Stefano Spindola, M. M. José, A. Oliveira, Flávio Nakasato Cação, F. G. Cozman
{"title":"蓝色亚马逊流域葡萄牙语问答系统中注意机制的可解释性","authors":"Stefano Spindola, M. M. José, A. Oliveira, Flávio Nakasato Cação, F. G. Cozman","doi":"10.5753/eniac.2021.18302","DOIUrl":null,"url":null,"abstract":"The Brazilian Exclusive Economic Zone, or the \"Blue Amazon\", with its extensive maritime area, is the primary means of transport for the country's foreign trade and is important due to its oil reserves, gas and other mineral resources, in addition to the significant influence on the Brazilian climate. We have manually built a question answering (QA) dataset based on crawled articles and have applied an off-the-shelf QA system based on a fine-tuned BERTimbau Model, achieving an F1-score of 47.0. More importantly, we explored how the proper visualization of attention weights can support helpful interpretations of the system's answers, which is critical in real environments.","PeriodicalId":318676,"journal":{"name":"Anais do XVIII Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2021)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":"{\"title\":\"Interpretability of Attention Mechanisms in a Portuguese-Based Question Answering System about the Blue Amazon\",\"authors\":\"Stefano Spindola, M. M. José, A. Oliveira, Flávio Nakasato Cação, F. G. Cozman\",\"doi\":\"10.5753/eniac.2021.18302\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The Brazilian Exclusive Economic Zone, or the \\\"Blue Amazon\\\", with its extensive maritime area, is the primary means of transport for the country's foreign trade and is important due to its oil reserves, gas and other mineral resources, in addition to the significant influence on the Brazilian climate. We have manually built a question answering (QA) dataset based on crawled articles and have applied an off-the-shelf QA system based on a fine-tuned BERTimbau Model, achieving an F1-score of 47.0. More importantly, we explored how the proper visualization of attention weights can support helpful interpretations of the system's answers, which is critical in real environments.\",\"PeriodicalId\":318676,\"journal\":{\"name\":\"Anais do XVIII Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2021)\",\"volume\":\"23 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2021-11-29\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"1\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Anais do XVIII Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2021)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.5753/eniac.2021.18302\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Anais do XVIII Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2021)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.5753/eniac.2021.18302","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Interpretability of Attention Mechanisms in a Portuguese-Based Question Answering System about the Blue Amazon
The Brazilian Exclusive Economic Zone, or the "Blue Amazon", with its extensive maritime area, is the primary means of transport for the country's foreign trade and is important due to its oil reserves, gas and other mineral resources, in addition to the significant influence on the Brazilian climate. We have manually built a question answering (QA) dataset based on crawled articles and have applied an off-the-shelf QA system based on a fine-tuned BERTimbau Model, achieving an F1-score of 47.0. More importantly, we explored how the proper visualization of attention weights can support helpful interpretations of the system's answers, which is critical in real environments.