Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation
Tree-LSTM algorithm accommodates tree structure processing to extract information outside the linear sequence pattern. The use of Tree-LSTM in text generation problems requires the help of an external parser at each generation iteration. Developing a good parser demands the representation of complex...
| Main Authors: | , , |
|---|---|
| Format: | Other |
| Published: |
ACM Transactions on Asian and Low-Resource Language Information Processing
2022
|
| Subjects: |
| _version_ | 1797037856806404096 |
|---|---|
| author | Hastuti, Rochana Prih Suyanto, Yohanes Sari, Anny Kartika |
| author_facet | Hastuti, Rochana Prih Suyanto, Yohanes Sari, Anny Kartika |
| author_sort | Hastuti, Rochana Prih |
| collection | UGM |
| description | Tree-LSTM algorithm accommodates tree structure processing to extract information outside the linear sequence pattern. The use of Tree-LSTM in text generation problems requires the help of an external parser at each generation iteration. Developing a good parser demands the representation of complex features and relies heavily on the grammar of the corpus. The limited corpus results in an insufficient number of vocabs for a grammar-based parser, making it less natural to link the text generation process. This research aims to solve the problem of limited corpus by proposing the use of a Reinforcement Learning algorithm in the formation of constituency trees, which link the sentence generation process given a seed phrase as the input in the Tree-LSTM model. The tree production process is modeled as a Markov's decision process, where a set of states consists of word embedding vectors, and a set of actions of {Shift, Reduce}. The Deep Q-Network model as an approximator of the Q-Learning algorithm is trained to obtain optimal weights in representing the Q-value function.The test results on perplexity-based evaluation show that the proposed Tree-LSTM and Q-Learning combination model achieves values 9.60 and 4.60 for two kinds of corpus with 205 and 1,000 sentences, respectively, better than the Shift-All model. Human evaluation of Friedman test and posthoc analysis showed that all five respondents tended to give the same assessment for the combination model of Tree-LSTM and Q-Learning, which on average outperforms two other nongrammar models, i.e., Shift-All and Reduce-All. © 2022 Copyright held by the owner/author(s). Publication rights licensed to ACM. |
| first_indexed | 2024-03-14T00:10:28Z |
| format | Other |
| id | oai:generic.eprints.org:284475 |
| institution | Universiti Gadjah Mada |
| last_indexed | 2024-03-14T00:10:28Z |
| publishDate | 2022 |
| publisher | ACM Transactions on Asian and Low-Resource Language Information Processing |
| record_format | dspace |
| spelling | oai:generic.eprints.org:2844752024-01-02T08:46:46Z https://repository.ugm.ac.id/284475/ Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation Hastuti, Rochana Prih Suyanto, Yohanes Sari, Anny Kartika Information and Computing Sciences Tree-LSTM algorithm accommodates tree structure processing to extract information outside the linear sequence pattern. The use of Tree-LSTM in text generation problems requires the help of an external parser at each generation iteration. Developing a good parser demands the representation of complex features and relies heavily on the grammar of the corpus. The limited corpus results in an insufficient number of vocabs for a grammar-based parser, making it less natural to link the text generation process. This research aims to solve the problem of limited corpus by proposing the use of a Reinforcement Learning algorithm in the formation of constituency trees, which link the sentence generation process given a seed phrase as the input in the Tree-LSTM model. The tree production process is modeled as a Markov's decision process, where a set of states consists of word embedding vectors, and a set of actions of {Shift, Reduce}. The Deep Q-Network model as an approximator of the Q-Learning algorithm is trained to obtain optimal weights in representing the Q-value function.The test results on perplexity-based evaluation show that the proposed Tree-LSTM and Q-Learning combination model achieves values 9.60 and 4.60 for two kinds of corpus with 205 and 1,000 sentences, respectively, better than the Shift-All model. Human evaluation of Friedman test and posthoc analysis showed that all five respondents tended to give the same assessment for the combination model of Tree-LSTM and Q-Learning, which on average outperforms two other nongrammar models, i.e., Shift-All and Reduce-All. © 2022 Copyright held by the owner/author(s). Publication rights licensed to ACM. ACM Transactions on Asian and Low-Resource Language Information Processing 2022 Other NonPeerReviewed Hastuti, Rochana Prih and Suyanto, Yohanes and Sari, Anny Kartika (2022) Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation. ACM Transactions on Asian and Low-Resource Language Information Processing. https://dl.acm.org/doi/abs/10.1145/3490501 10.1145/3490501 |
| spellingShingle | Information and Computing Sciences Hastuti, Rochana Prih Suyanto, Yohanes Sari, Anny Kartika Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation |
| title | Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation |
| title_full | Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation |
| title_fullStr | Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation |
| title_full_unstemmed | Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation |
| title_short | Q-Learning for Shift-Reduce Parsing in Indonesian Tree-LSTM-Based Text Generation |
| title_sort | q learning for shift reduce parsing in indonesian tree lstm based text generation |
| topic | Information and Computing Sciences |
| work_keys_str_mv | AT hastutirochanaprih qlearningforshiftreduceparsinginindonesiantreelstmbasedtextgeneration AT suyantoyohanes qlearningforshiftreduceparsinginindonesiantreelstmbasedtextgeneration AT sariannykartika qlearningforshiftreduceparsinginindonesiantreelstmbasedtextgeneration |