Due to that, many researches has been done within this domain to solve the issue. Moreover, there are not many researches of text summarization for Malay text. Most of this researches focussing on extractive method. In this research, a model under recurrent neural network approach called sequence-to-sequence model with the combination of attention model from a TensorFlow library has been proposed to generate a summary from a single document in abstractive manner. This research is focussing on Malay text. This research has done a literature review of different abstractive summarization methods from previous researches. The benefits and limitations for each methods has been done too. This research conducts three different experimental setups to check for the accuracy of the model. Results for the three experimental setups are shown in graph. Each graph were compared to determine which experimental setup has the more accuracy compared to others. At the end, with the amount of datasets collected for this research, it is concluded that the accuracy is higher with a larger amount of datasets used to train the sequence-to-sequence model. It is hope in this research will give an idea of a bigger step to the researchers who are interested in this domain.