An RNN Model for Generating Sentences with a Desired Word at a Desired Position
Generating sentences with a desired word is useful in many natural language processing tasks. State-of-the-art recurrent neural network (RNN)-based models mainly generate sentences in a left-to-right manner, which does not allow explicit and direct constraints on the words at arbitrary positions in...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Faculty of Mechanical Engineering in Slavonski Brod, Faculty of Electrical Engineering in Osijek, Faculty of Civil Engineering in Osijek
2020-01-01
|
| Series: | Tehnički Vjesnik |
| Subjects: | |
| Online Access: | https://hrcak.srce.hr/file/340463 |
| _version_ | 1797207269448876032 |
|---|---|
| author | Tianbao Song Jingbo Sun Yinbing Zhang Weiming Peng* Jihua Song |
| author_facet | Tianbao Song Jingbo Sun Yinbing Zhang Weiming Peng* Jihua Song |
| author_sort | Tianbao Song |
| collection | DOAJ |
| description | Generating sentences with a desired word is useful in many natural language processing tasks. State-of-the-art recurrent neural network (RNN)-based models mainly generate sentences in a left-to-right manner, which does not allow explicit and direct constraints on the words at arbitrary positions in a sentence. To address this issue, we propose a generative model of sentences named Coupled-RNN. We employ two RNN's to generate sentences backwards and forwards respectively starting from a desired word, and inject position embeddings into the model to solve the problem of position information loss. We explore two coupling mechanisms to optimize the reconstruction loss globally. Experimental results demonstrate that Coupled-RNN can generate high quality sentences that contain a desired word at a desired position. |
| first_indexed | 2024-04-24T09:20:14Z |
| format | Article |
| id | doaj.art-50084f13618046659ac946bb2031848f |
| institution | Directory Open Access Journal |
| issn | 1330-3651 1848-6339 |
| language | English |
| last_indexed | 2024-04-24T09:20:14Z |
| publishDate | 2020-01-01 |
| publisher | Faculty of Mechanical Engineering in Slavonski Brod, Faculty of Electrical Engineering in Osijek, Faculty of Civil Engineering in Osijek |
| record_format | Article |
| series | Tehnički Vjesnik |
| spelling | doaj.art-50084f13618046659ac946bb2031848f2024-04-15T16:03:43ZengFaculty of Mechanical Engineering in Slavonski Brod, Faculty of Electrical Engineering in Osijek, Faculty of Civil Engineering in OsijekTehnički Vjesnik1330-36511848-63392020-01-01271818810.17559/TV-20190929153200An RNN Model for Generating Sentences with a Desired Word at a Desired PositionTianbao Song0Jingbo Sun1Yinbing Zhang2Weiming Peng*3Jihua Song4Beijing Normal University, No. 19, Xinjiekouwai St, Haidian District, Beijing, 100875, P. R. ChinaBeijing Normal University, No. 19, Xinjiekouwai St, Haidian District, Beijing, 100875, P. R. ChinaBeijing Normal University, No. 19, Xinjiekouwai St, Haidian District, Beijing, 100875, P. R. ChinaBeijing Normal University, No. 19, Xinjiekouwai St, Haidian District, Beijing, 100875, P. R. ChinaBeijing Normal University, No. 19, Xinjiekouwai St, Haidian District, Beijing, 100875, P. R. ChinaGenerating sentences with a desired word is useful in many natural language processing tasks. State-of-the-art recurrent neural network (RNN)-based models mainly generate sentences in a left-to-right manner, which does not allow explicit and direct constraints on the words at arbitrary positions in a sentence. To address this issue, we propose a generative model of sentences named Coupled-RNN. We employ two RNN's to generate sentences backwards and forwards respectively starting from a desired word, and inject position embeddings into the model to solve the problem of position information loss. We explore two coupling mechanisms to optimize the reconstruction loss globally. Experimental results demonstrate that Coupled-RNN can generate high quality sentences that contain a desired word at a desired position.https://hrcak.srce.hr/file/340463desired wordlexically constrainedRNNsentence generation |
| spellingShingle | Tianbao Song Jingbo Sun Yinbing Zhang Weiming Peng* Jihua Song An RNN Model for Generating Sentences with a Desired Word at a Desired Position Tehnički Vjesnik desired word lexically constrained RNN sentence generation |
| title | An RNN Model for Generating Sentences with a Desired Word at a Desired Position |
| title_full | An RNN Model for Generating Sentences with a Desired Word at a Desired Position |
| title_fullStr | An RNN Model for Generating Sentences with a Desired Word at a Desired Position |
| title_full_unstemmed | An RNN Model for Generating Sentences with a Desired Word at a Desired Position |
| title_short | An RNN Model for Generating Sentences with a Desired Word at a Desired Position |
| title_sort | rnn model for generating sentences with a desired word at a desired position |
| topic | desired word lexically constrained RNN sentence generation |
| url | https://hrcak.srce.hr/file/340463 |
| work_keys_str_mv | AT tianbaosong anrnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT jingbosun anrnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT yinbingzhang anrnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT weimingpeng anrnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT jihuasong anrnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT tianbaosong rnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT jingbosun rnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT yinbingzhang rnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT weimingpeng rnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition AT jihuasong rnnmodelforgeneratingsentenceswithadesiredwordatadesiredposition |