Speech act classification in Vietnamese utterance and its application in smart mobile voice interaction

We can observe the rapid development of the spoken humanmachine interface, thanks to the big progress in automatic recognition and text to speech technologies. Especially, the mobile virtual assistants are becoming more popular than ever. Recognizing user's intent in interaction is one of the biggest problems while the system is designed and is attracting the attention of researchers. Identifying utterance's speech act in an automated manner is important to reveal user's intention (such as informing, asking, requesting and expressing emotion), which can provide useful indicators to improve the performance of human-machine interaction. Automatic speech act classification has been studied in many languages such as English, Chinese, Slovakian, Arabic, but not yet in Vietnamese. In this paper, we present a speech act scheme suitable for Vietnamese utterances in mobile voice interaction. Also, we built a speech act classifier that achieved 84.36% accuracy. This classifier has been applied in our mobile virtual assistant for Vietnamese (VAV). The experiment result has demonstrated the ability of this classifier such as robust, compact and can work well on mobiles.

Title: Speech act classification in Vietnamese utterance and its application in smart mobile voice interaction
Authors: Ngo, T.-L.
Duong, Q.-V.
Pham, S.-B.
Phan, X.-H.
Keywords: Natural spoken language understanding, Speech act, Vietnamese spoken text processing, User intent identification
Issue Date: 2016
Publisher: Association for Computing Machinery
Citation: Scopus
Abstract: We can observe the rapid development of the spoken humanmachine interface, thanks to the big progress in automatic recognition and text to speech technologies. Especially, the mobile virtual assistants are becoming more popular than ever. Recognizing user's intent in interaction is one of the biggest problems while the system is designed and is attracting the attention of researchers. Identifying utterance's speech act in an automated manner is important to reveal user's intention (such as informing, asking, requesting and expressing emotion), which can provide useful indicators to improve the performance of human-machine interaction. Automatic speech act classification has been studied in many languages such as English, Chinese, Slovakian, Arabic, but not yet in Vietnamese. In this paper, we present a speech act scheme suitable for Vietnamese utterances in mobile voice interaction. Also, we built a speech act classifier that achieved 84.36% accuracy. This classifier has been applied in our mobile virtual assistant for Vietnamese (VAV). The experiment result has demonstrated the ability of this classifier such as robust, compact and can work well on mobiles.
Description: ACM International Conference Proceeding Series Volume 08-09-December-2016, 8 December 2016, Pages 396-402 7th Symposium on Information and Communication Technology, SoICT 2016; Ho Chi Minh; Viet Nam; 8 December 2016 through 9 December 2016; Code 125331
URI: http://dl.acm.org/citation.cfm?doid=3011077.3011119
http://repository.vnu.edu.vn/handle/VNU_123/34054
ISBN: 978-145034815-7
Appears in Collections:Bài báo của ĐHQGHN trong Scopus

Nhận xét

Đăng nhận xét

Bài đăng phổ biến từ blog này

Caves in Basalts in Krông Nô Area, Đắk Nông Province, Vietnam

Hình ảnh
Caves have always been creating a mysterious and fascinating beauty for geoheritages, natural heritages, as well as geoparks. Caves in sedimentary rocks, especially in karst in Vietnam are diverse, often breaking new records in length, depth and distribution region wide and world wide. Meanwhile, caves in basaltic rocks in Vietnam were discovered in 2007 in the area of Trinh Nữ waterfall, Krông Nô District, Đắk Nông province. The Trinh Nữ waterfall area is an attractive heritage complex on both natural and socio-cultural aspects. It comprises 3 waterfalls and the famous associated scenery in the Central Highlands, the craters, columnar basalts, resurgences of underground springs, primary forests, national historic cultural monuments, etc. Caves here are quite numerous and usually extend in certain direction. Classified by depth and water table, caves exist in 2 types: dry caves and wet caves. A majority of caves are 3-6m in diameter, some of which are up to over 10m with...
Đọc thêm

Phân tích, đánh giá cảnh quan phục vụ phát triển nông, lâm nghiệp và bảo tồn đa dạng sinh học các huyện biên giới Tây Nam tỉnh Nghệ An : Luận án TS. Kiểm soát và bảo vệ môi trường: 628501

Title:  Phân tích, đánh giá cảnh quan phục vụ phát triển nông, lâm nghiệp và bảo tồn đa dạng sinh học các huyện biên giới Tây Nam tỉnh Nghệ An : Luận án TS. Kiểm soát và bảo vệ môi trường: 628501 Authors:  Nguyễn, Văn Hồng Keywords:  Đa dạng sinh học;Tài nguyên thiên nhiên Issue Date:  2017 Publisher:  H.: Trường Đại học khoa họcTự nhiên Abstract:  Đã làm rõ tính đặc thù: khô,nóng; thiếu nước; thiếu đất sản xuất của 3 lớp cảnh quan, 6 phụ lớp cảnh quan, 6 kiểu cảnh quan, 9 hạng cảnh quan và 110 loại cảnh quan ở khu vực nghiên cứu. - Trên cơ sở phân vùng cảnh quan, tổ chức được 13 không gian lãnh thổ ưu tiên cho phát triển nông nghiệp, lâm nghiệp và bảo tồn đa dạng sinh học tại khu vực nghiên cứu. - Đã nghiên cứu cụ thể về tương tác giữa cảnh quan và đa dạng sinh học; trong đó sự đa dạng về cảnh quan là tiền đề của sự đa dạng về ổ sinh thái, chi phối tổ thành loài và đa dạng sinh học. Description:  151 tr. URI:  http://rep...
Đọc thêm

Food sources of macro-invertebrates in an important mangrove ecosystem of Vietnam determined by dual stable isotope signatures

Hình ảnh
Dual stable isotope signatures (δ 13C and δ 15N) were applied to determine the contribution of mangrove materials and other organic carbon sources to the invertebrate community in an ecologically important mangrove ecosystem of Vietnam. We have analyzed 181 specimens of 30 invertebrate species and found δ 13C and δ 15N ranging from -14.5 to -26.8‰ and from 1.3 to 12.1‰, respectively. From taxa measured for stable isotopes, polychaete, gastropods, bivalves, and grapsid crabs living in mangrove forest showed relative low δ 13C values, while fiddler crabs inhabiting in the land-water ecotone showed the highest δ 13C values. The δ 13C showed that just a few mangrove inhabitants directly relied on the mangrove materials. The wide ranges of δ 13C and δ 15N signatures indicated that the invertebrates utilized heterogeneous diets, comprising benthic microalgae, marine phytoplankton, particulate organic matter, sediment organic matter, mangrove detritus, and meiofauna and rotten ...
Đọc thêm