{"title":"Thai Paraphrasing Tool for Chatbot Intent Recognition Training","authors":"Darlyn Sirikasem, Siripen Pongpaichet","doi":"10.1109/ICSEC56337.2022.10049337","DOIUrl":null,"url":null,"abstract":"This paper introduced a novel tool to automatically generate Thai paraphrased sentences specialized for intent recognition training. This technique is devised to improve intent recognition accuracy and reduce the training time, especially during the creation of applications like Thai natural language-based games and chatbots. The switching if-conjunction clauses in question sentences and the thesaurus-based paraphrasing methods have been explored. For the evaluation, a group of participants used the prototype tool to develop Thai chatbots. Which were tested to recognize question-type messages in the Thai national identification card domain provided by another group. The results demonstrated that the proposed techniques increase the accuracy F1 score compared to the state-ofthe-art pattern-aided chatbot by approximately 39%. On average, expert chatbot developers scored 8.25 out of 10 points on the prototype tool satisfaction.","PeriodicalId":430850,"journal":{"name":"2022 26th International Computer Science and Engineering Conference (ICSEC)","volume":"76 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-12-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 26th International Computer Science and Engineering Conference (ICSEC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICSEC56337.2022.10049337","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
This paper introduced a novel tool to automatically generate Thai paraphrased sentences specialized for intent recognition training. This technique is devised to improve intent recognition accuracy and reduce the training time, especially during the creation of applications like Thai natural language-based games and chatbots. The switching if-conjunction clauses in question sentences and the thesaurus-based paraphrasing methods have been explored. For the evaluation, a group of participants used the prototype tool to develop Thai chatbots. Which were tested to recognize question-type messages in the Thai national identification card domain provided by another group. The results demonstrated that the proposed techniques increase the accuracy F1 score compared to the state-ofthe-art pattern-aided chatbot by approximately 39%. On average, expert chatbot developers scored 8.25 out of 10 points on the prototype tool satisfaction.