Matches in SemOpenAlex for { <https://semopenalex.org/work/W4380047099> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W4380047099 endingPage "e1425" @default.
- W4380047099 startingPage "e1425" @default.
- W4380047099 abstract "Aspect-based sentiment analysis tasks are well researched in English. However, we find such research lacking in the context of the Arabic language, especially with reference to aspect category detection. Most of this research is focusing on supervised machine learning methods that require the use of large, labeled datasets. Therefore, the aim of this research is to implement a semi-supervised self-training approach which utilizes a noisy student framework to enhance the capability of a deep learning model, AraBERT v02. The objective is to perform aspect category detection on both the SemEval 2016 hotel review dataset and the Hotel Arabic-Reviews Dataset (HARD) 2016. The four-step framework firstly entails developing a teacher model that is trained on the aspect categories of the SemEval 2016 labeled dataset. Secondly, it generates pseudo labels for the unlabeled HARD dataset based on the teacher model. Thirdly, it creates a noisy student model that is trained on the combined datasets (∼1 million sentences). The aim is to minimize the combined cross entropy loss. Fourthly, an ensembling of both teacher and student models is carried out to enhance the performance of AraBERT. Findings indicate that the ensembled teacher-student model demonstrates a 0.3% improvement in its micro F1 over the initial noisy student implementation, both in predicting the Aspect Categories in the combined datasets. However, it has achieved a 1% increase over the micro F1 of the teacher model. These results outperform both baselines and other deep learning models discussed in the related literature." @default.
- W4380047099 created "2023-06-10" @default.
- W4380047099 creator A5008859517 @default.
- W4380047099 creator A5010707924 @default.
- W4380047099 creator A5036579361 @default.
- W4380047099 date "2023-06-08" @default.
- W4380047099 modified "2023-09-27" @default.
- W4380047099 title "A semi supervised approach to Arabic aspect category detection using Bert and teacher-student model" @default.
- W4380047099 cites W2106401878 @default.
- W4380047099 cites W2166706824 @default.
- W4380047099 cites W2556605533 @default.
- W4380047099 cites W2591922431 @default.
- W4380047099 cites W2769317996 @default.
- W4380047099 cites W2770803436 @default.
- W4380047099 cites W2802301063 @default.
- W4380047099 cites W2895547478 @default.
- W4380047099 cites W2903000316 @default.
- W4380047099 cites W2916227226 @default.
- W4380047099 cites W2971855175 @default.
- W4380047099 cites W2984353870 @default.
- W4380047099 cites W2993843842 @default.
- W4380047099 cites W2998019206 @default.
- W4380047099 cites W3037469642 @default.
- W4380047099 cites W3077780682 @default.
- W4380047099 cites W3097075955 @default.
- W4380047099 cites W3101155479 @default.
- W4380047099 cites W3180048861 @default.
- W4380047099 cites W3208010654 @default.
- W4380047099 cites W3209984204 @default.
- W4380047099 cites W4206355005 @default.
- W4380047099 doi "https://doi.org/10.7717/peerj-cs.1425" @default.
- W4380047099 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37346563" @default.
- W4380047099 hasPublicationYear "2023" @default.
- W4380047099 type Work @default.
- W4380047099 citedByCount "0" @default.
- W4380047099 crossrefType "journal-article" @default.
- W4380047099 hasAuthorship W4380047099A5008859517 @default.
- W4380047099 hasAuthorship W4380047099A5010707924 @default.
- W4380047099 hasAuthorship W4380047099A5036579361 @default.
- W4380047099 hasBestOaLocation W43800470991 @default.
- W4380047099 hasConcept C119857082 @default.
- W4380047099 hasConcept C136389625 @default.
- W4380047099 hasConcept C138885662 @default.
- W4380047099 hasConcept C151730666 @default.
- W4380047099 hasConcept C154945302 @default.
- W4380047099 hasConcept C162324750 @default.
- W4380047099 hasConcept C187736073 @default.
- W4380047099 hasConcept C204321447 @default.
- W4380047099 hasConcept C2779343474 @default.
- W4380047099 hasConcept C2780451532 @default.
- W4380047099 hasConcept C41008148 @default.
- W4380047099 hasConcept C41895202 @default.
- W4380047099 hasConcept C44572571 @default.
- W4380047099 hasConcept C50644808 @default.
- W4380047099 hasConcept C86803240 @default.
- W4380047099 hasConcept C96455323 @default.
- W4380047099 hasConceptScore W4380047099C119857082 @default.
- W4380047099 hasConceptScore W4380047099C136389625 @default.
- W4380047099 hasConceptScore W4380047099C138885662 @default.
- W4380047099 hasConceptScore W4380047099C151730666 @default.
- W4380047099 hasConceptScore W4380047099C154945302 @default.
- W4380047099 hasConceptScore W4380047099C162324750 @default.
- W4380047099 hasConceptScore W4380047099C187736073 @default.
- W4380047099 hasConceptScore W4380047099C204321447 @default.
- W4380047099 hasConceptScore W4380047099C2779343474 @default.
- W4380047099 hasConceptScore W4380047099C2780451532 @default.
- W4380047099 hasConceptScore W4380047099C41008148 @default.
- W4380047099 hasConceptScore W4380047099C41895202 @default.
- W4380047099 hasConceptScore W4380047099C44572571 @default.
- W4380047099 hasConceptScore W4380047099C50644808 @default.
- W4380047099 hasConceptScore W4380047099C86803240 @default.
- W4380047099 hasConceptScore W4380047099C96455323 @default.
- W4380047099 hasLocation W43800470991 @default.
- W4380047099 hasLocation W43800470992 @default.
- W4380047099 hasOpenAccess W4380047099 @default.
- W4380047099 hasPrimaryLocation W43800470991 @default.
- W4380047099 hasRelatedWork W2022144023 @default.
- W4380047099 hasRelatedWork W2351428524 @default.
- W4380047099 hasRelatedWork W2386544544 @default.
- W4380047099 hasRelatedWork W2750880011 @default.
- W4380047099 hasRelatedWork W3024381485 @default.
- W4380047099 hasRelatedWork W3114038635 @default.
- W4380047099 hasRelatedWork W32283444 @default.
- W4380047099 hasRelatedWork W4226172683 @default.
- W4380047099 hasRelatedWork W4295267149 @default.
- W4380047099 hasRelatedWork W31012516 @default.
- W4380047099 hasVolume "9" @default.
- W4380047099 isParatext "false" @default.
- W4380047099 isRetracted "false" @default.
- W4380047099 workType "article" @default.